View
2
Download
0
Category
Preview:
Citation preview
University of Wollongong Thesis Collections
University of Wollongong Thesis Collection
University of Wollongong Year
The virtual orchestra: a systematic
method of realising music composition
through sample-based orchestral
simulation
Leif SundstrupUniversity of Wollongong
Sundstrup, Leif, The virtual orchestra: a systematic method of realising music compositionthrough sample-based orchestral simulation, DCA thesis, Faculty of Creative Arts, Universityof Wollongong, 2009. http//ro.uow.edu.au/theses/872
This paper is posted at Research Online.
http://ro.uow.edu.au/theses/872
The Virtual Orchestra: A Systematic Method of Realising Music
Composition through Sample-Based Orchestral Simulation
Presented in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
by
Leif Sundstrup
BMus (Hons), MMus
Faculty of Creative Arts
2009
Sundstrup 2
Acknowledgements
I wish to thank Professor Stephen Ingham for his encouragement and support throughout my
time as a student at the University of Wollongong. His calm nature and understanding was
especially appreciated during times of difficulty during my candidature.
The completion of my work as a doctoral candidate would never have come to fruition
without the tremendous support of my family. Their patience, acceptance, and support of my
continual absence of concentration on family matters - during times of study - were worth
more than words can describe. My understanding wife Vanessa and sons Jamie and Arran
deserve awards for tolerance and support during my preoccupied times of thought!
I dedicate this thesis to my parents Mary and Erik, who I am sure will be delighted that I have
finally submitted my doctoral project.
Sundstrup 3
Abstract
This thesis investigates a method of orchestral simulation using sample-based synthesis,
instrument modelling, and music performance rules. Music scores are produced using
Sibelius notation software and performed by FATSO (Film and Television Studio Orchestra).
FATSO is a virtual orchestra developed by the author using a combination of computer-
music software applications and expressive instrument modelling techniques capable of
producing convincing simulated orchestral performances of music scores.
Music performance rules are modelled on live human performance practice using both
Analysis-by-Synthesis and Analysis-by-Measurement techniques. The collected data is
analysed, and then implemented into a music score using the Sibelius live playback
transformation feature. After a music score is processed with human performance data, the
instrument sounds and playing techniques are realised by the Vienna Instruments sample
playback engine and GigaPulse convolution reverberation plug-in. The processed
performance data of the score is transmitted to Vienna Instruments via MIDI using sound-sets
created with the Sibelius sound-set editor. Consequently, a music score created using Sibelius
can be performed by FATSO with a high level of realism through detailed instrument
modelling and expressive music performance rules.
This thesis contains two parts. Part 1 discusses the background to sample-based orchestral
simulation and the main components of realistic and expressive orchestral modelling. Part 2
discusses methods used by the author for performance data acquisition, and the resulting
performance data implementation into the FATSO environment.
Sundstrup 4
Contents
1. Background to Sample-Based Orchestral Simulation ............................................8
1.1. Introduction ……………………………………………..………….………….. 8
1.2. Overview ……….....…................……………………..……………………......11
1.2.1. Sampling and Sound Synthesis ……...……………………..……...…11
1.2.2. Vienna Symphonic Library and Sample Playback Engine …...….…..13
1.2.3. Sibelius Notation Software ……………………..…..……………......14
1.2.4. Expressive Performance Modelling ….……....……………………....15
1.2.5. Acoustic Spatialisation and GigaPulse Convolution Reverberation ... 16
1.2.6. Review of the MIDI Protocol …………....…….…………................. 18
1.3. Expressive Human Performance Rules ………......……….....................…...….22
1.3.1. Introduction ..........................................................................................22
1.3.2. Performance Data Acquisition ……………………………….....……23
1.3.3. Ensemble Timing …………………………………….....………...….26
1.3.4. Intonation …………………………………....…….….……….......…27
1.4. Instrument Technique Rules …........………………………………............…...29
1.4.1. Articulation …………………………………………….…….…........29
Sundstrup 5
1.4.2. Note Repetition ………………………………………….…………...31
1.4.3. Performance Transitions …………………….……………………….32
1.4.4. Velocity and Timbre ……………………………...……..………...…33
1.5. Orchestral Environment ………….......…...……………………....………....…35
1.5.1. Orchestral Balance ………………………….....…………..…………35
1.5.2. Dynamic Pitch ………………………….………………………….....36
1.5.3. Hall Resonance and Instrument Localization ……….…………...…...37
1.6. Concluding Observations ...............................................................................….40
2. FATSO Performance Rules ......................................................................................42
2.1. Introduction …….....……………………....……..............………...……...…..42
2.1.1. Analysis-by-Measurement ……………………….……………...…....44
2.1.2. Analysis-by-Synthesis ………………………………….……...……..46
2.2. Performance Data Results……………......……..……………...............……….49
2.2.1. Timing Rules Data ………………………………………..……...…..49
2.2.2. Intonation Rules Data ……………………..…………………….........56
2.2.3. Instrument Timbre Rules Data ………..…………...…………………61
2.2.4. Note Repetition and Legato Rules Data …………..………….....……64
Sundstrup 6
2.3. Performance Rules Implementation …………..........................………………..67
2.3.1. Sibelius Playback Dictionary and SSE ……….……….………......….67
2.3.2. Sibelius LPT ……………………………………….…………...…….71
2.3.3. VI Performance Control …………...………………..……………..…74
2.3.4. Timing Rules Integration ………………………….….…….……......75
2.3.5. Intonation Rules Integration …………………….……….……….….78
2.3.6. Instrument Timbre Rules Integration ……………….….…………....79
2.3.7. Note Repetition and Legato Rules Integration …………..……….….80
2.4. Creating a Standardized Orchestral Environment …….......….…..............……82
2.4.1. Orchestral Layout ……………….…………………………...............82
2.4.2. Spatialisation……………………….……….…………….………..…85
2.4.3. Simulated Microphone Technique …………….………….……….…87
2.4.4. Dynamic Pitch …………………………………….……….…......…..89
2.5. Conclusion ……………......…………………………………...............…….….92
2.5.1. The Future of FATSO …………………………….………….……....92
2.5.2. FATSO as a Performance Resource …………………….……..…..... 93
2.5.3. Concluding Remarks ….……………………….……………….….…94
Sundstrup 7
Bibliography and Works Cited ………………………………….......…..…………...……..96
Appendix 1. Glossary of Abbreviations ……………………….…...…….…...……..........101
Appendix 2. Instrument Dynamic Pitch Charts ……………....……….……….….....….....102
Appendix 3. List of Original Compositions …………...…….……....……….…….......….104
Appendix 4. List of Tracks on Audio CD Recording ...........................................................105
Sundstrup 8
1. Background to Sample-Based Orchestral Simulation
1.1. Introduction
It is widely accepted within the current music industry community that computer simulations
of musical instruments and ensembles have been used for many years prior to 2009.
However, the continual advances in computer technology and software programming have
provided additional opportunities for musicians to produce high quality computer-music
playback without the need for a sophisticated comprehension of computer programming
techniques and access to unlimited financial resources.
Computer simulated music is heard everywhere: radio broadcasts, live productions - theatre,
dance, drama, clubs, shows, and concert performances - television, commercial CD/DVD
recordings, feature films, and internet related music. However, computer simulations of full
orchestral performances have only recently developed to high levels of realism due to
advances in computer technology and music software development. The twenty-first century
has brought an abundance of possibilities for creating realistic orchestral simulations using
powerful audio/MIDI sequencers, detailed instrument reproduction through sound synthesis,
and performance modelling using data acquisition and expressive performance rules.
With the increasing power of computers and rising capacity of data storage, full orchestral
simulations can be rendered in real-time as convincing virtual performances. With the aid of
powerful computer-music software, music compositions can be performed by a virtual
orchestra without requiring the support of live instrumentalists to add that ‘touch of realism’.
However, the task can be particularly complex and requires a thorough understanding of the
intricacies of each instrument in the orchestra, and their collective performance techniques. In
Sundstrup 9
addition, an understanding of expressive human performance practice is crucial for successful
production of realistic virtual orchestra performances using both instrument modelling and
expressive human performance analysis. Once an understanding of how an orchestra ‘sounds’
- the characteristics of individual instruments and effects of ensemble playing – performance
rules can be devised to replicate these attributes and implement them into the virtual orchestra
playback process.
FATSO’s computer generated performances are based on empirical methods of virtual
orchestra simulation using five dedicated software programs: Sibelius, Sibelius sound-set
editor (SSE), Vienna Ensemble (VE), Vienna Instruments (VI), and GigaPulse convolution
reverberation. Each software application forms a sequence from the original notated music
score created using Sibelius, to the final orchestral simulation performed by FATSO. The five
main applications generate the source, performance rules, articulation functions, instrument
techniques, and sample playback, sequentially. Accordingly, this thesis examines each
software application and the techniques used to quantify the composite processes required to
generate an impressive, convincing, and realistic music performance through sample-based
orchestral simulation.
The approach to high-level orchestral simulation discussed in this thesis is based on direct
playback from a notation program. It is an unusual method of orchestral simulation that
differs considerably to the current techniques used in the music industry. The standard
methods of music production and expressive orchestral simulation used in the music industry
involves MIDI data manipulation using third-party audio/MIDI sequencers such as SONAR,
Cubase, Logic, and Digital Performer. However, in this thesis, all expressive performance
rule parameters are implemented and sequenced within the Sibelius notation application. This
is an original approach to high-level orchestral simulation as audio/MIDI sequencers are not
Sundstrup 10
utilised in the creation of virtual orchestra performances. Consequently, once a score is
created using Sibelius, all performance rules and expression are created within the Sibelius
environment and directly sent to VI via MIDI using SSE settings.
Sundstrup 11
1.2. Overview
1.2.1. Sampling and Sound Synthesis
"Sound Synthesis is the process of producing sound. It can reuse existing sounds by
processing them, or it can generate sound electronically or mechanically (Russ, 4). "Sampling
is the process of recording a sound source one part at a time, each part of which is then
imported into a sampler" (McGuire, and Pritts, 1). Although early forms of sampling used
tape-based analog systems such as the Mellotron, it is now widely accepted in the music
industry that the creation and manipulation of sounds is predominantly performed by digital
means. Gilreath states in his Guide to MIDI Orchestration that: “The world of orchestral
sampling can be traced back to the 1960s. The Mellotron, which generated orchestral sounds
using pre-recorded strips of analogue tape, was the first 'sampler' available to the public”
(Gilreath, 521).
Electronic sound synthesis includes many types of sound creation and sound processing
methods such as subtractive synthesis, additive synthesis, wavetable synthesis, sample replay,
and physical modelling. This thesis refers to sound synthesis as computer manipulation and
replay of sampled orchestral instruments and their various articulations. Pioneers of sample-
based music composition included Pierre Schaeffer, Luciano Berio, and Karlhheinz
Stockhausen. "Musique Concrète is a French word which has come to be used as a
description of music produced from ordinary sounds which are modified using tape
techniques [...] Pierre Schaeffer coined the term in 1948 as music made from existing sonic
fragments" (Russ, 19). Luciano Berio and Karlheinz Stockhausen also used sampled sound
sources for Electronische Musik. "Early electronic composers found that manipulation of
Sundstrup 12
recorded sound materials opened a whole new palette of sound sources for musical
expression" (McGuire, and Pritts, 193).
Digital samplers began their existence in hardware form before the development of computer
sampling software. The Fairlight Computer Musical Instrument was developed in 1979
followed by the Ensoniq Mirage and the EMU Emulator samplers. Both the Mirage and
Emulator made instrument sampling accessible, and are still used for certain genres of music
production - especially dance music. Computer software applications have almost completely
eliminated hardware samplers due to their superior power and sample storage ability. “The
21st century has seen a wide adoption of software sample playback as an alternative to
hardware: either as plug-ins to software MIDI and audio sequencers, or as stand-alone
‘sample’ sequencers” (Russ, 26).
A major breakthrough in sample playback technology came when software developers
NemeSys released GigaSampler. “It was this product that included the streaming technology
necessary to move sampling to the next level” (Gilreath, 522). Disk streaming permitted
samples to be triggered directly from the hard drive and allowed massive sample collections
to be instantly ready for playback. Before disk streaming, sample storage was limited to the
small amounts of RAM available in the computer.
The current 2009 industry leaders of powerful orchestral sample libraries include the Vienna
Symphonic Library (VSL), Sonic Implants Symphonic Collection, and East-West Quantum
Leap Symphonic Orchestral Library. These libraries include a vast number of orchestral
instruments and thousands of articulations with many dynamic levels. “The Vienna
Symphonic Library is the most ambitious orchestral sample library ever produced” (Gilreath,
Sundstrup 13
630). VSL includes over 600 gigabytes of instrument sound samples and articulations, and is
the chosen sample playback device used to simulate orchestral performances by FATSO.
1.2.2. Vienna Symphonic Library and Sample Playback Engine
VSL is a highly sophisticated orchestral sample library that is integrated within the VI sample
playback engine. The VI sample playback engine streams the sampled instruments of VSL
from computer hard drives. There are over twenty-five instrument groups currently available
including a colossal amount of articulations, dynamics, performance repetitions, and true
legato samples. The sample playback engine can be operated in three modes: Virtual Studio
Technology (VST), Audio Units, or Stand-alone. In this thesis, the VST plug-in format is
used to integrate VI into Sibelius using a VI host application called Vienna Ensemble (VE).
VST is an audio/MIDI format developed by Steinberg for integrating third-party software
applications into host sequencers and audio/MIDI applications. Consequently, VI can be
integrated directly into Sibelius without the requirement of additional sequencers,
audio/MIDI programs, or further audio/MIDI hardware equipment.
The VI sample playback engine works on three levels of sample integration: pre-sets,
patches, and matrices. Pre-sets are company-configured collections of instrument
articulations ready to use. Patches are the individual sampled sounds/articulations comprising
the sample library instruments. Matrices are used to arrange arrays of two-dimensional cells
that hold patches. There are 144 cells - each able to hold two patches - available for every
instrument matrix and can be triggered by key-switches, program changes, speed control, and
velocity control.
Sundstrup 14
1.2.3. Sibelius Notation Software
As revolutionary as the computer word processor has been for document typesetting, music
notation software has similarly transformed the way orchestral composers work. With the
development of powerful notation applications have come unique and sophisticated, yet
economical ways of producing professional manuscript scores ready for publication. The two
industry leading notation programs currently available in 2009 are Finale by MakeMusic, and
Sibelius by Sibelius Software. In this thesis, Sibelius is used as the primary environment for
both music notation and orchestral simulation due to its powerful MIDI controller and key-
switching capabilities.
Sibelius has been designed to play back a score with a reasonably accurate interpretation of
the notated music. However, the result is, in my view, neither expressive nor convincing, and
needs further processing to generate the required realism appropriate for believable orchestral
simulation. The expressive human performance formulas used to process a Sibelius score are
implemented using the Live Playback Transformation (LPT) feature in Sibelius.
Until the release of Sibelius version No. 5 in April 2007 that included VST hosting, most
scores written using Sibelius required export to a third-party sequencer to add performance
expression and permit successful integration with sample play-back applications. Sequencers
are still used by many of the major film scoring studios to add expressive performance play-
back. However, with the use of LPT in Sibelius, all expressive control can be accomplished
within the Sibelius environment subject to the music originating in score form.
The performance rules implemented using LPT require a context or source phrase. This may
consist of a series of notes including their articulation, pitch, dynamic, intonation, note
Sundstrup 15
duration, and tempo indication. The performance rules discussed in this thesis are based on
the manipulation of five separate note parameters: note length, rhythmic placement,
intonation, timbre, and dynamics - volume, attack, release and decay.
1.2.4. Expressive Performance Modelling
“Interpretation is one of the most important aspects of music performance” (Friberg, 2).
Research into expressive human performance practice has revealed that the unique qualities
of musical expression are a direct result of deviations from a mathematically perfect
interpretation of notated music. These deviations are based on both intentional and
indiscriminate nuances in expressive music performance. There are many research teams
working in the area of expressive human performance and developing rule-based systems for
automatic expressive music interpretation in computer-music technology. The research team
at the Department of Speech, Music and Hearing (KTH) at the Royal Institute of Technology
in Sweden, have been formulating rules for musical expression using Director Musices, a
software application developed at KTH.
Director Musices is a program that converts music scores into performances using rules based
on research conducted at KTH. “Rules in the program model performance aspects such as
phrasing, articulation, and intonation, and they operate on performance variables such as
tone, inter-onset duration, amplitude, and pitch” (Bresin, Friberg, and Sundberg, 1).
Many software notation applications - including Sibelius - generate simple performance rules
for score playback based on instrument articulations, symbols and score indications.
However, in my view, the quality of realism is not convincing as many sophisticated rules are
required for detailed modelling of instrument performance that do not currently exist in any
Sundstrup 16
notation applications to the required extent. Subsequently, performance rules are usually
formulated and processed using third-party music software applications other than Sibelius
unless a performance rule method can be implemented using the LPT feature within the
Sibelius environment. In this thesis, the latter method is used.
Two classes of performance models are used: expressive human performance, and instrument
technique. The methods used to model expressive human performance are based on
intonation, dynamics, and instrument/ensemble timing. The methods used to model
instrument technique are based on timbre, articulation, sample repetition, and performance
transitions. These systems of performance modelling are based on both random and
predictive rule-based deviations from the notated score correlating to human performance
practice.
The rules created to model human player expression discussed in this thesis are based on
research conducted by the author. The author’s rule systems for instrument performance
modelling are derived from two methods of performance analysis: Analysis-by-Measurement
and Analysis-by-Synthesis. The Analysis-by-Measurement method of human performance
data collection is based on the author's original research analyses of performances by ROSO
(Royal Oman Symphony Orchestra).
1.2.5. Acoustic Spatialisation and GigaPulse Convolution Reverberation
The sound of an orchestra is based on the interaction of many different instruments, which
are used in both solo situations and as ensemble groups. It is affected by the spatial expansion
of the ensemble and by the influence of the surrounding acoustics. Acoustic measurement
procedures allow an analysis and description of the tonal characteristics of individual
Sundstrup 17
instruments with regard to sound intensity and reverberation. Appreciation of the
complexities of an orchestral sound requires an understanding of the effect of sound delay
times within an orchestra, including the associated problems of playing in time, as well as
differences in hall reflection times for different instrument locations on stage.
Given that sound travels at about 1,100 feet per second through dry air at 20 degrees Celsius,
not every performer in an orchestra is going to hear a specific sound at an identical time as
another performer in the orchestra. Furthermore, the audience will recognize instrument
sounds – both direct and reflected - later than most orchestral performers. Add to this the
various sound reflections from the acoustic properties of the performance space, and it
becomes obvious that the perception of sounds emanating from different orchestral
instruments reach individual players with varying levels of intensity and sonic qualities.
Sounds emanating from a sound source to a listener have a delay time of about 1 millisecond
per foot from the source. Consequently, a player seated 40 feet from a sound source on stage
will hear the sound 20 milliseconds later than a player seated only 20 feet from the same
sound source. This is a significant difference that affects the entire sound of a large
symphony orchestra due to the natural deviations in note onset/offset times.
To model this phenomenon, the virtual players in FATSO are allocated a natural time delay
based on their distance from front/centre stage where the conductor usually stands. This is an
ideal position to set time delay settings appropriate for an audience in an average concert hall
with the front row audience seated at least 15 feet from front stage next to a virtual stereo
microphone. Once these settings are formulated, a convolution reverberation can be added in
addition to the direct sound delays between instrument sections to add a natural performance
space for the virtual orchestra. Some convolution reverberations, such as GigaPulse, control
the necessary timing differences as a part of the reverberation processing algorithms.
Sundstrup 18
GigaPulse is a VST convolution reverberation plug-in developed by NemeSys. Unlike most
synthetic reverberations available today, GigaPulse reconstructs reverberation using actual
sampled spaces including halls and stages. FATSO employs four separate instances of
GigaPulse to render the VI playback performances simulating separate virtual locations in a
medium concert hall. These separate locations are based on the impulse response data
collected at different stage positions in the hall rendered by GigaPulse. The impulse
responses imitate the varying hall acoustics displayed at different stage positions.
1.2.6. Review of the MIDI Protocol
MIDI (Musical Instrument Digital Interface) was developed in the early nineteen-eighties to
allow different electronic musical devices to communicate with each other using digital
messages in binary code. Originally used to communicate between various hardware
synthesizers, controllers, and effects modules, MIDI is also used within most computer
sequencing applications, recording software, and notation programs. MIDI uses sixteen
independent channels on which data can be transmitted and received by various devices
through a MIDI cable. Each channel can send detailed messages independent of the other
channels and can therefore be utilized to control different devices/instruments
simultaneously. Consequently, each MIDI cable - or virtual cable within a computer
environment - can control up to sixteen independent digital instruments simultaneously.
MIDI travels in only one direction down a MIDI cable and is transported from one device to
another through a MIDI port. Most MIDI devices have a ‘MIDI In’ port and a ‘MIDI Out’
port to send and receive MIDI messages. In this thesis, a virtual MIDI cable connecting a
‘MIDI Out’ port to a ‘MIDI In’ port will be referred to as a 'MIDI tunnel'.
Sundstrup 19
MIDI is the protocol used to communicate between Sibelius and VI. It is utilised to transfer
all programmed notational and expressive music information from a Sibelius score to VI in
real-time, to simulate a musical performance based on the score’s macro information - printed
symbols, rhythmic notation, and expressive indicators - and micro information - controller
information embedded in the score based on expressive music performance data. The MIDI
protocol allows Sibelius to send messages to VI including user-inputted information detailing
how to perform a Sibelius score. Although each MIDI tunnel can control sixteen instrument
channels, a Sibelius score can often employ more than forty separate instrument samples in a
full orchestral composition. Therefore, more MIDI tunnels are required to control all the
instruments separately and can be easily setup to accommodate each individual instrument
with only one important factor to consider - computer resources.
MIDI messages send information based on two classes of data information: system messages
- controlling data pertaining to all the channels on a MIDI port - and channel messages -
controlling data appropriate to individual channels independently of each another. System
messages are analogous to a Sibelius score and control generic messages including score
start/continue/stop commands, and overall volume. Channel messages are analogous to each
instrument in a Sibelius score and control detailed performance parameters including note
on/off commands, articulation keys-switches, pitch messages, velocity, pitch bend, pan, and
modulation performance controllers amongst many others. Each instrument channel message
can also include comprehensive control over note attack, decay, and release times, instrument
brightness, fine-tuning, and any other sound manipulation commands that are available via
MIDI channel messages within the synthesizer, sampler, or software application in use.
A MIDI controller sends various messages to a MIDI device with real-time information
regarding performance parameters. The most common MIDI controller is a controller
Sundstrup 20
keyboard or any keyboard with a ‘MIDI in’ and ‘MIDI out’ port. Sibelius is programmed to
send appropriate MIDI controller messages to VI instead of using a controller keyboard. In
this thesis, channel messages used to control VI from Sibelius include the following
commands: note-on messages, note-off messages, pitch bend, program changes, key-
switches, and expressive controller messages.
Note-on messages sent from Sibelius tell VI which of the sixteen channels available through
a MIDI tunnel to address and when to begin a note - including information regarding pitch,
duration and velocity (dynamics). The pitch numbers can vary from 0 to 127 - C2 to G8 on a
keyboard where C4 is middle C - and the velocity numbers can vary from 0 (silence) to 127
(maximum velocity). Note-off messages sent from Sibelius tell VI when to terminate a note-
on command. Pitch-bend messages raise or lower the pitch of notes and alter the frequency of
a note by up to a tone - higher or lower - within the VI interface. Program changes sent from
Sibelius to VI are used to change instrument patches during a performance. For example, a
program change can tell a virtual clarinet player to change to a virtual saxophone by
switching the sample patch, or tell a virtual violin section to play sul ponticello by switching
to an appropriate sample patch at a user allocated point in the score. Key-switches are a
relatively recent development in the control of virtual instruments and make it possible to
quickly change instrument parameters by using the keys on a controller keyboard that are out
of the pitch register of the relevant instrument. Key-switches are often used to change
instrument articulations during a performance and can be allocated within the Sibelius
environment without requiring a controller keyboard. Control changes (CCs) allow many
different parameters of a MIDI channel to be adjusted in detail. There can be up to 128
different assignable parameters for each MIDI channel transported via a MIDI tunnel. Each
parameter extends from 0 to 127 in values appropriate to the specific device function. The
Sundstrup 21
most frequently used controllers to adjust parameters in VI are CC 1 (modulation), CC 7
(volume), and CC 10 (pan). Assignable CC’s can also control fine adjustments of note attack,
release, sustain, and decay amongst many other possible parameters.
Sibelius can be programmed to send any CC data that VI has been setup to interpret, and
includes a vast collection of performance parameters and key-switches for each individual
instrument within Sibelius. SSE is used to construct performance parameters based on
instrument type and articulation. All notation symbols and indications can also be allocated
specific instrument performance variations to model live instrument performance practice.
Sundstrup 22
1.3. Expressive Human Performance Rules
1.3.1. Introduction
Expressive behaviour – whether intentional or unintentional - in every form of
communication has been a source of rigorous scientific research. In the field of music, much
research has focused on expressive music performance practice and musical interpretation.
However, interpretation includes both intentional and unintentional expressive actions.
Much research into expressive human performance practice has revealed that the unique
qualities of musical expression are a direct result of deviations from a mathematically precise
interpretation of a notated score. Accordingly, the expression in musical performance
“consists in aesthetic deviation from the regular – from pure tone, true pitch, even dynamics,
metronomic time, rigid rhythms, etc” (Seashore, 9). The deviations in human performance -
from the information symbolized in a score - are quantified by means of human performance
analysis. These performance deviations are categorized into four variables: “frequency,
intensity, duration, and form” which are equivalent to “pitch, loudness, time, and timbre”
(Seashore, 29).
Humanization – the act of simulating human qualities – in music expression is based on the
four elements of human performance practice classified by Seashore. Through human
instrumental performance practice and analysis, a rule-based system is developed and
implemented into a musical score using LPT in Sibelius. The rules created are based on the
reality that performers do not play perfectly in tune, perfectly in time, or perfectly together.
There is both a random element in these performance deviations and a predictable
performance factor that must be considered when developing expressive performance rules.
Sundstrup 23
As the scope of this thesis does not allow for detailed research into intentional expressive
music performance (rubato, phrasing, and vibrato), the performance rules developed for
FATSO are based on unintentional performance deviations from a mathematically accurate
interpretation. These involve deviations in intonation, timing, dynamics and timbre.
However, the performance deviations - from a mathematically perfect interpretation of a
notated score - add vitality, warmth and humanization to an otherwise lifeless and
unconvincing computer generated performance. FATSO currently has limitations due to the
lack of intentional expression. However, due to the performance rules used based on
unintentional expressive performance deviations, FATSO is approaching levels of realism
acceptable to professional film, television, and radio producers.
Through research at ROSO by the author, performance data based on Analysis-by Synthesis
and Analysis-by-Measurement has produced a number of performance rules used in FATSO.
Performance rule models based on unintentional performance practice implemented into
Sibelius include: high/loud – increase sound level in proportion to pitch height;
duration/contrast – shorten relatively short notes and lengthen relatively long notes;
high/sharp – stretch all intervals in proportion to size; note duration – adjust note start and
note finish timings; pitch – alter note pitch with micro-tuning; and timbre – manipulate
dynamics, balance and note execution.
1.3.2. Performance Data Acquisition
There are two major ways to gather expressive performance data from a human musical
performance. The first is to record a live performance using various measurement devices
including MIDI, video, audio, or movement sensors. The second is to extract performance
data from an audio recording. The latter method can be extremely difficult unless the audio
Sundstrup 24
recorded uses the multi-track recording technique to isolate each individual player in an
ensemble. According to the paper 'Sense' in Expressive Music Performance: Data
Acquisition, Computational Studies, and Models, Binet and Courtier made the first attempts
to record expressive music performance data:
Of the first to record the movement of piano keys were Binet and Courtier
(1895) who used a 6-mm caoutchouc rubber tube placed under the keys that
was connected to a cylindric graphical recorder that captured continuous air
pressure resulting from striking different keys on the piano. They investigated
some basic pianistic tasks such as playing trills, connecting tones, or passing-
under of the thumb in scales with exemplary material. (Goebl, Dixon, De Poli,
Friberg, Bresin, and Widmar, 3)
Since these studies, performance data has been recorded using piano roll data, gramophone
recordings and more recently, electronic and MIDI pianos. Most of these techniques were
based on single instrument data acquisition, and it is only with the development of MIDI and
multi-track audio sequencers that multi instrument performance data has been recorded with
an acceptable level of accuracy to analyse.
The Machine Learning and Intelligent Music Processing Group at the Austrian Research
Institute for Artificial Intelligence have investigated expressive music performance:
by measuring expressive aspects such as timing, dynamics, etc. in large
numbers of recordings by famous musicians (currently: pianists) and using AI
technologies – mainly from machine learning and data mining – to analyze
these data and gain new insights for the field of performance research.
(Widmar, Dixon, Flexer, Goebl, Knees, Madsen, Pampalk, Pohle, Schedl, and
Tobudic, 1)
Two methods of human performance data acquisition are used as a foundation for the
performance rules created for FATSO: Analysis-by-Measurement and Analysis-by-Synthesis.
Analysis by Measurement quantifies data collected from live human performances. The
Sundstrup 25
performances are recorded into an audio sequencer using the multi-track technique to gather
information from each player in the ensemble. By placing microphones extremely close to
each instrument, acceptable individual audio images can be obtained from each player in the
ensemble. Specific performance information is collected from the recorded data and used as a
source for new human performance rules or as a foundation for the Analysis-by-Synthesis
method. The data collected for FATSO was acquired from multi-track recordings of ROSO.
Four classes of performance information were collected using Analysis-by Measurement:
dynamics, intonation, timbre, and note onset/offset timings. Note onset/offset represents
timing differences between each instrument in an ensemble performing notated music
indicated as rhythmically identical. The breadth of deviations in dynamics, intonation, timbre,
and ensemble timing are calculated from the least noticeable differences (LND) to the most
noticeable differences (MND).
Analysis-by-Synthesis uses an artificial performance that is realised using a hypothetical
performance rule and then evaluated by listening. As the performance rule is continuously
manipulated and improved by listening, a final natural sounding interpretation can be
developed. However, the success of Analysis-by-Synthesis is based exclusively on the skill of
the listener. Consequently, the performance rules developed for FATSO using Analysis-by -
Synthesis are based on the author’s musical ideals, giving FATSO a unique expressive sound.
Many of the hypothetical performance rules realised through Analysis-by-Synthesis are based
on data collected through Analysis-by-Measurement. The performance rules developed for
FATSO using the Analysis-by-Synthesis technique are based on experiments conducted on
scores created in Sibelius and processed using LPT. LPT allows manipulation of intonation,
dynamics, timbre, instrument attack/decay, and timing within the Sibelius environment.
Sundstrup 26
1.3.3. Ensemble Timing
There are two influences on ensemble timing that need to be modelled for a convincing
virtual orchestra performance. The first influence is based on acoustics and the speed of
sound. The second influence is based on each player’s rhythmic accuracy, musicianship and
expressive performance practice. Of course, these influences concern deviations in timing
accuracy from a mathematically correct performance of a notated score and contribute
enormously to the realism of a computer generated orchestral performance. The ensemble
timing data collected through Analysis-by-Measurement include note onset/offset timings
and note duration. Ensemble timing rules are developed based on performance deviations
under various musical conditions based on tempo and musical style. There is also a random
element of timing deviation based on a performer’s precision and flexibility.
The acoustic situation of instrument players greatly differs from that of the listeners. From
the position of the conductor, the delay time of the direct sound between the individual
instruments may rise to 35 milliseconds, and between the outer players of an orchestra up to
45 milliseconds. The relation between a player's own instrument and the level of the other
instruments is very important. Thus, the level of other instruments may be supported by
reflections arriving during a time interval of more than 30 milliseconds after the direct sound.
Shorter delay times lead to sound levels of other instruments perceived as higher in relation
to the apparent sound level of their own instrument.
Through Analysis-by-Measurement, timing deviations between instrument performers were
analysed based on multi-track recordings of ROSO. The recordings were created at the Royal
Guard of Oman Music Studio Hall, which generates a moderate reflection time of about one
second. The performances were recorded into ProTools audio software providing a graphic
Sundstrup 27
user interface (GUI) time-line of each instrument’s performance. The timing rules created are
based on unintentional deviations and do not include expression based on tempo changes,
musical phrasing, or intentional rubato. The performance rules developed focused on
deviations from LND to MND in milliseconds. The deviations between LND and MND were
devised from each player’s note onset/offset data during performances at various tempi and
dynamics.
As a general observation, slower tempi produced greater deviations in note onset/offset
timings. Consequently, the formulas developed for the timing rules within Sibelius are based
on notated length rather than note duration (in milliseconds), where each note is divided into
many segments based on one crotchet equalling 256 divisions called ticks. Using
performance rules that adjust notated length and note onset/offset values coincides with the
observation that quicker tempi produce smaller deviations in ensemble timing. Accordingly,
slower tempi produce greater deviations in ensemble timing and as a result, each note tick
produces a greater effect on note onset/offset times according to tempi.
1.3.4. Intonation
According to Pejrolo and DeRosa in their book Acoustic and Midi Orchestration for the
Contemporary Composer, “One of the biggest problems with a virtual MIDI orchestra is the
fact that it is always perfectly in tune, especially if you are using synthesized patches […]
This is the reason why it is recommended that you make your virtual ensemble sound a bit
'worse' than it could through a gentle use of the detune parameter” (Pejrolo, and DeRosa,
142). This is a very basic recommendation that can have a profound effect on an orchestral
simulation, depending on the methods of detuning used.
Sundstrup 28
Three methods of detuning are used for FATSO:
• Detuning each string section and solo instrument by a small amount as a set parameter
for the entire score.
• Random detuning between solo instruments for the duration of the score.
• Intonation performance rules applied to FATSO based on analysis of the ROSO
recordings.
The intonation performance rules used for FATSO include high/sharp, loud/sharp, and
random intonation data.
The intonation performance rules are not applied within the separate strings sections - first
violins, second violins, violas, cellos, and basses - as each ensemble section sample already
contains the natural variability between players when recorded. However, the intonation
performance rules are applied to the separate ensemble sections within the entire string
section. For example, intonation performance rules are applied to the cello section, but not to
individual players within the section. This situation exists because each section was initially
sampled as a whole, and individual instruments cannot be separated from each other within
the string ensemble section samples.
Sundstrup 29
1.4. Instrument Technique Rules
1.4.1. Articulation
One of the most significant components of an instrument’s personal character comprises of
the initial transients of an articulated tone. Other than the harmonic content - that determines
an instrument’s unique sound quality – the prime influence on persuasive expressive
performance simulation is note onset articulation - also known as note attack. “The term
articulation is used to describe the amount of legato/staccato with which a note is being
played. It is defined as the ratio between the note duration (i.e. sounding duration) and the
IOI” (Friberg, Bresin, and Sundberg, 150). The IOI (inter-onset interval) referred to by
Friberg, Bresin, and Sundberg concerns the time from one note’s decay to the next note’s
attack. Accordingly, articulation can be expressed as either the onset/offset of separate tones,
or the onset/offset of one tone in relation to another. Consequently, articulation refers to
single notes, note repetition, and performance transitions as discussed later in this chapter.
Before the current orchestral sample libraries were available – including thousands of
instrument articulations – the most common notated articulations - including staccato,
marcato, tenuto, and accent - needed to be simulated by changing each note’s attack, decay,
sustain, and release (ADSR) times using envelope generators such as oscillators and filters.
An envelope generator “is a multi-stage controller that allows the synthesizer to control over
time the amplitude of a waveform” (Pejrolo, and DeRosa, 133). However, with the
development of detailed orchestral sample libraries, each type of instrument articulation is
sampled with all the natural transients included for each dynamic level. Pejrolo and DeRosa
justify the importance of sample-based orchestral libraries for realistic orchestral simulation:
Sundstrup 30
The advantage of using sample-based sounds instead of synthesized ones lies
in the fact that while a synthesizer tries to re-create complex acoustic
waveforms artificially through the use of one or more oscillators and filters, a
sampler uses random access memory (RAM) to store the original recording
(samples) of an acoustic set of waveforms that can be played (triggered) using
a generic MIDI controller. This technique has the huge advantage that if you
have enough RAM, and if the original samples were recorded and
programmed accurately, the results can be astonishing. (Pejrolo, and DeRosa,
120)
As a result, a notated tone in a music score with an attached articulation symbol - staccato,
tenuto, or an accent - can be programmed to trigger a real sampled sound of a similar
articulation at the equivalent dynamic level on the chosen instrument. Sibelius can be
programmed to trigger the appropriate instrument articulations in VI based on the notated
articulation and dynamic indication in a score using SSE. SSE uses key-switches, program
changes, and controller messages to alternate between various articulations based on the
information sent from Sibelius using the instrument performance rules pre-programmed
within SSE.
Although a generic template of articulation rules can be programmed using SSE, there are
also unintentional expressive performance rules that also affect the variations within each
performed articulation. “Tones can be played longer or shorter than their nominal duration.
Furthermore, tones can be played with different sound levels, or with different vibrato, or
with different tone attacks, etc” (Jerkert, 6). From findings based on Analysis-by-
Measurement - discussed in Part 2 of this thesis - the author considers deviations in
articulation to be both intentional and unintentional. The performance rules created for
FATSO are based on unintentional deviations in both note attack and IOI.
Sundstrup 31
1.4.2. Note Repetition
Note repetitions are often the cause of synthetic ‘machine gun’ effects heard on many film
and television shows using synthesised orchestral mock-ups. Many computer generated
simulations use the same sample patch for each repeated note that has the same pitch and
dynamic. The result sounds artificial as the same sampled note is performed in sequence and
does not contain the intricate deviations in tone, dynamics, and articulation that each repeated
note performed by a live instrumentalist or instrument section would display. The audio CD
recording of Voyage – performed by the standard Sibelius sample playback and general
Sibelius performance rules – gives a prime example of the 'machine gun' effect (see
Appendix 4).
One way to increase the realism of repeated notes is to adjust each sampled note’s dynamic
level, timbre, and length by a very small amount. However, in the latest sophisticated
orchestral sound libraries, note repetition simulation is achieved by using variant samples for
each repeated note in a sequence. For example, semiquaver notes - performed at various
dynamic levels - are recorded several times to provide the slight but necessary variations for
every repeated note. The VI sample playback interface provides up to nine variations of
repetition for each sampled note at every recorded dynamic level. As each variation of a
sampled note is recorded using live musicians, the reality is enhanced, and as long as the
score is orchestrated using knowledgeable skills appropriate for a live orchestra, the note
repetitions performed by both solo virtual instruments and virtual instrument sections became
more convincing. The note repetitions used in VI are not limited to separate tones including
sustain, staccato, portato, tenuto, and accents. An enormous library of sampled repetitions
also exist for legato phrases, portamento, and performance trills, covering many possible
Sundstrup 32
orchestration techniques used in a notated score, except for various modern extended
performance techniques such as bowing behind a violin bridge or string harmonic glissandi.
1.4.3. Performance Transitions
Before the release of VSL – originally hosted by the GigaStudio software sampling
workstation before the VI engine was developed – instrument tones performed by a virtual
instrument were of the ADSR type. “At least since von Helmholtz, musical notes have been
split into a central region called the steady state, which is preceded by an attack and followed
by a decay” (Strawn, 867). The manipulation of an instrument’s ADSR could – to a
reasonable extent - simulate single articulations required for single tones. However, the
difficulty in simulating connected tones - as in legato and tenuto - has challenged sample-
based sound designers for decades. Strawn further verifies the importance of note transitions:
But much of an instrument’s tell-tale “sound” lies in how the notes are
connected, and thus in how musical phrases help create the instrument’s
“signature”. It is thus important to examine more than one note at a time if the
nature of musical sound is to be fully understood. (Strawn, 867)
A performance transition includes the ending decay of one note and the beginning attack of
the next note. One method of simulating a legato phrase comprising two notes is to lessen the
decay of the first note and lessen the attack of the second note to smoothly join the tones
together. Although this achieves a moderately effective simulation of legato, it has, until
recently, never successfully accomplished an accurate and convincing note transition.
However, due to the recent development of huge computer storage devices, orchestral sample
libraries now have the resources and memory storage accessibility to not only store sampled
single notes, but also masses of performance samples of true legato transitions.
Sundstrup 33
At the forefront of sampling technology - concerning sampled note transitions - is VSL, with
its true legato samples. All legato transition intervals - between a minor second and an octave
- have been sampled at different dynamic levels for many of the solo instruments and string
ensemble sections. This makes adjustments to ADSR envelopes unnecessary for true legato
playback, as notated phrases and note transitions - indicated by a slur in Sibelius -
automatically trigger true legato interval samples within VI using parameters set in SSE.
1.4.4. Velocity and Timbre
The sound levels of orchestral instruments display varying changes in timbre according to
both the dynamic level and pitch register of the produced tones on an instrument. As is
clearly observed in a live orchestral performance, the louder an instrument or ensemble
section performs a passage of music, the brighter the passage sounds due to the increase of
harmonic content in each instrument’s timbre. The brass section of an orchestra displays the
greatest transformation of timbre according to volume. Not only can brass instruments
accomplish the warmest and smoothest sounds of the entire orchestra - at low dynamic levels
in the mid register of each instrument - they can also achieve the brightest and hardest sounds
in an entire orchestra - at high dynamic levels. This can be simulated to various extents using
software frequency filters to adjust the sample’s harmonic content according to the pitch and
dynamic notated in a score. However, with the advance of detailed sample libraries, the
changes in instrument dynamics/timbre are captured at various dynamic levels during the
sample recording session and often do not require the necessity of a frequency filter to adjust
the harmonic content of an instrument’s dynamic spectrum.
Although each tone on a virtual instrument can be pre-set to simulate the sound of a live
instrument by using appropriate instrument samples and mapping templates, the evolving
Sundstrup 34
transformations in dynamics and timbre over time – such as a single note crescendo – is a
little more complicated and requires the use of cross-fading. Cross-fading is a technique used
in MIDI orchestration and synthesis to blend different samples together. It is often used to
change from one sampled sound to another over time, as in a single note
crescendo/diminuendo. For example, if a crescendo is programmed to affect a single note
over time - without the use of filters to brighten the sound - the result will be similar to
turning the volume level up. With cross-fading, as the volume increases over time, a sample
at a specific dynamic level will fade into another sample recorded at a greater dynamic level
and so on. Consequently, the evolving dynamic changes over time reproduce the natural
timbral changes observed in a live performance. The same effect is used in reverse order to
simulate decrescendos.
Changes in micro-level dynamics and timbre exist throughout the duration of an orchestral
performance. These changes can be programmed - on a random level - to simulate the slight
imperfections and nuances experienced in a live performance. Not only can micro-level
adjustments to notes be made using slight variations of velocity, they can also be shaped by
adjusting the filter settings - over time - to create small deviations to the sampled attack and
timbre of each note. Subsequently, the abnormally constant timbre experienced in many
computer generated performances will be eliminated, and the notated phrases performed by
FATSO acquire a natural and humanistic vitality not experienced in a perfect computer
generated orchestral simulation.
Sundstrup 35
1.5. Orchestral Environment
1.5.1. Orchestral Balance
The balance between individual instruments and ensemble sections within an orchestra has a
major affect on the combined sound produced. It is complicated enough to evenly balance an
orchestra’s complex sound in live situations, let alone simulate a convincing homogeneous
equilibrium within a virtual orchestra environment. Furthermore, it is very difficult to
stipulate exactly the finer details of balance, as every situation throughout an orchestral work
calls for an evolving sound level change between instruments and ensemble sections to
emphasize the emerging musical moments and their varying tone colours. An important
attribute of orchestral balance to appreciate is the dynamic range of each individual
instrument. The softest and loudest tones produced on individual instruments can vary
drastically in timbre from one another. Consequently, consideration of the timbre for each
instrument at contrasting dynamic levels must be assessed to appropriately fine-tune a natural
orchestral balance.
It is appropriate to mention here that most sample library instruments are recorded at the
highest input level on the recording machine to capture the cleanest possible audio signal
with the highest signal-noise ratio. Consequently, the softest tone on a clarinet will playback
at the same sound level as the loudest tone on a trumpet. Accordingly, all recorded instrument
dynamics – whether soft or loud – result in the same output level when played back and must
be re-adjusted after the sampling process to accurately represent the natural dynamic levels of
a live instrument. Accordingly, phrases notated to be performed at a soft dynamic – such as
piano – will trigger a sound originally recorded at a soft dynamic level as played by the
musician. The same outcome applies to loud dynamic levels in sampled instrument tones and
Sundstrup 36
must be adjusted to imitate the natural timbral changes experienced at loud dynamic levels on
an orchestral instrument.
1.5.2. Dynamic Pitch
Research into acoustics, instrument timbre, and sound intensity, has revealed much
information about the dynamic range of individual orchestral instruments. As mentioned
earlier, the dynamic range is based on the sound pressure level differences between the
softest tones an instrument can produce and the loudest it can produce - measured in decibels.
There are three classifications of dynamic range used in an orchestral context: individual
instruments, instruments in groups, and the entire orchestra. The importance of allocating a
correct dynamic range for each instrument in a virtual orchestra is paramount if the dynamic
balance and appropriate instrumental timbre is to be convincing. In his thesis The Sound of
an Orchestra, Meyer found that “In general strings are slightly softer than the woodwinds,
and these are softer by about 10 dB than brass instruments” (Meyer, 203). These
measurements were evaluated by the average of two kinds of performance situations: fast
scales and individual tones. However, this is a general overview and a more detailed analysis
can be found in Dynamic Spectrum Changes of Orchestral Instruments by David A. Luce.
His research is not only based on the dynamic range of individual orchestral instruments, but
included analyses of the dynamic range of each instrument at different pitch registers. These
results complicate the implementation of an accurate dynamic range for instruments into a
virtual orchestra, as the dynamic pitch - dynamic sound level according to instrument pitch -
of each orchestral instrument varies dramatically and no computer orchestration text, as far as
the author’s research extends, stipulates the implementation of these variables into a virtual
orchestra environment.
Sundstrup 37
It is uncomplicated to allocate a general dynamic range of each instrument within a virtual
orchestra environment. However, to apply the appropriate degree of dynamics relative to the
pitch of an instrument is more complex. For example, the dynamic range of a flute has a
much greater sound intensity in its higher register than in its lower register. Accordingly, an
analysis of its dynamic pitch reveals that a flute producing a dynamic of fortissimo in its low
register is actually softer than producing a pianissimo in its high register.
This cannot be achieved by using MIDI velocity messages alone, as they are only able to
allocate a general level of volume between 0 and 127. Consequently, if the MIDI protocol is
used to adjust the relative dynamics based on the range of the instrument within Sibelius, the
incorrect timbre patches will be triggered. This can result, for example, in a pianissimo tone
programmed to playback in the high register actually triggering a mezzo forte or forte sample
due to the MIDI velocity dictating a certain dynamic which is not pitch specific. Therefore,
another method of allocating the dynamic pitch for each instrument must be used and is
discussed in Part 2 of this thesis.
1.5.3. Hall Resonance and Instrument Localization
The final component of convincing orchestral simulation by FATSO concerns spatialisation
and the panoramic listening environment. “The importance of placing a sample-based
sonority in a natural and realistic environment is significant” (Pejrolo, and DeRosa, 148). For
listeners to hear an orchestral simulation with convincing realism, they must be put in a
listening environment that imitates that of a live orchestral performance. Two main factors
determine a natural live performance environment: instrument localization – performer
seating positions - and performance space resonance.
Sundstrup 38
To successfully imitate an orchestra in a natural performance space, each virtual member of
the orchestra must be positioned in the same location as in a live orchestra. “The position of
the sections in relation to one another is an emotive and much discussed subject” (Adey, 16).
However, although orchestral seating positions have altered throughout the past, there are
standard orchestral seating configurations that many orchestras use and are accordingly
replicated by FATSO.
Another important aspect of instrument localization regards the directivity of sound. Every
instrument in the orchestra exhibits a different sound radiation pattern that has an enormous
influence on sound source localization. Meyer reveals an important aspect of the directivity
of sound radiation:
An omnidirectional radiation is found only at the lowest frequencies of each
instrument covering about one octave of the fundamentals. […] Discussing all
orchestral instruments it can be said that there is no omnidirectional radiation
higher than 500 Hz.
The higher frequency components of the brass instruments are concentrated in
a narrow angle around the axis of the instrument. The flute acts as a dipole
having the sound radiation from the embouchure and the first open side hole;
there is no omnidirectional radiation. Reed instruments have, at middle
frequencies, a distributed radiation of all open side holes (like loud-speaker
arrays) affecting preferential directions perpendicularly to the axis of the
instrument. The directivity of strings rests on the phase distribution of the
vibrating parts of the body, especially of the belly. Therefore its dependence
on frequency is particularly significant. Even directions of the strongest
radiation vary, depending on the frequency. (Meyer, 205 - 206)
The human ear can localize higher pitched instruments with greater accuracy than lower
pitched instruments. However, there are other factors regarding sound directivity that
contrast this phenomenon. For example, both the trumpet and trombone display a very
narrow radiation of sound towards front stage, and from the audiences listening position,
exhibit a greater direct sound than reflected sound in a medium to large acoustic space – such
Sundstrup 39
as a concert hall. However, the French horn – which directs most of its sound away from the
audience listening position and towards the back of the acoustic space – exhibits a greater
reflected sound than direct sound in the same acoustic space. These differences between the
direct and reflected sounds of instruments not only cause complex timbral transformations,
but also direct the human ear to an instrument’s two-dimensional localization: panoramic
position and panoramic depth. Considering that all the instruments in the VSL sample library
were recorded in a relatively dry space, the spatial reflections of the recording studio were
not captured in the raw samples and subsequently need to be simulated using appropriate
spatialisation modelling – room/hall reverberation replication and harmonic equalization – to
reproduce the natural environment produced by a real acoustic space.
It is worth briefly discussing the psychological influence of stage and hall noise added to an
orchestral simulation. These noises include the sounds emanating from both the performers
on stage and the audience in the hall. The onstage noises include page turns, performers
breathing during musical phrases, floor clatter from feet, seat shuffles, and various random
accidents such as instruments knocking music stands or mutes dropped on the floor.
Audience noises include applause, seat shuffles, crackling confectionary wrapping paper,
audience murmuring, and general hall hum.
Although environment noises can add a further degree of performance realism to an
orchestral simulation, the FATSO performances on the accompanying audio CD do not
include environment noises as it was considered a negative addition to an already high-level
virtual orchestra. The author desires FATSO to gain recognition as a convincing virtual
orchestra through the techniques and performance rules discussed in this thesis.
Sundstrup 40
1.6. Concluding Observations
This overview of sample-based orchestral modelling has introduced the main concepts of
realistic computer generated orchestral simulation. With the tremendous progress of
computing power and the development of sophisticated music applications, the technology
necessary to achieve convincing computer generated orchestral simulations is available and
continuing to improve. However, the essential ingredient for realistic orchestral simulation is
based on methods of humanization using expressive instrument performance rules that -
similar to artificial intelligence programming - are difficult to accomplish effectively.
Furthermore, the idea of FATSO is to perform all sequencing manipulation within a software
notation program as opposed to an audio/MIDI software sequencer application.
Surprising as it may seem, methods of humanization do not necessarily concern what is
added to computer generated orchestral simulations to improve realism, but what is taken
away: precision of performance, and deviation of interpretation. Often computer generated
orchestral performances will play music compositions with perfect timing and intonation,
yielding the synthetic sound associated with many unconvincing orchestral simulations. In his
dissertation Instrument Differences in Characteristics of Expressive Musical Performance,
Timothy Walker clarifies the distinction between a music score and a live orchestral
performance:
A score can be thought of as a “quantized” representation of music, in that
each note has a clearly-defined length (e.g. eighth-notes, quarter-notes) and
categorical or relational instructions for articulation, dynamics, and tempo. A
performance, however, contains continually varying dimensions of relative
onsets, articulation, intensities, tempo, and timbral properties. (Walker, 9)
Sundstrup 41
Although intentional human expressive performance practice – including musical rubato,
tempo fluctuation, and phrasing – is an important element of humanization, the unintentional
deviations in timing, intonation, and dynamic variation, have been found by the author to be
the most significant components of orchestral simulation to emulate realistic orchestral
performances.
Accordingly, Part 2 of this thesis focuses on 'unintentional' expressive deviations – in
comparison to a mathematically precise interpretation - in solo, ensemble, and orchestral
performance. However, it must be acknowledged that the implementation of 'intentional'
expressive performance rules into FATSO would add more realism and musicality to a final
orchestral simulation but requires considerably more research than the scope of this thesis
allows.
Sundstrup 42
2. FATSO Performance Rules
2.1. Introduction
Part 2 of this thesis discusses particular methods of orchestral simulation used by the author
based on Analysis-by-Measurement and Analysis-by-Synthesis. Additional research
undertaken by the author in order to quantify performance data collection - based on audio
recordings by ROSO - determined the initial performance rules developed for FATSO.
Performance data collected from ROSO recordings was analysed and categorized from LND
to MND in relation to timing, intonation, articulation, dynamics, timbre, and performance
transitions. Once the performance data was examined, performance rules were realised within
the Sibelius environment, manipulated using Analysis-by-Synthesis, and subsequently
implemented into a Sibelius score using LPT. Although fine detail in performance practice
was analysed in some areas of data collection, much of the data was further processed to
induce more appropriate performance rules suitable for FATSO.
A summary of orchestral environment simulation is included in Part 2 of this thesis. Although
more concise than investigations into methods of performance expression and instrumental
technique, is a necessary component of this thesis as a matter of thoroughness in the subject
of orchestral simulation. Accordingly, an examination of orchestral layout, simulated
microphone technique, dynamic pitch, and spatialisation are discussed. Detailed findings
concerning dynamic pitch are discussed due to its profound influence on the balance, timbre,
and dynamic colour of the orchestra and individual instrument sections. Furthermore, the
dynamic pitch of an instrument is a rarely discussed topic in the field of virtual orchestras and
is considered by the author to be an important ingredient in successful orchestral simulation.
Sundstrup 43
An investigation into the system of performance data implementation applied by the author
shapes a significant component of Part 2 of this thesis and focuses on the utilization of the
software applications engaged to create FATSO. The systematic procedures used are based
on a chain of performance data manipulations accomplished by each software application
programmed to simulate expressive performance behaviour and instrument modelling
techniques (see fig. 1). Consequently, each software application utilised to shape FATSO -
using unintentional expressive performance rules and instrument techniques - will be
examined. However, only performance rules implemented into FATSO will be discussed in
relation to the data collected from the ROSO recordings.
Fig. 1. FATSO - Systematic procedures flow chart.
It is beyond the scope of this thesis to include the orchestral percussion section in the
performance data collection, analysis, and implementation, as the excerpts recorded by
ROSO for analysis did not include extensive percussion instruments and many of the
performance rules do not apply to the percussion section. The percussion used in
compositions performed by FATSO are allocated basic orchestral timing rules only.
VE
VI GigaPulse FATSO Output
Sibelius Score
Sibelius Dictionary Sibelius SSE Sibelius LPT
Sundstrup 44
The accompanying audio CD of simulated performances includes the following
compositions:
• Concerto Classique: Concerto for harp and orchestra – 21”
• Prelude, Intermezzo & Finale: Work for symphonic wind ensemble – 10”
• Four Bagatelles: Work for medium orchestra – 9:30”
• Theme and Variations: Work for solo harp – 8”
• Voyage: Work for large orchestra – 11”
The submitted works use instrument performance techniques that FATSO is currently able to
simulate and therefore avoid many of the extended techniques used in music composition that
are often employed in musical works created through the twentieth and twenty-first centuries.
However, the works created by the author include many aspects of standard instrument
techniques that provide an excellent challenge for FATSO to display its technical ability.
All of the above listed works composed by the author were performed by FATSO except for
Voyage, which was performed by the standard Sibelius sample playback and general Sibelius
performance rules. This particular computer generated orchestral simulation has been
included on the accompanying audio CD as an example of the basic level of orchestral
simulation expected from current notation programs (see Appendix 4).
2.1.1 Analysis-by-Measurement
The initial performance data collected as a foundation for the expressive rules used in
FATSO were based on recordings of ROSO. The orchestra was recorded performing various
orchestral excerpts using thirty-two microphones - close placement method - into Pro Tools.
Sundstrup 45
Pro Tools is currently the industry standard for digital recording onto computer and displays
detailed graphic information of the recorded sound wave data. Once recorded, the sound
wave data for each instrument and string ensemble section was analysed using waveform
information capturing dynamics, note onset/offset timings, and pitch. Information concerning
performance deviations from a mathematically precise interpretation was then analysed and a
comparison was made between each instrument performance and the original score
information. All instrument and ensemble section deviations were calculated in relation to the
average orchestral pitch core and timing pulse.
Pitch and timing information of each instrument performance was collected by processing the
instrument’s recorded waveform shown in Pro Tools using Melodyne - a pitch and timing
analyses/editing plug-in by the company Celemony. For the purpose of this thesis, Melodyne
was used to analyse and display detailed pitch and timing information of an individual
instrument’s performance. The Melodyne GUI displays all pitch deviations and attack
transients as interpreted from the recorded instrument waveforms shown in Pro Tools and is
the starting point for observing fluctuations in instrument intonation and timing throughout a
performance. Instrument volume information is clearly shown within the Pro Tools
waveforms time-line and is a tremendous source of information regarding variations in
dynamic contrast during an instrument performance. Once the performance data was
collected using Analysis-by-Measurement, the performance information was then analysed
using Analysis-by-Synthesis to adjust the collected data to meet the requirements of FATSO
and its simulated high orchestral standard.
Implementing the raw performance data into FATSO resulted in an average virtual orchestra
simulation due to the high deviations in performance practice observed in the ROSO
recordings. Consequently, the data was refined using Analysis-by-Synthesis to induce a
Sundstrup 46
higher performance standard of FATSO whilst still adhering to the fundamental performance
data and instrument deviations observed. Due to the scope of this thesis and the almost
unlimited extent of performance data analysis possible, only the techniques used to analyse
and implement performance rules into FATSO will be discussed. Therefore, only the general
findings appropriate to the performance rules implemented into a Sibelius score will be
examined in detail.
The following rules have been devised for FATSO based on performance data collected from
the ROSO recordings.
• Timing Rules: instrument specific variables, note duration (onset-offset) and random
timing.
• Intonation Rules: high/sharp, loud/sharp, and random intonation.
• Instrument Timbre Rules: note attack, articulation, velocity control, and timbre
filtering.
• Note Repetition and Legato Rules: VI settings and speed control.
2.1.2. Analysis-by-Synthesis
After the initial performance data was collected using Analysis-by-Measurement, further
analyses were performed using Analysis-by-Synthesis. There were two levels of Analysis-by-
Synthesis used: pre-Sibelius score rules implementation and post-Sibelius score rules
implementation. Pre-Sibelius score rules implementation concerned the analysis and
manipulation of the recorded instrument data prior to importing into a Sibelius score. Post-
Sibelius score rules implementation concerned the analysis and manipulation of the recorded
instrument data after importing into a Sibelius score. If the initial data collected from the
ROSO recordings indicated poor performance aspects concerning instrument technique, the
Sundstrup 47
data was modified before it was imported into a Sibelius score. However, if data that
appeared to be of a sufficient standard to implement into a Sibelius score was found to be
unacceptable when performed by FATSO, it was consequently manipulated using Analysis-
by-Synthesis - post-Sibelius implementation. Many of the rules based on the collected
performance data were developed in both pre-Sibelius and post-Sibelius performance rules
importation.
The development of performance data by Analysis-by-Synthesis is discussed in more detail in
Part 2. In general, Analysis-by-Synthesis was used to adjust events by listening and
determining which rules worked successfully for FATSO. The rules were adjusted to suit the
FATSO environment as some data collected was based on average musicianship and
consequently manipulated to provide a more professional simulation. As a result, the initial
performance data based on observations of the ROSO recordings was manipulated to produce
more constructive performance rules and give FATSO a unique individual character with a
higher standard than displayed in the ROSO recordings. There was of course a fine line
between humanization and perfection, and the author’s goal was to simulate the highest
quality orchestra possible without losing the vital elements of humanization which contribute
to the realism of FATSO.
The precision of the performance rules implemented into a Sibelius score in relation to the
data collected from the ROSO recordings varied considerably depending on Analysis-by-
Synthesis. There were many situations where the data collected from the ROSO recordings
displayed excessive deviations in timing and intonation, resulting in poor performance
technique within the FATSO environment. Consequently, the performance data was
manipulated – by narrowing the deviations data to an acceptable level – to induce an
improved standard of instrument technique. However, the varying extent of performance
Sundstrup 48
deviations observed for each instrument was kept in proportion in most situations. When
occasional instrument performance accidents occurred due to obvious human error, only
some of the data was retained to add infrequent instrument mishaps in a FATSO
performance.
Sundstrup 49
2.2. Performance Data Results
2.2.1. Timing Rules Data
The timing rules used for FATSO include instrument specific variables, note duration (onset-
offset), duration contrast, and random timing deviations. The timing data became so
complicated to analyse that only general rules could be formulated for FATSO. The
recordings highlighted the reality that the deviations of timing within instrument and section
performances were astronomical, even though the recordings appeared adequately performed.
Furthermore, to pinpoint where the deviations occurred - in relation to the ensemble sections,
conductor, and whole orchestra - became impossible. However, deviations between
individual instruments and sections could be analysed to a reasonable extent. For example, if
two players perform a duet together and one player is continually ahead or earlier than the
other, who is accountable for the timing deviations. Add a third player to the situation and the
problem, surprisingly, becomes more complicated to analyse. Consequently, rules based on
timing observations concern both the timing differences between players within sections, and
noticeable timing deviations as a result of specific performance events based on tempo
fluctuation, interval execution, and random timing.
In this thesis, the average difference of all instrument timing deviations from a notated score
– in comparison with each other during the same score events - is referred to as the Timing
Core (TC). The TC can be described as the observed centre of timing as a work progresses
and gives the overall pulse and rhythmic identity. Many of the timing deviations observed
were classified as random, and varied considerably amongst each performer. This appeared to
be a product of the quality of musicianship and instrument performance technique unique to
each player. Consequently, rules were adjusted using Analysis-by-Synthesis to
Sundstrup 50
counterbalance some of the more obvious player orientated performance deviations based on
average musicianship.
Distinctly apparent timing deviations - with reference to the whole orchestra – were observed
during the initial start of a work or section, changes of tempo, tempo acceleration, and tempo
deceleration. The most obvious deviations in timing occurred at the beginning of works, new
movements, and sections of music that contained a change of initial tempo. It was found that
the orchestral musicians took a generous amount of time – depending on tempo – to adjust to
the speed of the conductor and colleagues within the orchestra or ensemble group. Timing
deviations in lower pulsed music were not as audible as in faster pulsed music. However, the
observed data displayed greater deviations in instrument timings in slower pulsed music.
In works of faster pulsed music, the orchestra took up to 4 seconds to find the TC between
musicians and ensemble sections, especially in the brass section that displayed very late
timing deviations. This was most prominent in the horn section. The timing deviations
concerning the brass – in relation to the other ensemble sections - could be described as a
rubber band effect as the core orchestra initially stretched ahead of the brass group causing a
delay in the brass sound compared to the TC. The brass performers would then counteract the
noticed delay by accelerating at a faster tempo than the core orchestra to catch up. There was
also an element of the core orchestra decreasing tempo to wait for the brass to catch up and
was an indication of a psychological collaboration. Although the brass displayed the most
obvious deviations from the TC, all instruments and ensemble sections contributed to the
timing fluctuation around the TC during the beginning of new passages and sections of music
at different tempi.
Sundstrup 51
Similar timing deviations from the TC occurred during tempo accelerations and
decelerations, where the players took some time to find the TC. In the case of decelerations
approaching a fermata, the large deviations in timing were not rectified in any of the similar
situations observed on the ROSO recordings. However, the timing deviations were not large
enough to stand out as poor musicianship, only large compared to timing deviations observed
during settled performance situations. In many cases, the violins and violas – especially the
first violin section - were always ahead in relation to the TC, and more often than not, the
brass – especially horns – were often behind in relation to the TC. The woodwinds displayed
varying degrees of deviation between the timing differences of strings and brass. They also
exhibited the most deviations in timing with relation to the string ensemble and brass
sections.
An important characteristic of timing deviations observed in the ROSO recordings concerned
interval transitions. It was recognised by the author that larger intervals initiated a tendency
for the secondary tone of an interval to be executed late in proportion to the primary tone.
Furthermore, larger intervals caused greater deviations in the delay of the secondary tone in
proportion to the execution speed. Intervals performed at slow speeds displayed greater
timing deviations than intervals performed at fast speeds. However, timing deviations - based
on interval width and speed - were less prevalent in descending intervals than in ascending
intervals. The strings displayed the least amount of timing deviations during transitions
between large intervals. However, the brass – especially the trombones and horns – displayed
increasing timing delays as the interval transitions became larger. The woodwinds displayed
negligible timing delays during transitions until the intervals reached over an octave and a
half. At this point, various timing deviations were introduced – especially in the flutes and
bassoons.
Sundstrup 52
Another cause of delayed timing deviations emerged during the passing of melodies and
countermelodies between both individual instrumentalists and ensemble groups. The timing
delays were caused by the instrumentalists or ensemble sections accepting a musical line later
than it was delivered. As a consequence, a small break was induced between the passing of
thematic material between individual instruments and ensemble sections. It was mostly
noticeable when thematic lines were passed between instrumentalists or ensemble sections
separated by large distances on stage. The woodwinds displayed minor delay times due to the
small distances between each player. However, the brass and strings displayed greater
deviation times between each instrument and ensemble sections due to the greater distance
between them on stage. This was not only evident when observing thematic material passed
from a player in the horn section to a player in the trumpet section, but equally as apparent in
the many situations when the violin section passed thematic material to the cello section. It
was assumed that although the front desks of violins and cellos were relatively close together,
the consequential time deviations observed were a result of the entire ensemble section
causing increasing timing deviations as the desks of both the violins and cellos progressively
increased in distance towards the back of each ensemble section.
The fluctuations in timing deviation during settled passages of music – after the initial starts
of sections without accelerations or decelerations – were measured with reference to the pulse
and tempo of the music. Slower tempi produced greater timing deviations and faster tempi
produced smaller timing deviations compared to the TC. However, this was only observed
during passages of music that lay comfortably within the ability of the performers as timing
deviations became more prominent during technically difficult passages of music – especially
at fast tempi. As a general observation, as more performers were involved in a section of
Sundstrup 53
music, the timing deviations became less audible, even though the timing data displayed an
increase in deviations with the addition of more players.
The complexities of rhythmic placement were another source of obvious timing deviations
observed in the ROSO recordings. In the case of tuplets – the simplest being the triplet - the
timing deviations increased in proportion to the length of the tuplet. Groups of tuplets falling
on one beat - such as a triplet quavers executed in the time of one crotchet – displayed
minimal timing deviations compared to settled time. However, triplet crotchets executed in
the time of two non-tuplet crotchets displayed an increase amount of time between the first
and second triplets within the timing group. This created a distinct timing delay of all tuplets
that were notated in the length of more than one beat. Furthermore, the audible timing delays
during tuplets increased as the tempo and musical pulse decreased.
Obvious rhythmic timing deviations occurred during syncopated passages and were
especially noticeable during repetitive off-beat quavers. However, the timing deviations
compared to the TC were varied among different instruments and ensemble sections. The
deviations of rhythmic placement were much more prevalent between ensemble sections than
instruments within each ensemble. This situation often separated the timing deviations
between the various sections of the orchestra as each performer played with much greater
accuracy within the instrument sections in comparison to the deviations displayed between
each ensemble section. Consequently, the larger deviations in rhythmic timing occurred
between the various orchestral sections rather than between individual players within each
ensemble section.
The instrument specific variables rule contains timing information data based on interval and
rhythmic placement. Larger intervals induced greater time delays on the secondary tones and
Sundstrup 54
were applied to instruments and ensemble sections at different percentages according to data
observed on the ROSO recordings. Complicated rhythmic passages, syncopated beats, and
complex tuplets also induced substantial timing deviations as discussed in the section above.
However, the intricacies of timing deviations caused by difficult rhythmic passages were so
complex that only general rules were developed to integrate into FATSO.
The note duration rule is based on articulation length and includes both random and
prescribed deviations in note duration. All instrument and ensemble sections displayed
random levels of note length deviation throughout a performance. It was also observed that
each player displayed different interpretations of note lengths in general. For example, the
first trumpet often sustained note lengths to their full duration in contrast with the second
oboe that sustained note lengths to an average of three quarters of the note duration. This was
more predominant on shorter note lengths but still obvious at the conclusion of long sustained
notes. Consequently, the note duration rule manipulates the length of sustained notes in
accordance to the observations of the ROSO recordings (see fig. 2).
Note Durations Rule
-100
-80
-60
-40
-20
0
20
40
60
80
100
Pic
colo
Flu
te 1
Flu
te 2
Oboe 1
Oboe 2
Cla
rinet 1
Cla
rinet 2
Bass C
larin
et
Basso
on 1
Basso
on 2
Contra
Bass
oon
Horn
1
Horn
2
Horn
3
Horn
4
Tru
mpet 1
Tru
mpet 2
Tru
mpet 3
Tro
mbone 1
Tro
mbone 2
Bass T
rom
bone
Tuba
Harp
Vio
lin 1
Vio
lin 2
Vio
la
Cello
Bass
No
te L
en
gth
%
Long Notes
Short Notes
Fig. 2. Long and short note lengths of instrument players.
Sundstrup 55
Long notes were considered sustained notes with durations longer than 1 second. For any
note that lasted longer than 1 second, the data collected was based on the note finish time as
if the note was only 1 second, no matter how long the sustained note may have been. Short
notes were considered notes with durations smaller than 1 second and were often related to
various articulation types. The short note data was based on separate non-legato, non-
staccato sustained notes.
The duration contrast rule is based on observations that performers were inclined to lengthen
relatively long notes and shortened relatively short notes. However, this was only observed
during passages including few performers and appeared to be a result of enhanced musical
expression. In general, passages of music performed by massed instruments displayed various
degrees of note shortening, whilst still adhering to proper rhythmic placement.
The random timing deviations rule concerns the fluctuations in timing from the TC for each
instrument and string ensemble section. Through observation of the ROSO recordings, the
degree of general timing deviations from the TC was applied to each instrument in FATSO.
The random deviations were measured in milliseconds and an average of both slow pulsed
and fast pulsed tempo deviations were used as an instrument or string ensemble rule specific
to FATSO (see fig. 3). The allocation of random timing data - initially measured in
milliseconds - was implemented into a Sibelius score as tick based data as described in the
following chapter.
Sundstrup 56
Random Timing Deviations Rule
-100
-80
-60
-40
-20
0
20
40
60
80
100
Picc
olo
Flu
te 1
Flu
te 2
Oboe 1
Oboe 2
Cla
rinet 1
Cla
rinet 2
Bass
Cla
rinet
Bass
oon 1
Bass
oon 2
Contra
Basso
on
Horn
1
Horn
2
Horn
3
Horn
4
Tru
mpet 1
Tru
mpet 2
Tru
mpet 3
Tro
mbone 1
Tro
mbone 2
Bass
Tro
mbone
Tuba
Harp
Vio
lin 1
Vio
lin 2
Vio
la
Cello
Bass
Mill
ise
co
nd
s
Ahead of TC
Behind TC
Fig. 3. Timing deviations of instrument players.
The timing differences between instruments were much larger than expected, as some
instruments displayed deviations of around 100 milliseconds based on the TC. Considering
most people can recognize time delays that are over 40 milliseconds, it could be assumed that
the variations of timing deviation collected from the ROSO recordings would suggest a very
poor sounding performance. However, the timing differences were not as noticeable in the
ROSO recording due to the homogeneous assemblage of all the instrument timing deviations.
There were also occasional deviations of up to 300 milliseconds that were not considered in
the random timing deviations rule as they were a reflection of average musicianship and not
appropriate for integration into FATSO
2.2.2. Intonation Rules Data
The intonations rules used for FATSO include high/sharp, loud/sharp, and random intonation.
The rules were based on data processed in the Melodyne plug-in. The performance data was
analysed according to an orchestra tuned to A440 where the tone A - above middle C - equals
Sundstrup 57
440 Hertz. The Melodyne plug-in clearly displays intonation curves concerning wave data
imported from Pro Tools, and is used to process the wave files recorded by ROSO. Melodyne
displays the volume data as waveforms, note data as pitch charts, and intonation data as
continuous horizontal lines through the volume data waveforms (see fig. 4).
Fig. 4. Melodyne plug-in wave information.
The findings regarding intonation data collected from the ROSO recordings informed that the
MND in intonation deviations were often short term, as performers corrected intonation
problems when time allowed. As a result, the most obvious intonation deviations appeared in
Sundstrup 58
fast passages of music where the performers did not have time to correct any noticeable
intonation problems. The same situation occurred when performing large interval leaps where
the performers only corrected noticeable tuning problems when time allowed, but left the
noticeable intonation deviations in the initial execution of the notes. This observation
concerned the strings, wind, and brass, and was one of the more difficult set of rules to
implement into FATSO.
In this thesis, the average difference of all instrument intonation deviations - in comparison
with each other during the same score events - is referred to as the Intonation Core (IC). The
IC can be described as the observed centre of tuning as a work progresses, and gives the
overall result of all combined tuning deviations. Although it would be easy to analyse
intonation deviations with reference to A 440 – the frequency of initial orchestral tuning – the
continuous changes of the IC throughout a work would make the regular comparison to A
440 unreasonably complex. Consequently, intonation deviations are analysed in relation to
the IC rather than the initial orchestral tuning pitch of A 440.
It was discovered that intonation deviations varied between the different sections of the
orchestra – strings, wind, and brass – in relation to each other. The wind section as an
ensemble group generally played at a higher average pitch than the other orchestral sections.
It was most noticeable when they featured in a musical section that relied on each performer
within the section to keep a steady IC, instigating a gradual increase in pitch due to the
absence of strings to maintain the tuning at the original pitch. However, different musical key
and mode structures caused different levels of intonation deviations between the instruments
and ensemble sections. It was also observed that musical keys that prevented the string
section performers from using regular open strings in a musical passage or movement of a
work, instigated a slight rise in pitch and was most audible when the strings had passages of
Sundstrup 59
music alone, and fixed pitch instruments - such as the harp - entered after a medium amount
of time at the initial pitch where the orchestra originally tuned at A = 440.
The ROSO recordings displayed a minor rise in pitch according to the width of interval
performed. It was observed within the strings and woodwind that for every interval of an
octave higher, there was a general rise in pitch of about 2 cents. The brass exhibited a
general increase of pitch – 2 to 4 cents - for every interval of an octave higher. In the case of
the brass, this was an extensive and noticeable deviation in wide intervals and was
subsequently manipulated using Analysis-by-Synthesis to provide a more appropriate rule for
FATSO. In general, it was observed than rising passages of music tended to increase in pitch
and declining passages of music tended to decrease in pitch.
One of the most noticeable deviations of intonation concerned dynamic levels, particularly
during crescendos and diminuendos. Generally, the brass displayed a large increase in pitch
at higher dynamic levels that was most noticeable during a crescendo leading to a dynamic
intensity of forte or above. This rise in pitch was proportionate to the increase of brightness in
the timbre of each brass instrument. The strings displayed minor changes in intonation based
on dynamic levels. However, obvious rises in intonation occurred under very loud conditions
as a result of string tension within the string ensemble sections – particularly the
contrabasses. The woodwinds displayed a minor decrease in pitch during loud performance
dynamics except for the flutes that displayed an increase in pitch as the dynamic level
increased above forte. The intonation of the flutes was particularly evident and was assumed
to be a product of average instrument technique. Consequently, the intonation data was
manipulated using Analysis-by-Synthesis to generate a more appropriate rule for FATSO.
Sundstrup 60
The remainder of intonation deviations observed were categorized as random. However,
various instrument specific deviations in relation to the IC were used as further intonation
rules in FATSO. For example, the second flute displayed a greater level of intonation
deviation throughout a work than any other woodwind instrument, and to an appropriate
extent was applied to the second flute in FATSO. The instrument specific extents of
intonation deviations observed in the ROSO recordings were used as individual instrument
rules for FATSO. However, the degree of random intonation deviations observed in
individual instruments was limited to acceptable levels of variation more appropriate for
FATSO. It is worth noting that the intonation deviation rules were based on deviations from
the IC. This did not account for the equal temperament scale that has tuning deviations up to
16 cents between some intervals compared to pure intervals. However, the implementation of
the data into FATSO had a positive affect on humanization in the final FATSO performances
due to the minor intonation deviations noticeable between each instrument.
The loud/sharp rule is based on the random intonation rule and adds various degrees of pitch
increase - according to performance dynamic – to each instrument and ensemble section as
discussed in the previous section. The high/sharp rule is based on the random intonation rule
and adds various degrees of pitch increase - per octave interval - to each instrument and
ensemble section according to the data collected from the ROSO recordings. However, the
rule is only applied to dynamics under forte to avoid conflicts with the loud/sharp rule.
The random intonation rule allocates a specific percentage of random pitch deviation to each
instrument and ensemble section. The rule assigns specific degrees of pitch deviation – in
relation to the IC - according to data collected from the ROSO recordings (see fig. 5).
Sundstrup 61
Random Intonation Deviations Rule
-50
-40
-30
-20
-10
0
10
20
30
40
50
Pic
colo
Flu
te 1
Flu
te 2
Oboe 1
Oboe 2
Cla
rinet 1
Cla
rinet 2
Bass
Cla
rinet
Bass
oon 1
Bass
oon 2
Contra
Bassoon
Horn
1
Horn
2
Horn
3
Horn
4
Tru
mpet 1
Tru
mpet 2
Tru
mpet 3
Tro
mbone 1
Tro
mbone 2
Bass
Tro
mbone
Tuba
Harp
Vio
lin 1
Vio
lin 2
Vio
la
Cello
Bass
Pit
ch
(cen
ts)
Sharp (IC)
Flat (IC)
Fig. 5. Random intonation deviations of instrument players.
2.2.3. Instrument Timbre Rules Data
The instrument timbre rules used for FATSO include note attack, articulation, velocity
control, and timbre filtering. Analyses of instrument and ensemble section timbral changes
observed on the ROSO recordings drew attention to the constant fluctuations of sound
variations throughout an entire performance. Timbre rules were designed based on
examinations of particular tone changes due to instrument design, performance technique,
and musical events. The string section of the orchestra displayed the least amount of random
timbral changes due to the fact that the ensemble sections were made up of many individual
players - each with their own unique tone deviations - and when grouped together contributed
to a much more even sound. However, although each individual string player displayed a
large amount of timbral change throughout a performance, the individual string playing data
was not relevant as far as FATSO was concerned as only full ensemble section samples were
used from the VSL library.
Sundstrup 62
One of the more obvious causes of timbral change in all instruments was a result of the
performance dynamics: pianissimo, mezzo forte, fortissimo, etc. In all situations, greater
dynamic levels produced a brighter sound due to the increase of higher overtones in each
instrument according to the volume level. The observed changes of timbre according to
dynamic levels mostly affected the brass, followed by the woodwind and then the strings. The
brass displayed the greatest contrast between low and high dynamic levels - from the warmest
and purest sounds at low dynamic levels to the brightest and hardest sounds at high dynamic
levels. Within the string ensemble sections, the cellos displayed the most audible timbral
changes according to dynamics and pitch. However, each separate string on all the string
instruments possesses its own individual tonal character. The timbral differences between
separate strings on a string instrument are already embedded within the multi dynamic
instrument samples of the VSL library.
It must be noted that the most obvious timbral changes were based on note articulations such
as staccato, détaché, and marcato. This is obviously the whole point of written articulations:
to add colour and expression to a musical work. However, within the various articulations
executed by performers lay micro changes in tone colour and it could be said that no two
similar notes performed by a live instrumentalist ever have exactly the same timbre, unlike
raw articulation samples used in many computer generated simulations. Along with the vast
range of articulations available to change instrumental tone colour – including modern
extended techniques - is the use of mutes within the string, brass, and to a minor extent,
woodwind instruments. The timbral changes influenced by the application of instrument
mutes ranged from string mutes - that reduce the harmonic content of the sound - to the many
varieties of brass mutes that transformed the sound by either reducing or increasing the
harmonic content.
Sundstrup 63
The note attack rule is a simple rule that adds a minor amount of randomization to the attack
transients of an instrument’s initial articulation. It is applied to the beginning of sustained
tones only as the VSL sample library already contains variations of attack embedded in the
instrument samples. Although the sustain samples also have embedded variations of attack
for each sampled tone, a judicious amount of extra attack variation was found to enhance the
naturalness of the orchestral sound due to increased timbral separation of similar sonic
characters.
The articulation rule was developed using the Sibelius dictionary, SSE, and VI settings. The
rule allocates appropriate instrument articulations according to performance information in a
Sibelius score. The rule involves complex matrix settings within VI using speed control,
velocity cross-fading, and cell cross-fading. It was straightforward to allocate articulations
indicated in a score to available articulations in the VSL sample library. However, there were
varying approaches to notated articulations by the performers depending on tempo and style.
For example, strings players may perform a written staccato in different styles based on ease
of bowing technique and tempo indication. Consequently, a single staccato indication could
receive many performance interpretations including staccato, spiccato, and fast/slow portato.
This situation increases the multifarious articulation allocations necessary to program in VI
using cross-fading and speed control.
The velocity control rule involves micro velocity deviations on selected groups of repeated
notes to conform to the data observed in the ROSO recordings. In essence, every articulation
executed by an instrument or ensemble section displayed random deviations of dynamic level
which increased during extremes of dynamics. Small deviations in volume levels were
noticed at moderate performance dynamics – piano to forte. However, larger irregularities in
volume levels were observed at extremes of dynamics – lower than piano or higher than
Sundstrup 64
forte. This was most noticeable in the woodwind section – especially the oboes. However, the
strings displayed small deviations during extremes of dynamics.
The timbre filtering rule is used as both a random rule on similar instruments, and a
prescribed rule to further enhancement timbral changes in relation to dynamic contrast. The
timbre filtering rule improves the aural separation of like instruments - such as two trumpets,
four horns, or three clarinets – by slightly adjusting the harmonic content of each
instrument’s sound at a small random percentage using the low-pass frequency filter in VI.
The same low-pass filter is used to decrease the high frequency components observed in the
brass section during performances at high dynamic levels. In most situations, the brightness
displayed by the brass at high dynamic levels was appropriate and resounding. However,
various stylistic situations were observed in the ROSO recordings where the brass attacks and
sonic qualities were mellowed by the performers’ technique – intentionally or unintentionally
- even at high dynamic levels. This observation was considered worthy to implement into the
FATSO environment in relation to the horns and trumpets depending on the required
performance style.
2.2.4. Note Repetition and Legato Rules Data
The note repetition and legato rules were created using settings within VI appropriate for
natural sounding note repetition and transition (legato) phrases. The sample patches
programmed in VI are based on observations in the ROSO recordings. The data was reduced
to an appropriate simplicity to suit the capability of the VSL sample library and VI controller
functions. The rules are based on matrix settings within VI including dynamic cross-fading,
cell cross-fading, and speed control.
Sundstrup 65
Note repetitions – repeating the same note more than once in sequence – demonstrated a very
high level of articulation variation between each note repeated in a sequence on all
instruments and string ensemble sections. The up and down bowings were clearly evident on
all string ensemble sections as were both double and triple tonguing on the wind and brass
instruments. The alternation of string bowing showed a moderate amount of dynamic contrast
in repeated notes due to the natural tendency of bowing contrasts: heavier down-bows and
lighter up-bows. The wind and brass showed minor changes in dynamic contrast when
performing single tongued repeated notes, opposed to double and triple tongued repeated
notes that displayed tremendous variations in dynamic contrast due to the multi-tonguing
technique. Repeated notes performed using double and triple tonguing showed strong
dynamic pronunciations on the first of each group of two or three notes due to the heavy
accent of ‘Ta’ compared to the following ‘Ka’ as observed in the multi-tonguing note
repetition passages.
It was also noticed that greater speeds of note repetition increased deviations in articulation
and timing. The articulations of fast note repetitions sounded brighter due to the enhanced
harmonic content and less accurate note execution. Other than the noticeable difference
between various string up and down bowings, the string ensemble section displayed fewer
deviations in timbre, articulation, and timing than the wind and brass.
Legato note transitions were of particular interest as the string ensemble sections of the
orchestra displayed semi detached note transitions in legato phrases that stretched over an
octave, which in many cases was a result of the string fingering technique and string cross-
over. Legato passages within small intervals were very smooth and at larger intervals became
noticeably detached when the transitions included various string cross-over fingering.
Passages of music also included elements of portamento during note transitions on the same
Sundstrup 66
string. The wind displayed well connected legato note transitions in both small and large
intervals. However, as the legato intervals expanded, increased timbral formants appeared
between each note and were an audible indication of a live performer. The brass legato
transitions sounded similar to the wind but also included an enhanced delay time between
notes as the legato intervals became larger. There was also a greater dynamic change in the
note formants between legato transitions than the strings or winds.
Sundstrup 67
2.3. Performance Rules Implementation
2.3.1. Sibelius Playback Dictionary and SSE
Before any instrument performance information can be correctly transmitted from Sibelius to
VI, correct macro information - noteheads, symbols, lines, technique, expression and
articulations – must be setup in the Sibelius dictionary. The Sibelius dictionary works in
conjunction with SSE to send pre-programmed performance information/identification to VI.
The Sibelius dictionary allows almost all performance information to be allocated
identifications (IDs) that Sibelius understands. SSE allows these IDs to be interpreted into
MIDI information that VI understands. The Sibelius dictionary also allows controller
information to be categorized and consequently interpreted using SSE settings for successful
communication with VI.
The Sibelius dictionary has six dialogue pages for allocating score events to appropriate
MIDI commands:
• Staff text – for playing instructions that only apply to a single staff.
• System text – for playing instructions that apply to all instruments.
• Staff lines – for lines such as trills, slurs, hairpins etc. that apply to a single staff.
• Articulations – for articulations that apply to both single staffs and all instruments.
• Noteheads – for effects indicated by various types of noteheads such as harmonics or
percussion techniques, etc.
• Symbols – for all graphic instrument, technique and performance symbols.
SSE is a separate Sibelius application that acts as an articulation interpreter between Sibelius
and VI. It is used to set up all peripherals and capabilities of a specific sample library or
Sundstrup 68
synthesiser module, in this case VSL. SSE allows pre-configured settings to trigger the
appropriate articulation patches in VI based on information supplied by Sibelius. All
instrument articulations and performance symbols in a Sibelius score can be interpreted
correctly by VI based on the user settings in SSE. Consequently, every articulation, dynamic,
and performance style for each instrument and string ensemble section must be programmed
within SSE for an accurate interpretation by VI based on Sibelius score information.
An important aspect of both the Sibelius dictionary and SSE is the Sibelius SoundWorld and
sound IDs. Sibelius SoundWorld is a utility that is used to categorize instrument timbres.
Each instrument timbre and articulation is given a unique sound ID such as
strings.violin.staccato, woodwind.flutes.piccolo.flutter-tongue, or brass.trumpets.legato.
Further to generic sound IDs, performance techniques - such as mute, sul ponticello, or
harmonic - can be added to the instrument sound ID as patch changes. Generally, each sound
ID begins with an instrument family group, followed by a particular instrument name, and
when specified, a particular playing technique unique to each instrument. The Sibelius
dictionary allocates sound IDs to every instrument and its available techniques. The sound
IDs are then interpreted into appropriate MIDI commands and key-switches that VI
understands through sound-sets created using SSE (see fig. 6).
Fig. 6. Example of sound ID changes in a Sibelius score.
Sundstrup 69
The scope of this thesis does not allow a detailed examination of the Sibelius dictionary and
SSE. However, for the purpose of understanding their use and integration into FATSO, the
following example shows the process of setting up the playback behaviour of a simple
trumpet articulation.
• An accent is placed on a note in the trumpet staff of a Sibelius score.
• ‘Accent’ is selected in the Sibelius dictionary under ‘Articulations’ (see fig. 7).
• ‘+accent’ is selected in the sound ID change in the first dialogue box.
• The dynamic, attack and duration of the note can be adjusted. However, often these
parameters are left unselected as they are controlled within the VI playback engine.
Fig. 7. Sibelius articulation dialogue page.
Sundstrup 70
• All Wind is selected – that includes trumpet - within the SSE application in the switch
types dialogue box once the instrument has been listed. The sound ID ‘+accent’ is
selected and a key-switch is chosen in the ‘Actions’ dialogue box and allocated an
appropriate key-switch - such as the note F - in an octave out of the trumpet
performance range (see fig. 8).
Fig. 8. Sibelius switch types in SSE.
• VI is then programmed to play the correct accent patch according to the allocated key-
switch set in the SSE switch types dialogue page.
Sundstrup 71
This process is used for all instrument articulations and CCs - such as the modulation and
pitch wheels. Although it appears straight forward, there is a considerable amount of data to
be programmed that amounts to thousands of articulations, CCs, and extra controller
messages. If one articulation is programmed incorrectly, it is possible that the whole database
of programmed information can be harmfully influenced, leading to unexpected notational
misinterpretations throughout the entire orchestra.
2.3.2. Sibelius LPT
At the heart of expressive performance rules implementation into the Sibelius environment is
the Live Playback Transformation feature (LPT). LPT allows the programmer to adjust the
value of every note’s timing and velocity within the score. The transformations can be
applied to individual notes or selected groups of notes and musical phrases as well as
individual notes within a chord. There are two main dialogue boxes that allow adjustments to
selected notes and passages in a score: velocities and timings (see fig. 9a and 9b).
Fig. 9a. LPT velocities dialogue box.
Sundstrup 72
Fig. 9b. LPT timings dialogue box.
With LPT activated in Sibelius, small vertical columns appear above each note processed in a
score (see fig. 10).
Fig. 10. LPT velocity level column.
The columns depict the velocity of each note and can be adjusted in the velocities dialogue
box, Sibelius playback panel, or by selecting with the computer mouse. It is also possible to
glide over a series of notes using the mouse to add velocity curves. Unfortunately, timing
adjustments are not indicated by columns and need to be manipulated using the timings
dialogue box or in the Sibelius playback panel (see fig. 11).
Sundstrup 73
Fig. 11. Sibelius playback panel.
The velocities dialogue box, timings dialogue box and Sibelius playback panel affect all
selected notes and passages in a score.
LPT also allows MIDI controller information to be added to a Sibelius score in real time by
over-dubbing the notational information with real-time expressive performance data using
MIDI controller messages and Sibelius flexi-time input. This allows real-time input of
controller information able to manipulate intonation, note attack, cross-fading, and other
expressive information based on the data collected from the ROSO recordings.
Sundstrup 74
2.3.3. VI Performance Control
The Vienna Instruments playback engine allows sample assignment via two dimensional
matrices, dynamic cross-fading, cell cross-fading, and speed control. These powerful
functions give precise control over the way instrument samples are managed and provide a
number of complex sample manipulation controls to alter and combine instrument patches
with detailed precision.
Every VI matrix has 144 cells that can each load up to two instrument articulation samples.
Appropriate articulation samples are then triggered by incoming MIDI information as
programmed with SSE and the Sibelius dictionary. Each matrix cell can be triggered by
various events including key-switches, dynamic cross-fade, cell cross-fade, and speed
control. Key-switches trigger cells according to appropriate MIDI commands programmed in
SSE and the Sibelius dictionary. Dynamic cross-fades trigger cells according to the volume
received MIDI volume of an instrument instruction. Cell cross-fades can be triggered by
various parameters. However, for the purpose of performance rules implementation into
FATSO, cell cross-fades are triggered by the modulation wheel.
Dynamic cross-fades blend samples according to MIDI velocity commands. Many of the
VSL samples were recorded at various dynamic levels from pianissimo to fortissimo. Due to
the varying content of harmonics in each dynamic sample, cross-fading is used to combine
each sample together - by overlapping the patches at selected points - to provide a
homogenous transition from pianissimo to fortissimo. As a result, the contrast between
various dynamic levels remain natural and is most noticeable during the execution of
crescendos and diminuendos where the natural instrument timbral changes occur during the
process of dynamic changes.
Sundstrup 75
The VI cell cross-fade function allows the blending of two inserted patches within a single
matrix cell. For the matrices programmed for FATSO, the second samples in each cell are
either patches with enhanced vibrato or patches with initial detuning at the beginning of
string ensemble samples. The cross-fading is controlled by the modulation wheel and
implemented using flexi-time input.
The VI speed control function allows preset samples to be triggered according to the how fast
individual notes are executed. Speed control is selected to trigger various matrix cells instead
of key-switches and automatically changes patches according to the speed the samples are
triggered. This allows automatic assignment of appropriate patches such as slow legato,
medium legato, and fast legato depending on the speed on the individual notes in a Sibelius
score. The matrices programmed for FATSO include three different articulation speeds of
staccato, legato, sustain, and portato, that are all automatically triggered within VI.
2.3.4. Timing Rules Integration
The timing rules used for FATSO include note duration, instrument specific variables, and
random timing. The rules are based on note start and note finish times as collected from the
ROSO recordings. The performance data for each instrument is compared to both the original
score information and the orchestral performance data. Timing rules integration into a
Sibelius score cannot be applied using real-time controller information as is used to embed
many of the other performance rules. Each note or musical phrase within a Sibelius score
requires separate adjustments using LPT based on the timing rules used for FATSO.
The timing deviations are implemented into a Sibelius score by using either the timings
dialogue box or Sibelius playback panel. There are six available timing transformations
Sundstrup 76
offered by the LPT: scale live durations, set live durations n% of the notated durations,
constant live durations, move earlier, move later, and scale start positions relative to notation.
Scale live durations allows the programmer to adjust the length of all selected notes by an
allocated percentage of their full length: make the selected notes’ live duration longer or
shorter. Set live durations n% of the notated durations allows the programmer to adjust the
length of all selected notes by an allocated percentage of their full length despite any
previously changed timing information on the selected notes. Constant live durations sets the
duration of all selected notes to the appropriate number of ticks - where one quaver equals
128 ticks. Move later allows the programmer to adjust the start time of all selected notes to
sound later by a prescribed amount of ticks. Move earlier allows the programmer to adjust the
start time of all selected notes to be earlier by a prescribed amount of ticks. Scale start
positions relative to notation allows the programmer to allocate note start positions - later or
earlier – without affecting the note finish position. The programmer can also decide whether
to keep the durations the same or only affect the start of selected notes.
The timing measurements, based on ROSO recordings, were converted from milliseconds to
ticks. However, whilst milliseconds are a segment of time, ticks are a division of notes where
one crotchet is always equal to 256 ticks. Consequently, when deviations are measured in
ticks, slower tempi will cause greater timing deviations in milliseconds as each tick will last a
longer amount of time. Faster tempi cause smaller timing deviations in milliseconds as each
tick will last a smaller amount of time. This situation worked well for integrating timing
deviations into a Sibelius score, as it was found in previous research that timing deviations in
orchestral performances became greater as the tempo and pulse became slower in passages of
steady time. Consequently, the timing conversion was based on observations of timing
deviations in orchestral performance at a standard tempo of 120 beats per minute where one
Sundstrup 77
crotchet is equal to 500 milliseconds and converted to 256 ticks. Consequently, all timing
deviations measured in milliseconds were implemented into a Sibelius score as half the
numerical value when converted to ticks.
It was clear that at reduced tempi, each tick was going to have a more obvious effect on the
timing and was - to an acceptable extent - in line with the observation of the ROSO
recordings. However, at tempi slower than 50 beats per minute, it was necessary to limit the
scale of timing deviations to a smaller quantity, as the deviations implemented as tick based
information became too obvious, and not in line with the data collected from the ROSO
recordings.
For timing deviations observed in wide intervals, both the primary and secondary notes were
manipulated. The length of the primary note was expanded whilst the start of the secondary
note was delayed and lengthened in proportion to the data collected from the ROSO
recordings. This was only implemented in situations of single interval transitions as a series
of intervals caused unacceptable delay times based on the accumulated timing deviations.
For timing deviations observed at the beginning of sections, tempi changes, and during
accelerations and decelerations, the timing deviation of each note for all instruments and
sections was adjusted by a small a random amount. The breadth of deviation was
progressively narrowed throughout the duration of a bar to simulate the time it took for
players to successfully find the TC as observed in the ROSO recordings. The same method
was used for timing deviations observed during accelerations and decelerations.
The Random timing rules were applied to each note in a Sibelius score with timing and note
duration information based on the random selection of numbered balls picked out of a bucket.
Sundstrup 78
The number of balls available was selected in accordance to the random deviation parameters
given by the various timing rules. For example, an instrument with a random timing deviation
of between -10 milliseconds behind and +40 milliseconds ahead of the TC, would have two
sets of balls in the bucket: ten balls numbered from zero to minus ten, and forty balls
numbered from zero to forty. For each note that was allocated a random deviation, a ball was
arbitrarily selected from the bucket to give the applied parameter.
However, in reference to the observed fluctuations in timing deviations from the ROSO
recordings, the random selection of balls was limited to a 10 millisecond maximum deviation
(+/- 10 milliseconds) for each random selection to prevent unnatural and non-contextual
increases in timing deviations. This method was used for all random timing rules and was
implemented by adjusting note start and finish times.
2.3.5. Intonation Rules Integration
The intonation rules were implemented into a Sibelius score using flexi-time input which is a
real-time MIDI input system within Sibelius using LPT. One great feature of flexi-time is the
ability to add controller data such as pitch-bend, modulation, dynamics, and any other
controller data recognized by a VI. It is possible to over-dub an instrument part with
controller data without changing the notation and symbol information in a Sibelius score. All
intonation rules were implemented into a Sibelius score using this technique except for
instrument pitch transformations programmed in VI. The instrument pitch transformations
allocate a slightly different pitch for every instrument and string ensemble section for the
entire work performed. Each instrument pitch change was allocated based on the random
intonation data collected from the ROSO recordings.
Sundstrup 79
The random intonation rules were applied to a Sibelius score based on the data collected from
the ROSO recordings. The pitch wheel was programmed to send controller information
within the random deviations data set for each instrument. The percentage of random pitch
deviation – measured in cents - allocated to each instrument was programmed in the
performance control settings in VI. The degree of intonation deviation placed in a Sibelius
score is controlled by the pitch wheel according to the intonation performance rules. Every
instrument in a Sibelius score was allocated a micro variation of pitch change according to
the random intonation rules. They followed the observations of general pitch measurements
such as woodwinds sounding higher in pitch more often than the rest of the orchestral
sections. The appropriate pitch changes - based on data collected from the ROSO recordings
– were programmed in the performance control settings of VI in the master tuning selection
box.
2.3.6. Instrument Timbre Rules Integration
The instrument timbre rules used for FATSO include note attack, articulation, velocity
control, and harmonic filtering. Note attack, velocity control, and harmonic filtering is
applied to a Sibelius score using real-time controller data. Articulation rules are implemented
through VI settings as discussed at the beginning of this chapter.
Variations in note attack, velocity control, and harmonic filtering were implemented into a
Sibelius score using preconfigured controller information based on the data collected from
the ROSO recordings. Three separate controllers were used in real-time to add the
performance information to a Sibelius score using flexi-time input. All general information
was based on random micro level deviations. However, the amount of random variation
changed according to the performance data collected as discussed in previous chapters.
Sundstrup 80
2.3.7. Note Repetition and Legato Rules Integration
The note repetition and legato rules used for FATSO are based on VI settings and speed
control. Both note repetition and legato are often the most obvious indications that an
orchestral performance is simulated. VSL has overcome difficulties in convincing legato
performance simulation by providing complete samples including the legato note transitions
in four contrasting dynamic levels including intervals from a semitone to over an octave.
Consequently, during FATSO playback, all legato passages – within an octave - are true
samples of legato phrases by both individual instruments and string ensemble sections. The
problems with note repetition have also been overcome by the use of several different
samples for each single note. Therefore, each repeated note is a different sample and not just
a clone repetition of a single sample that often produces the 'machine gun' effect. This adds
the subtle variations necessary to achieve a high-level simulation of repeated notes such as a
sequence of semiquavers at the same pitch, as each semiquaver will display slightly different
sonic qualities as observed in live performances.
VSL has sampled legato intervals covering both slow and fast performance intervals. For
FATSO, long note phrases use a legato sample patch with progressive vibrato and as the
phrases become quicker, the vibrato sets in earlier according to the speed control settings.
Speed control is a powerful utility in VI that allows the user to program automatic patch
changes according to the speed of the notes being performed. Consequently, as notes become
faster, VI automatically switches to a faster sampled legato patch to achieve maximum
realism based on the settings programmed by the user.
Speed control is also used for note repetition to automatically switch samples according to
note speed. For example, staccato notes performed by a string section can be programmed to
Sundstrup 81
automatically switch patches from a slow staccato, to medium staccato, and then a fast
spiccato according to the speed of the notes. This worked very well in conjunction with
Sibelius as VI automatically adjusts to tempo changes and the speed of note transitions
without any further performance manipulation within Sibelius.
Sundstrup 82
2.4. Creating a Standardized Orchestral Environment
2.4.1. Orchestral Layout
To simulate an accurate orchestral sound emanating from a stage, the seating positions of
each virtual instrument player and sections within the string ensemble – first violins, second
violins, violas, cellos, and basses – are placed in an appropriate position within the orchestral
sound field. Once the initial sound source allocations are positioned according to a chosen
standardized orchestral layout, the panoramic reproduction of musicians seated in the
appropriate positions on stage are then processed by convolution reverberation using
GigaPulse - including virtual microphones that reproduce simulated equivalents to a variety
of selected acoustic microphones. The positions allocated on stage have both width and depth
dimensions as sound sources display incremental timing delays in relation to the distance
from front stage.
FATSO uses an orchestral seating setup currently used by many orchestras in the world (see
fig. 12). The main variations in the stereo image concern the percussion that are placed in a
variety of different positions at rear stage - determined by the work performed, logistics
decided by the percussion leader, and conductor preferences. Many other seating
arrangements can be used by FATSO. However, the standard international orchestral seating
allocations give a stereo image experienced by most people who listen to orchestras. When
soloists are accompanied by FATSO, they are placed front/middle stage next to the conductor
position.
Sundstrup 83
1 – Violins, 2 – Violas, 3 – Cellos, 4 –Contrabasses, 5 – Harp, 6 – Flutes, 7 – Oboes,
8 – Clarinets, 9 – Bassoons, 10/11 – Trumpets, 12/13 – Horns,
14/15 – Trombones, 16 – Tuba, 17 – Timpani, 18/19 – Percussion
Fig. 12. FATSO orchestral seating arrangement.
VE is used to host each VI and includes dedicated panning capabilities for each sample
source using the powerpan plug-in. Not only can each instrument be placed anywhere within
a stereo field, the width of each instrument’s stereo sample can be increased or decreased to
further enhance the sound properties contained in the instrument’s sonic image. To maximize
the affect of reverberation that GigaPulse has on processing the stereo output from VE, the
stereo width of each sample is adjusted according to the various amounts of direct sound that
radiates from each instrument and string ensemble section. Accordingly, instruments that
display a greater direct sound compared to reflected sound – such as trumpets and trombones
- are allocated a narrow sound width within the stereo field. Instruments that display a greater
reflected sound compared to direct sound – such as French horns and tuba – are allocated a
broader sound width within the stereo field. Each string ensemble section has a balanced
direct and reflected sound, which would suggest a neutral allocation of stereo sample width.
However, due to the size of string ensemble sections and the large space they occupied on a
Sundstrup 84
virtual stage, a large stereo width was found by the author to create a much more natural
placement of string ensemble sections within the panoramic stereo image (see fig. 13).
Fig. 13. Powerpan placement of the viola section.
As mentioned in Part 1, each string ensemble section is based on one stereo sample image
and not on separate sample images for each individual player within the section.
Consequently, the sound emanating from the occupied area of stage by one string section
sample is adjusted to cover a wide stereo width to imitate the large area covered by a string
ensemble section sound source. Consequently, the sound of a string ensemble section is not
located at one point on stage as would be a solo instrumentalist, but spread over the area the
ensemble section occupies on a real stage.
Sundstrup 85
2.4.2. Spatialisation
In the FATSO computer setup, both VE and VI are run on a server computer with a 64 bit
operating system. Although VE can host effect plug-ins, they need to be 64 bit compatible
and currently there are no major convolution reverberations available that are 64 bit
compatible. Consequently, GigaPulse is used to simulate hall spatialisation through the
Sibelius effect inserts on the main Sibelius computer which runs on a 32 bit operating system
and can host 32 bit VST plug-ins such as GigaPulse (see fig. 14). Version 5 of Sibelius can
only use four separate effect plug-ins on each of the available audio buses. One further plug-
in can be inserted on the main output bus in Sibelius but is usually reserved for mastering
effects.
Fig. 14. Basic GigaPulse graphic user interface.
To simulate the natural space experienced in a concert hall, four separate instances of
GigaPulse are used to create both hall reverberation and instrument depth - how far an
Please see print copy for image
Sundstrup 86
instrument section is from the listening source. For example, from an audience listening
position in front of an orchestra, the percussion will sound further away than the strings and
must consequently be positioned using appropriate reverberation settings.
Each instance of GigaPulse can render only one instrument or ensemble section on a virtual
stage. Consequently, the four available instances of GigaPulse within Sibelius are
programmed to simulate the spatial location of strings, woodwind, brass, and percussion. The
round icons in the GigaPulse interface represent the stage microphones and can be set from a
prescribed distance from the sound sources. The square icons represent positions on a virtual
stage where live impulse responses (hall samples) were recorded for the convolution
reverberation used in GigaPulse. The stereo pair of left and right microphones - L in yellow
and R in blue – are each linked to one chosen position on stage each. The distance between
the yellow and blue positions on stage represent both the spread across stage and depth down
stage. Consequently, the stage positions for FATSO are selected as a representation of where
each of the main four orchestral sections are placed in a live performance environment (see
fig. 15a, 15b, 15c, and 15d).
Fig. 15a. Woodwind section. Fig. 15b. Brass section.
Sundstrup 87
Fig. 15c. Percussion section. Fig. 15d. String section.
A medium hall simulation was chosen to process the orchestral stereo image as larger
simulated hall environments hindered the clarity and sonic qualities of both individual
instruments and string ensemble sections within the orchestra.
In addition to placing sound sources in various locations on a virtual stage, parameters can be
further adjusted to fine tune hall resonance, creating a natural orchestral sound and sonic
placement. One powerful feature of the GigaPulse application is the ability to easily adjust
the perspective - sound source distance from the virtual microphones – of each instance of
GigaPulse. This is extremely useful to further control the depth of each orchestral section on
the virtual stage and is particularly useful for sonically locating the orchestral percussion
section.
2.4.3. Simulated Microphone Technique
It is obvious that any audio recording of an acoustic instrument or ensemble must have been
sampled by a microphone or pickup at the first stage of capturing the sound events. The main
consideration of microphone choice rests on capturing a pure representation of all the
orchestral instruments with minimum background noise. Some sample libraries – such as the
Sundstrup 88
East West Symphonic Orchestra – include the hall reverberation in the samples. However, the
VSL instruments and ensemble sections were recorded with minimum ambiance and require
further processing with a simulated reverberation.
Although the VSL samples can be directly processed with hall simulation using GigaPulse,
one more step – often overlooked in orchestral simulation – in producing a virtual recording
regards the type of virtual microphones used to record the simulated orchestra. GigaPulse
includes a unique addition to its convolution hall simulations called ‘Microphone
Replacement’. A change of colour – similar to using selected virtual microphones - can be
attained by utilizing an equalizer plug-in but requires time consuming adjustments and a
further load on computer resources. The GigaPulse microphone replacement utility does not
cause a significant load on the computer and is embedded within the GigaPulse convolution
plug-in.
The microphone replacement utility lists two sets of microphones: the original microphone
used, and the virtual replacement microphone. The microphone selected in the original
microphone list gives GigaPulse the characteristics of the original microphone used for
recording. The microphone selected in the microphone replacement list produces the
characteristics of the microphone modelled by GigaPulse. Microphones display various
frequency responses and sound colours depending on brand and model. Although inserting an
additional process into the chain of orchestral simulation may appear to be another
opportunity for audio degradation, using only a stereo pair of virtual microphones - based on
the characteristics of real and existing microphones – can add a more natural sound to the
processed sample instruments depending on the models chosen. For FATSO, the original
microphones were set to neutral – to retain a flat frequency response - as the original VSL
samples were recorded with a transparent sound. The selected replacement microphones are
Sundstrup 89
Neumann U87s and are used as a stereo pair. The U87 gives FATSO a warmer sound without
audible injury to the sonic clarity.
2.4.4. Dynamic Pitch
Initial data concerning the general dynamic range of instruments was organized based on the
softest and loudest volumes displayed in dynamic experiments concerning volume intensities
of each instrument. The data was based on the softest and loudest possible tones at any
appropriate pitch to achieve the greatest dynamic contrast. The input levels for each
microphone were set equal – based on the loudest combined sound of the entire orchestra.
This allowed accurate dynamic comparisons between each instrument and ensemble section –
without distortion - for integration into FATSO. The dynamic levels observed could not be
used as an indication of the sonic power of each instrument as each instrument’s sound
intensity diminished in proportion to distance. However, the volume levels for each
instrument could be analysed in relation to each other and the dynamic pitch spectrum could
be evaluated with an acceptable precision appropriate for implantation into FATSO (see fig.
16).
General Dynamic Range of Orchestra Instruments
0
10
20
30
40
50
60
70
80
90
Flute Oboe Clarinet Bassoon F Horn Trumpet Trombone Tuba Violins Violas Celli Basses
Volu
me (
dB
)
Dyn. Range
Fig. 16. Dynamic range of orchestral instruments.
Sundstrup 90
The Dynamic Pitch for each instrument player in the orchestra was based on the softest and
loudest tones –within musical limits – at two dynamic levels: pianissimo and fortissimo.
These were performed at five pitch register levels relative to each instrument: very low, low,
medium, high, and very high. The sessions were recorded under the same conditions as the
orchestral recordings but without other players present. The findings were graphically
represented on charts for each instrument (see Appendix 2). The instrument dynamic levels
were most intense between the middle and high pitch registers specific to each instrument
except for the cellos and basses, which generally displayed a decrease in sound level relative
to an increase in pitch (see fig. 17a and 17b).
Dynamic Pitch - Celli
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Basses
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Fig. 17a. Dynamic pitch of cellos. Fig. 17b. Dynamic pitch of basses.
The general dynamic range of each instrument was adjusted within VI based on data
collected from the ROSO recordings and used as a standard template for each Sibelius score.
The settings available within VI allowed adjustments to both the lowest and highest dynamic
levels for each individual instrument within the patch control environment. Yet, it was not
possible to allocate specific sound intensities from pianissimo to fortissimo dependant on
pitch. One possible solution was to use a software dynamics compressor/equalizer that could
be set to adjust the resultant sound intensities based on a selected pitch spectrum set for each
individual orchestral instrument. Unfortunately, powerful computer resources were needed to
Sundstrup 91
run a separate software dynamic compressor/equalizer for every instrument in the virtual
orchestra leading to possibly forty-five instances! Furthermore, the results would not sound
natural due to the massive amounts of compression and expansion necessary to alter the
original sample tones to the correct pitch velocity levels. Consequently, each note and phrase
notated in a Sibelius score is manipulated using LPT in Sibelius to achieve an appropriate
dynamic pitch for each instrument according to the dynamic pitch data collected from the
ROSO recordings.
Sundstrup 92
2.5. Conclusion
2.5.1. The Future of FATSO
Currently, FATSO is capable of performing general orchestral scores and has the ability to
render various instrument techniques. However, FATSO is not yet capable of performing
many of the current extended instrument techniques used in contemporary music
composition, and a composer must keep this in mind if FATSO is going to perform a score
proficiently. Unfortunately, this affects the creative process as the composer must keep within
the bounds of the capabilities of FATSO. It will not be long until detailed sample libraries
release new instrument techniques that can be used in a Sibelius score and successfully
rendered by FATSO.
A substantial part of this thesis discussed performance rules implementation based on
unintentional expressive performance practice. As observed when listening to FATSO, the
unintentional expressive rules add considerable humanization and natural warmth to
performances and are fundamentally responsible for the quality of the virtual orchestra.
However, intentional expressive performance behaviour (rubato, musical phrasing, and
vibrato) would further enhance the realism and musicality of FATSO and could convince
even the most experienced musicians that FATSO is a live orchestra.
As observed in this thesis, the performance rules implementation can be tremendously time
consuming, requiring a large amount of controller data based on performance rules formulae
to be added to a Sibelius score after the score has been notated. Consequently, a more
economical method of performance rule implementation should be investigated in the future.
An improved technique of performance rules implementation could be devised using the
Sundstrup 93
Sibelius manuscript language. The Sibelius manuscript language is a programming language
used to write plug-ins for use within Sibelius. The language provides the ability for a
programmer to alter many of the features in Sibelius using a programming environment based
on the Simkin language. Plug-ins could be developed to automatically add expressive
performance data to a Sibelius score and replace the time consuming method of adding
controller data with LPT.
Plug-ins could be utilized to perform complicated score manipulations and add intentional
expressive performance rules (rubato, musical phrasing, and vibrato) along with unintentional
expressive performance rules. Once the plug-ins are completed, any score could be analysed
and processed with expressive performance rules formulae with only one keyboard
command. Consequently, every score written in Sibelius could playback automatically with
expressive performance rules embedded within the Sibelius score without the need to add
controller data manually. Furthermore, future performance rules could also include
performance styles based on musical era, tradition, and even simulations of existing live
orchestras such as the Vienna Philharmonic or London Symphony. Timing plug-ins could be
developed for elastic quantising of individual instrument phrases to simulate macro timing
variations between instrument and ensemble sections. Intonation plug-ins could automatically
add intricate intonation variations based on harmonic intervals and key structures. A dynamic
pitch plug-in could be one of the most powerful tools to process instrument dynamic ranges
according to pitch.
2.5.2. FATSO as a Performance Resource
The author considers orchestral simulation to be a valuable resource for composers of
orchestral music. Although the author believes that live performance is the pinnacle of
Sundstrup 94
performing arts, it is still better to offer opportunities for an audience to listen to new
compositions performed by a simulated orchestra than not hear them at all. Of course, it is
questionable whether people go to symphonic concerts and choose audio recordings based on
the repertoire or the performance ensemble. However, FATSO can offer performance
opportunities for many works composed today that would otherwise never be heard. This
raises the question whether FATSO could become an important recording ensemble available
worldwide, offering worthwhile opportunities for both gifted composers and soloists.
It can be clearly seen that music technology - like all technology - appears to change direction
more quickly than the consumer can master the technology within the trends of the time.
FATSO is a virtual orchestra that is current today. However, 'tomorrow' may not even list
FATSO in the historical documents of 'yesterday'. This situation is the driving force behind
FATSO and its continuing development, and the future may even bring FATSO to the radio
and commercial recording arena as a vital resource for composers using the sound of an
orchestra as their compositional painting pallet.
2.5.3. Concluding Remarks
Composing music for real instruments is a very intricate art and orchestrating the music can
be as challenging as the composition process itself. The creative process involves writing
each note, dynamic, phrase, and information on how each 'real' player is to accomplish an
acceptable execution of each score part. Translating the information designed for real players
into a virtual orchestral environment can be a particularly difficult task. There are two
separate creative processes involved: artistic creativity – writing the music - and
technological creativity – simulating the orchestra. Consequently, a thorough understanding
Sundstrup 95
of music, orchestras, performance practice, and technology is crucial to successfully achieve
a high-level orchestral simulation.
Although the techniques of orchestral simulation used for FATSO differ from the current
methods established within the music industry, it is anticipated that the procedures used -
based on the integration of Sibelius and VI - provide a fascinating glimpse into the future
potential of sequencing music within a software notation application. It is to be hoped that the
approaching advances in music technology can offer FATSO the required resources to
automate all expressive performance rules based on increasing research into orchestral
performance practice and instrument simulation techniques.
Sundstrup 96
Bibliography and Works Cited
Books
Adey, Christopher. Orchestral Performance: A Guide for Conductors and Players. London:
Faber and Faber, 1998.
Applebaum, Samuel. The Art and Science of String Performance. Sherman Oaks, CA:
Alfred, 1986.
Backus, John. The Acoustic Foundations of Music. Great Britain: W.W. Norton and
Company Inc, 1969.
Beauchamp, James W. Analysis, Synthesis, and Perception of Musical Sounds. New York:
Springer Science + Business Media, 2007.
Cann, Simon. How to Make a Noise. Surrey, Eng: Coombe Hill Publishing, 2007.
Cann, Simon, and Klaus P. Rausch. Sample This. Surrey, Eng: Coombe Hill Publishing,
2007.
Del Mar, Norman. Anatomy of the Orchestra. Berkley: University of California Press, 1983.
Duffell, Daniel. Making Music with Samples. San Francisco: Backbeat Books, 2005.
Everest, Alton E. Master Handbook of Acoustics. New York: McGraw-Hill, 2001.
Fletcher, Neville H., and Thomas D. Rossing. The Physics of Musical Instruments. New
York: Springer-Verlag, 1998.
Gerle, Robert. The Art of Bowing Practice. London: Stainer & Bell Ltd, 1991.
Gilreath, Paul. The Guide to MIDI Orchestration. Marietta, GA: MusicWorks Atlanta, 2006.
Sundstrup 97
Hauck, Werner. Vibrato on the Violin. London: Bosworth Edition, 1975.
Howard, David M., and Jamie Angus. Acoustics and Psychoacoustics. Massachusetts: Focal
Press, 2006.
Jonson, Keith. Brass Performance Pedagogy. London: Prentice Hall International, 2002.
Karlin, Fred, and Rayburn Wright. On the Track: A Guide to Contemporary Film Scoring.
New York: Routledge, 2004.
Mazzola, Guerino. The Topos of Music: Geometric Logic of Concepts, Theory, and
Performance. Switzerland: Birkhauser Verlag, 2002.
McGuire, Sam, and Roy Pritts. Audio Sampling: A Practical Guide. Massachusetts: Focal
Press, 2008.
Pejrolo, Andrea, and Richard DeRosa. Acoustic and MIDI Orchestration for the
Contemporary Composer: A Practical Guide to Writing and Sequencing for the Studio
Orchestra. Massachusetts: Focal Press, 2007.
Prager, Michael. Sampling and Soft Synth Power! Boston: Thomson Course Technology
PTR, 2005.
Read, Gardener. Thesaurus of Orchestral Devices. Westport, CT: Praeger, 1953.
Rona, Jeff. The Reel World: Scoring for Pictures. San Francisco: Miller Freeman Books,
2000.
Russ, Martin. Sound Synthesis and Sampling. Massachusetts: Focal Press, 2004.
Seashore, Carl E. Psychology of Music. New York: McGraw-Hill, 1938.
Snoman, Rick. The Dance Music Manual. Massachusetts: Focal Press, 2004.
Sundstrup 98
Strange, Patricia, and Allen Strange. The Contemporary Violin: Extended Performance
Techniques. Los Angeles: University of California Press, 2001.
Weisberg, Arthur. Performing Twentieth-Century Music. New Haven: Yale University Press,
1993.
Journals, Papers, Periodicals and Articles
Belkin, Alan. Improving Orchestral Simulation through Musical Knowledge. 2008.
Benoit, A., and J.P.Chick, Computer Simulation of the Acoustic Impedance of Modern
Orchestral Horns. John Neumann Institute for Computing, 2006.
Bresin, Roberto, Anders Friberg, and Johan Sundberg. Director Musices: The KTH
Performance Rules System. Royal Inst. of Tech., Stockholm, 2000.
Chafe, Chris, Michael Gurevich, Grace Leslie, and Sean Tyan. Effects of Time Delay on
Ensemble Accuracy. Proc. of Int. Symposium on Musical Acoustics, March 31 – April 3
2004, (ISMA 2004), Nara, Japan. Stanford University: Centre for Computer Research in
Music and Acoustics, 2004.
Clark, Melville, and David Luce. Intensities of Orchestral Instrument Scales Played at
Prescribed Dynamic Markings. Presented at the 16th
Annual Meeting,MIT, October, 1964.
Fabiani, Marc, and Andrew Friburg. A prototype for rule-based expressive modifications of
audio recordings. Int. Symposium on Performance Science. Royal Inst. of Technology,
Sweden. ABC , 2007.
Friberg, Anders. A Quantative Rule System for Musical Performance. Royal Inst. of Tech.,
Stockholm, 1995.
Friberg, Anders, Roberto Bresin and Johan Sundberg. Overview of the KTH rule system for
musical performance. Dept. of Speech, Music and Hearing, KTH, Stockholm, 2006.
Sundstrup 99
Friberg, A., J. Sundberg, and L. Fryden. Preferred quantities of expressive variation in music
performance. Dept. for Speech, Music and Hearing, Quarterly Progress and Status Report,
1989.
Goebl, Werner, et al. 'Sense’ in Expressive Music Performance: Data Acquisition,
Computational Studies, and Models, 2005.
Hashida, Mitsuyo, Noriko Nagata, and Haruhiro Katayose. jPOP-E: An Assistant System for
Performance Rendering of Ensemble Music. Proc. of the Conference of New Interfaces for
Musical Expression, New York, N.Y., 2007.
Hellkvist, Andreas. Implementation of Performance Rules in Igor Engraver. June, 2004.
Jaffe, David A., Ensemble Timing in Computer Music. Proc. of the Int. Computer Music
Conference, Paris, France, October, 1984.
Jerkert, Jesper. Measurements and models of musical articulation. Dept. of Speech Music and
Hearing, October 2003.
Langner, Jorg, Reinhard Kopiez, and Christian Stoffel. Realtime Analysis of Dynamic
Shaping. 6th
Int. Conference on Music Perception, Keele, England, August, 2000.
Laurson, Mikael , and Mika Kuuskankare. Instrument Concept in ENP and Sound Synthesis
Control. Sibelius Academy, Helsinki. Journees d’Informatique Musicale, May, 2002.
Livingstone, Steven R., Ralf Muhlberger, and Andrew Brown. Playing with Affect: Music
Performance with Awareness of Score and Audience. Australian Computer Music
Conference, Brisbane. 2005.
Luce, David. Dynamic Spectrum Changes of Orchestral Instruments. 1st Convention of
Audio Engineering Soc., Los Angeles, May, 1975. Moog Music Inc., Williamsville, N.Y.,
1975
Sundstrup 100
Masri, Paul, and Andrew Bateman. Improved Modeling of Attack Transients in Music
Analysis-Resynthesis. Digital Music Research Group, U. of Bristol.
Meyer, Jurgen. The Sound of the Orchestra. J.Audio.Eng.Soc. Vol. 41, No.4, 1993.
Ramirez, Rafael, and Amaury Hazan. A Learning Scheme for Generating Expressive Music
Performances of Jazz Standards. Music Technology Group, Pompeu Fabra University,
Spain.
Strawn, John. Orchestral Instruments: Analysis of Performed Transitions. 78th
Convention of
the Audio Engineering Soc., Anaheim, CA. May 1985.
Timmers, R. On the contextual appropriateness of performance rules. Masstricht:
ShakerPublishing, 2002.
Timoney, Joseph, Thomas Lysaght, and Marc Schoenwiesner. Implementing Loudness
Models in Matlab. Proc. of the 7th
Int. Conference on Digital Audio Effects, Naples, Italy.
October, 2004.
Walker, Timothy M. Instrumental Differences in Characteristics of Expressive Musical
Performance. Ohio State University, 2004.
Widholm, G. The Sound-Characteristic of the Vienna Philharmonic Orchestra Investigated
with Digital Measuring Methods. Presented at the 88th
Convention, Montreux, March, 1990.
Widmer, Gerhard, et al. The Machine Learning and Intelligent Music Processing Group at the
Austrian Research Institute for Artificial Intelligence (OFAI), Vienna, 2006.
Sundstrup 101
Appendix 1 - Glossary of Abbreviations
ADSR – Attack, Decay, Sustain, and Release
CC – Control Change
FATSO – Film and Television Studio Orchestra
GUI – Graphic Use Interface
IC – Intonation Core
IOI – Inter-onset Interval
LND – Least Noticeable Difference
LPT – Live Playback Transformation
MIDI- Musical Instrument Digital Interface
MND – Most Noticeable Difference
ROSO – Royal Oman Symphony Orchestra
SSE – Sibelius Sound-set Editor
TC – Timing Core
VE – Vienna Ensemble
VI – Vienna Instruments
VSL - Vienna Symphonic Library
VST – Virtual Studio Technology
Sundstrup 102
Appendix 2 – Instrument Dynamic Pitch Charts
Dynamic Pitch - Flute
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Oboe
0
10
20
30
40
50
60
70
C2 C3 C4 C5 C6 C7 C8
PitchV
olu
me
(d
B)
ff
pp
Dynamic Pitch - Clarinet
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Bassoon
0
10
20
30
40
50
60
70
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - French Horn
0
10
20
30
40
50
60
70
80
90
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Trumpet
0
10
20
30
40
50
60
70
80
90
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Sundstrup 103
Dynamic Pitch - Trombone
0
10
20
30
40
50
60
70
80
90
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Tuba
0
10
20
30
40
50
60
70
80
90
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Violins
0
10
20
30
40
50
60
70
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Violas
0
10
20
30
40
50
60
70
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Celli
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Dynamic Pitch - Basses
0
10
20
30
40
50
60
70
80
C2 C3 C4 C5 C6 C7 C8
Pitch
Vo
lum
e (
dB
)
ff
pp
Sundstrup 104
Appendix 3 – List of Original Compositions
1. Concerto Classique (2008)
Duration – Approx. 21 minutes
Instrumentation – solo harp and orchestra
CD audio performance – FATSO
2. Prelude, Intermezzo & Finale (2002)
Duration – Approx. 10 minutes
Instrumentation – versions for symphonic wind ensemble, brass band and orchestra
First performance – Kew Band (brass band version) and NSW Police Band
(symphonic wind ensemble version)
CD audio performance – FATSO (orchestral version)
3. Four Bagatelles (2008)
Duration – Approx. 9:30 minutes
Instrumentation – medium orchestra
CD audio performance – FATSO
4. Theme and Variations (2009)
Duration – Approx. 8 minutes
Instrumentation – Eb clarinet, Bb clarinet, bass clarinet & harp
CD audio performance - FATSO
5. Voyage (2004):
Duration – Approx. 11 minutes
Instrumentation – large orchestra
First performance – Orchestra Victoria
CD audio performance – Standard Sibelius sample playback (not FATSO)
Sundstrup 105
Appendix 4 – List of Tracks on Audio CDRecording
Concerto Classique (FATSO)
Track No. 1 – Fantasy 7' 41"
Track No. 2 – Romance 7' 19"
Track No. 3 – Scherzo 6' 02"
Prelude, Intermezzo & Finale (FATSO)
Track No. 4 – complete work 10' 07"
Four Bagatelles (FATSO)
Track No. 5 – Lento 2' 58"
Track No. 6 – Vivace 1' 48"
Track No. 7 – Allegro 1' 45"
Track No. 8 – Andante 2' 54"
Theme and Variations (FATSO)
Track No. 9 – Theme 1' 54"
Track No. 10 – Variation 1 1' 36"
Track No. 11 – Variation 2 1' 42"
Track No. 12 – Finale 2' 55"
Voyage (Standard Sibelius sample playback)
Track No. 13 – complete work 10' 59"
CONCERTO CLASSIQUE for Harp and Orchestra
Leif Sundstrup
2008
Submitted in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
2009
Instrumentation
2 Flutes
1 Oboe
2 Clarinets
1 Bass Clarinet
1 Bassoon
2 Horns
2 Trumpets
2 Trombones
Timpani
Glockenspiel
Xylophone
Tubular Bells
Solo Harp
Violin 1
Violin 2
Viola
Cello
Contrabass
Note on Trills
Oboe
Half- tone trill
Oboe
Whole-tone trill
Transposing score
Moderato (h=104)
CONCERTO CLASSIQUEFor Harp and Orchestra
Copyright © 2008 by Leif Sundstrup
LEIF SUNDSTRUP
1.
Fantasy
Flute 1
Flute 2
Oboe
Clarinet 1in Bb
Clarinet 2in Bb
Bassoon
Horn 1in F
Horn 2in F
Trumpet 1in C
Trumpet 2in C
Trombone 1
Trombone 2
Timpani
Solo Harp
Violin 1
Violin 2
Viola
Violoncello
Contrabass
mp
mp
mp
mp
p
fp
p
p
fp
p
p
fp
p
p
fp
p
p
fp
p
p
fp
p
mf
fp
p
pizz
ff
mp
pizz
f
pizz
f
pizz
f
ff
mp
arco
mp
pizz
f
arco
mp
7
A
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf ff
7
mf ff
7
mf
ff
mp
3
mf
ff
7
mf
ff
7
mf
ff
mp
6
mf
ff
mp
mf
ff
mp
f
ff
mf
ff
mf
ff
mp
mute
f
ff
mp
mute
mf
ff
arco
mp ff
mp
7
arco
mp ff
mp
7
arcomp ff
mp
7
ff div.
mp
ff
mp
2
13
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf espress.
p
mp espress.
mf espress.
p
mp espress.
mp
mp
pp
open
pp
open
mp
p
mp
p
mp
p
mp
p
pizz
pizz
p
3
19B
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
ff
ff
3
33
3 3 7
mp
ff
ff
3 3 3 3 3 7
mp
ff
ff
7
mf
ff
ff
7
mp
ff
ff
7
mf
ff
ff
7
ff
ff
ff
ff
ff
ff
ff
p
ff
7
p
ff
7
p
ff
7
arco
mp
ff
ff
7
arco
mp
ff
ff
7
4
25
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
7
mf
7
mf
7
mf
mp
7
mf
mp
7
ff
fp
mf
7
ff
fp
mp
ff
fp
mp
f
mp
p
7
f
mp
p
7
ff
fp
mp
p
7
ff
fp
mp
7
ff
fp
pizz
mp
7
5
32
C
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
mf
ff
mf
ff
mf
ff
mf
ff
mf
ff
mf
f
ff
f
ff
f
ff
ff
3 3 3 3 3
Timp.
3 3
subito ff
subito ff
div.
subito ff
subito ff
arco
subito ff
6
D39
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
flutter
pp
flutter
pp
flutter
pp
flutter
pp
flutter
pp
ff
ff
ff
ff
ff
ff
MMMOMNMM
f
F§
mf
Ab
L.H. L.H. sim.
A§
EbBb
Gb
pizz
mp
div.
pizz
mp
div.
pizz
mp
div.
pizz
mp
div.
pizz
mp
7
44
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
mp
mp
mp
mp
mp
L.H. L.H. sim.
G§
AbDb
F# E§
F§Cb
C§
EbFb
L.H. L.H. sim.
mp
arco
mp
arco
mp
arco
mp
arco
8
48
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
p
f
p
f
p
f
p
f
p
f
p
f
p
f
p
fp
fp
fp
F§
ff
LMLOLMML mf
F# E§
F§Cb
C§
6 6
pizz
mp
f
(pizz)
mp
f
9
51
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
f
p
f
p
f
mf
f
p
f
p
f
f
p
mp
f
f
p
mp
f
f
mute p
open
f
mute
p
open
f
p
mp
f
f
p
mp
f
mf
ff
EbF#
LMLOLNML
pizz
mp
f
pizz
mp
f
pizz
mp
f
mp
f
mp
f
10
E54
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mf
F§
D§
B§
3
3
3
3
mp
mp
mp
11
58
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
mp
mp
mp
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
E§
Bb
D§Eb
L.H.
E§
B§L.H.
D#
arco
mp
pizz
p
p
p
p
12
63F
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f espress.
f espress.
f espress.
f espress.
f espress.
f
f
ff
f
ff
f
ff
f
ff
f
f
f
arco
mf
f
3
arco
mf
f
3
arco
f
arco
f
arco
f 3
13
71
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
ff
6 6 6
mf
ff
6 6 6
mf
ff
6 6
6
mf
ff
6 6 6
mf
ff
6 6 6
ff
ff
ff
f
ff
f
f
f
mp
ff
3
D§
ff
mp
ff
f
mp
ff
f
mp
ff
mp
ff
mp
ff
14
78G
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
6 6 6
ff
6 6 6
ff
mp
ff
mp
mp
mp
ff
ff
mp
ff
3
B#
mf
L.H. L.H. sim.
Eb
6 6 6 6 6 6 6 6 6
ff
pizz
p
6 6 6
ff
pizz
p
6 6
6
pizz
p
pizz
p
pizz
p
15
83
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
f
mf
6 6 6 5 7
mp
f
mf
6 6 6 5 7
mp
f
mf
5
mp
f
mf 6
6 6 5 7
mp
f
mf
6
6 6
5 7
mf
mp
3
5
mf
mp 3 5
mf
mp
3 5
mp
mp
mp
mp
B§
L.H.
L.H. sim.
Bb
B#
L.H. L.H.
sim. 6 6 6 6 6 6
arco
p
3
arco
p 3
div.arco
p
3
16
87H
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
3
3
3
mp
3 3 3
3
3
3
B§
f
Bb
Fb
§
E# A#
6 6 6 6 6 6 6 6 6 3
gli
ss.
3 3 3 mp
3
3
3 mp
3 3
3
17
92
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
3
mp 3
mp
3
mp
3
Ab
D# §
A# Eb Ab
B§
Bb
E# A#
3
33
gli
ss.
mf
mf
pizz
mf
div.
mf
mf
18
I97
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
mf 3
f
mf
3
mf
3
mf
3
mf
mf
3 3 3 mf 3
3
3
3 mf
3
3
3
3
mf
3
3 3 3 mf
3
mf
3
Eb
E#
Eb
Ab
E# A#
B§Eb
E#
3
3 33
f
f
arco
f
f
f
19
101
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
3 3 3
3 3 3 3
3
3 3 3
3
3
3
3
3 3 3 3
3
3
3
3
3
3
3
3
3 3 3
3
3 3 3 3
3
20
105
poco rit. J Poco meno mosso
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf ff
7
mf ff
7
mf ff
7
f
f
f
f
ff
f
ff
mf
f
mf
f
f
EbA§
ff
B§E§ F#
mp espress.
gliss.
ff
mf
pp
ff
mf
pp
div.
ff
mf
pp
div. arco
ff
mf
pp
arco
ff
mf
pp
21
110
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
BbEb
F§
Ab
3
22
118
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
espress.
poco cresc.
mf
poco a poco cresc.
3 3 3
mp
mf
p
mp
mf
p
div.
mp
p
div.
mp
p
mp
p
23
124
rall.
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f espress.
ff
ff
ff
f espress.
poco cresc.
ff
ff
mf espress.
poco cresc.
f espress.
ff
ff
ff
ff
ff
ff
ff
ff
B§
Bb
ff
gliss.
gliss
.
A§E§
ff
ff
ff
ff
ff
24
K Meno mosso (quasi cadenza)
130
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
MMLOMMMM
mp
ff
mf
F#C#
3
sffz
div.
pp
div.
pp
div.
pp
pizz
mf
div.
ff
pizz
mf
ff
25
136
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
Gb
B§G§
mpR.H. R.H. sim.
poco a poco cresc. A#
26
137
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
F§C§
Bb sffz
mp
Eb
B§
DbFb
nails
27
L Lento (h=60)
139
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mf
mp
mf
mf
mp
mf
p
mf
p
mf
mp
mf
p
mf
p
mf
mp
mp
Gb
NMMOMNMM
D#F#
mp
mp
div.
p
arco div.
p
arco
p
28
142
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
mp dolce
p
p
mp dolce
p
p
mp dolce
p
p
mp dolce
p
p
p
p
p
p
poco a poco cresc.
D§
p
p
pizz
p
p
pizz
29
M145
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
5
f
5
f
5
mp
f
mp
f
mp
mp
mp
mp
mp
f
mp
mp
arco div.
mp
arco div.
mp
mp
30
N148
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
mp
ff
ff
ff
ff
ff
mp
ff
mp
f
mp
NMMOMNMM
D# f
ff
ff
ff
ff
ff
mp
pizz
arco
ff
31
151
accel.
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp poco cresc.
mp poco cresc.
mp poco cresc.
mp poco cresc.
mf
mp
mp
6 66 6
6
pp
pp
pp
pp
mp
pizz
arco
pp
32
O Tempo 1 (h=104)
156
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
f
f
f
MMNOMMML
AbB#
ff
ff
ff
ff
ff
33
159
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mf
L.H. L.H. sim. Eb
6 6 6 6 6 6 6 6 6
ff
ff
ff
p
pizz
p
pizz
34
162
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
6 6 6 5 7
mp
6 6 6 5 7
mp
5
mp
6
6 6 5 7
mp
6
6 6
5 7
mp
5
mp
5
mp
5
B§
L.H. L.H.
sim.
Bb
6 6 6
35
165
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
mf
f
mf
f
mf
f
mf
f
mf
mp
3
3
3
3
mp 3 3 3 3
mp
3
3
3
3
mp
mp
mp
mp
mfB#
L.H. L.H.
sim.
B§
6 6 6 6 6 6 6 6 6 6 6 6
p
3 3 3 3
p
3 3
3
3
arco div.
p
3 3
3
3
36
P
169
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
3
3
3
3 3
3
3
3
f
3
3
3
3
3
3
3
3
f
3
3 3
3 3
3
3
3
f
7 7 3 7
f
7 7 3
7
f
3
3
3
3
3
3
3
3
f
3
3 3
3 3
3 3
3
f
3
3 3
3 3
3 3
3
f espress.
f espress.
fE§
MMLOLMML
Bb
f espress.
f espress.
f
7 7 3
7
arco
f
7 7 7
arco
f
7 7 7
37
173
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
f
B§E§
Eb
pizz
mp
arco
pizz arco
pizz
arco
pizz
mp
arco
pizz
arco pizz
arco
mp
pizz div.
mf
arco
mp
pizz
arco
mf
mp
pizz
mp
pizz arco
pizz arco
pizz
5
mp
pizz arco
pizz arco
pizz
5
38
178Q poco a poco accel.
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
f
f
f
f
f
f
mp
f
3
3 3
Bb
A§B§F#
MMLOLMMM
pizz arco
p
ff
f
pizz arco
p
ff
f
mf
arco
mp
pizz
arco
p
div.
ff
f
arco
pizz
arco
p
div.
ff
f
arco
pizz
arco
p
ff
f
39
184R
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
mute
f
mute
3
BbF§
mf
F# E§
3 3 3 3
3 3
3 3
p
f
p
f
p
f
p
f
p
f
40
190
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
41
197
S h = 144
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
mute
f
open
mute
f
open
f
open
f
open
MMMOMMMM
mf
gliss.
gliss.
Gb
Eb
gliss.
gliss.
E§
pizz
pizz
pizz
42
204
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Tub. B.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
mp
mp
mp
mp
mp
mp
mf
f
MMMOMMMM
mp
mp
arco
mp
mf
(pizz)
mf
(pizz)
43
210
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
mf
mf
f
f
Gb
Eb
E§
pizz
f
pizz
f
pizz
f
(pizz)
f
(pizz)
f
44
216
T
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
ff
ff
ff
ff
ff
ff
MMMOMMML
f
EbBb
div.
mp
div.
mp
mp
arco
mf
mp
45
222
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
mp
f
mp
f
mute mp
f
mute
mp
B§
A§Bb
F#
F§B§
Ab
G#
3 3 3
3
B#
arco
p
mp
f
pizz
arco
p
mp
f
pizz
div.
arco
p
mp
f
pizz
p
mp
f
pizz
(pizz)
46
U
227
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
mf
f
mf
f
f
mp
mp
f
mp
mp
f
mp
open
f
mp
open
mp
mp
mf
mf
B§
Cb D§
MMMOMMMM
mf
EbBb
arco
mp
f
pizz
arco
mp
mp
arco
mp
f
pizz
arco
mp
mp
arco
mp
f
pizz arco
mp
div. col legno
mp
arco
mp
f
pizz arco
mp
col legno
mp
mp
47
232
V
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
48
238
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
MMMOMMMM
ff
gliss
.
G# F#
mf cresc.
ff
5
pizz
mf
ff
mf cresc.
ff
5
pizz
mf
ff
pizz
mf
ff
49
243
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
ff
ff
F§
fff
6 6
ff
snap pizz
arcoff
pizz
fff
arco
ff
snap pizz
arcoff
pizz
fff
ff
snap pizz
arco
ff
pizz
fff
arco
ff
snap pizz
arco
ff
pizz
fff
arco
ff
snap pizz
arco
ff
pizz
fff
50
Largo (q=48)
2.
Romance
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Tub. B.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp espress.
3
mp
mf
MMLOLMML
p
E§
mute div.
pp
nat.
mp
mp
mute div.
pp
nat.
mp
mp
mute div.
pp
nat.
mp
mp
pp
mp
mute div.
pp
nat.
mp
mp
pp
mp
arco
mp espress.
3
mp
mp
pp
mp
51
12poco accel. A Andante (q=72)
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
3
mf
3
mp
3
mp
3
�f
p
MMLOLMLM
GbA§ mp
3 3
mp
pp
mp
p
mf
mf
pp
mp
p
p
mp
p
52
B18
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp espress.
mp espress.
mf
3
mf
3
mf
3
mf
3
mf
mf
mp
mp
mp
mp
G§
Ab
Gb
G§ E§
B§A§
Bb
Eb
5 3
53 3 3 3
p
p
p
pizz
p
pizz
53
24
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp espress.
mp espress.
mp espress.
L.H.
L.H.
mf
F#
F§
Gb
5
6
pizz
mp
pizz
mp
div. pizz
mp
arco
mp
arco
mp
54
29
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
Ab
arco
mp cresc.
arco
p cresc.
mf espress.
mf espress.
55
33
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
5 3
f
5 3
f
5
3
f
3 3
f
3 3
f
f
f
f
ff
mp
Gb
5 3 3
G§
f
3 3 5 3
f
3 3
5
3
arco
mf cresc.
f
f
f
56
38
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mf
3
mf
3
mp
p
mf
3
mf
3
mp espress.
f
mp espress.
f
f
f
A§B§
Bb
Eb mf
F#B§
F§
3
3
G§
E§
B§E§
p
pizz
f
p
p
pizz
f
p
p
pizz
f
p
p
pizz
p
p
pizz
p
57
43
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
mf
mf
mp
mp
mute
mp
open
mute
mp
open
BbEb
E§
B§ C#
arco
p
arco
p
58
C46
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp dolce
3
3
5
p
p
mp dolce
3
3
5
mf
C§
Db
mp dolce
3
3
5
mp dolce
3
3
5
p
p
(pizz)
p
59
48
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
3
3
mf dolce
3
mf dolce
3
D§
Cb
3 fp
3
fp
fp
fp
60
50
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
pp
pp
f
f
f
f
f
f
mf
mf
mf
Bb
Ab
C§
p
f
p
f
p
f
mf espress.
f
arco
f
61
54D
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
MMMOMMMM
A§B§
mf
Bb
§
b
B§
Bb
B§
Bb
3 5 3 5 3 5 3
p
p
p
p
5
p
pizz arco
62
61
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
fp
fp
fp
fp
mp
mp
mp
mp
f
5 5 5
Ab
5 5
mf
fp
mf
fp
mp
5 5 5
mf
fp
mf
fp
mp
5 5 5
fp
mf
fp
mp
5 5 5 5
mf
fp
mf
fp
mp
mf
fp
mf
fp
mp
63
68E
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
MMLOLMLM
Eb Gb A§ mf
G§
Ab
Gb
3 3 5 3
53 3
pp
mp
pp
mp
mf
mp
mf
mp
mf
mp
64
74
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp espress.
mp espress.
mf
3
mf
3
mf
3
mf
3
mf
mf
mp
mp
mp
mp
G§ E§
B§A§
Bb
Eb
3
3 5
6
pizz
mp
pizz
mp
div. pizz
mp
pizz arco
mp
pizz
arco
mp
65
79
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
F#
F§
Gb
mf espress.
mf espress.
66
83
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
Ab
arco
mp cresc.
arco
p cresc.
arco
mf cresc.
67
F87
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
5 3
f
5 3
f
5
3
mp espress.
mp espress.
f
3 3
f
3 3
mf
mf
mf
mf
ff
G§
Gb
G§ E§
B§A§
Bb
5 3 3
3
3
f
p
3 3 5 3
f
p
3 3
5
3
f
p
f
p
pizz
f
p
pizz
68
93
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mf
3
mf
3
mp
mf
3
mf
3
f
f
f
f
Eb mf
E§B§
F#
F§
EbBb
pizz
f
p
pizz
f
p
pizz
f
p
p
p
69
97G
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
mp dolce
3
mf
p
mf
p
p
mp dolce
3
mp
mp
mp
mp
E§B§ C#
C§
arco
p
mp dolce
3
arco
p
mp dolce
3
p
p
(pizz)
p
70
100
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
3
35
3
35
Db
D§
3
35
3
35
71
102
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
pp
pp
f dolce
3
f dolce
3
Cb
C§Eb
mf
Bb
fp
mp
fp
mp
fp
mp
fp
mf espress.
72
104
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
pp
pp
mp
mp
Ab
A§
mp
mp
mp
mp
73
106
H
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
3
3
3
5 9
ff
3
3
3
5 9
ff
3
3
3
5 9
ff
3 3
3
5
9
ff
3
3
3
5
9
ff
3
3
3
5
9
ff
3
3
3
5
ff
3
3
3
5
f
ff
3
f
ff
3
f
ff
3
f
ff
3
BbAb
cresc.
ff
MLMOMNLM
ff
ff
ff
ff
ff
arco
74
111
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff molto espress.
fp
ff molto espress.
fp
ff molto espress.
fp
ff molto espress.
fp
ff molto espress.
fp
ff molto espress.
fp
ff
ff
ff
ff
ff
ff
B§ Cb
f
E#
BbF§
C#G#
Db Ab
f
près de la table
D§F#
mf B§
A§
F# Gb
A§
fff molto espress.
fp
pizz
sfz
mf
arco
ff
mf
pizz
fff molto espress.
fp
sfz
pizz
mf
arco
ff
mf
pizz
fff molto espress.
fp
pizz
sfz
mf
arco
ff
pizz
mf
fff molto espress.
fp
pizz
sfz
mf
arco
ff
pizz
mf
fp
pizz
sfz
mf
arco
ff
pizz
mf
75
rit. 116
I A tempo
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
E§
C§ mp
ètouffez
G§
mute
p
div. arco
pp
nat.
mute
p
div. arco
pp
nat.
mp
arco
div.
p
mp
arco p
mp
arco p
76
q. = 76
3.
Scherzo
124A
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
fp
pp
4 4
f
fp
pp
4 4
f
fp
pp
4 4
ff
f
fp
pp
4 4
ff
f
fp
pp
4 4
ff
f
fp
pp
4
4
mf
fp
pp
fp
pp
mf
fp
pp
fp
pp
mf
fp
pp
f
4 4
mf
fp
pp
f
4 4
mf
fp
pp
fp
pp
mf
fp
pp
fp
pp
ff
ff
4 4
MMLOLLMM
mf F§
ff
p
ff
p
ff
ff
p
ff
p
ff
p
77
131
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
p
mp
p
mp
mp
mp
mp
mp
Db
GbD§
Ab
A§
B§G§
Ab
Bb A§
Db
b
3
G#
mf
pizz
p
mf
pizz
p
div.
mf
pizz
p
pizz
pizz
78
137
B
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
4 4
mp
4 4
Gb
E§
G§
f
D§
B§ Gb
A§ Db
Ab ff
LMLOLMMM
3 3 3 3
4
arco
mp
f
pizz
arco
mp
arco
mp
f
pizz
mp
4 4
arco
mp
f
pizz
arco
mp
mf
arco
f
pizz
mp
4 4
mf
arco
pizz
arco
mp
79
144
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
pp
2
f
pp
2
f
pp
2
f
pp
2
4 4
4
4
B§G§ A§ f
f
ff
arco
f
ff
4 4
f
ff
arco
f
ff
4
4
f
ff
80
151
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
mf
mp
mf
mp
mf
mp
mf
B§Ab Gb
D#E§
G§C#
b
C§ Db
Eb
Bb
81
156
rit.
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
ff
32
ff
32
ff
mp
f ff
3
ff
mp
f ff
3
ff
ff
ff
3
B§ D§G§
ff dim.
Bb DbGb
div.
f
div.
f
div.
f
div.
f
pizz
ff
arco
pizz
ff
arco
82
C Meno mosso (q. = 68)
162
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
p
p
leggiero
leggiero
3
p
leggiero
leggiero
3
p
p
p
mute
p
open
mute
p
open
p
mp
A§D§
leggiero
B§E§
Eb E§
Ab
Bb
Eb
nails
A§
E§3
3
pizz
p
pizz
p
pizz div.
p
pizz
p
pizz
p
83
D166
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
leggiero
mf
G#
G§
GbDb
Eb
Ab
nails
A§D§
mp
arco
mp
arco
arco
84
E Tempo primo (q. = 76)
170
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
ff
mp
ff
mp
ff
mp
mp
mp
mp
ff
ff
Cb
ff
G§ Db
C§ AbB§
Gb
D#E§
G§C#
§b Eb
Db
arco
ff
arco
ff
ff
ff
ff
85
176
F
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
mf
mf
mf
mf
mute
open
mf
mute
open
D§
Bb
G#C#
C§A§
mf
G§ Fb
F§
G#Db
p
p
p
div.
p
p
86
181
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
D§Gb Ab
A§
B§G§
Ab
Bb A§
Db
b
Gb
3 3
mf
pizz
p
mp
arco
mf
pizz
p
mp
arco
mf
pizz
p
mp
arco
pizz
mf
arco
pizz
mf
arco
87
187G
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
mp
mf
p
p
mp
mf
mf
p
mf
p
mf
p
D§�G§�
MMLOLMMM
Ab
C#
C§E§
C#
§ A§
G#
G§
pizz
f
mf
arco
pp
pizz
f
mf
arco
pp
pizz
f
arco
pp
pizz
f
mf
arco
mf
sul pont.
mp
nat.
pizz
f
arco sul pont.
mp
mp
nat.
88
194
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
mp 4
mf
mp
mf
mp
mp
mp
mp 4
Ab
A§
Ab
EbC#
C§E§
C#
C§A§
G#
pizz
mf
arco
p
pizz
mf
arco
p
mp
4
mf
pizz mp
arco
pizz
mf
mp
arco
4
89
200
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
3 fp
f
5
3 fp
f
5
3 fp
f
5
3fp
5
3 fp
5
4
fp
f
4 4
f
fp
f
fp
fp
4
fp
4 4
f
Gb
G§
B§C#F# G#
D#
EbBb
MMMOLMML
gli
ss.
pizz
mf
ff
arco
pizz
mf
ff
arco
4
fp
ff
4 4
pizz
mf
ff
arco
4
fp
ff
4 4
90
206
H
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mf
mp
mf
mp
mf
mp
mp
Bb
mf
B§
E§
ff
pizz
mf
ff
pizz
mf
ff
pizz
mf
pizz
arco
ff
pizz
p
arco
mf
pizz
arco
ff
mf
91
214
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mf
mp
mf
MMMOMMMM
ff
Eb
Ab
E§
arco
p
p
(pizz)
44 4 4
arco
p
p
pizz
4
4 4 4
p
92
I
220
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
f
ff
f
ff
ff
ff
ff
mf
3
f
ff
3
arco
ff
3
ff
3
arco
ff
3
ff
3
93
226
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
f
p
f
p
f
p
f
p
f
p
f
p
ff
ff
f
MMLOLMML
BbEb
94
J230
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
mp
mp
p
p
f
A§
AbDb
Gb
mp espress.
mp espress.
div.
mp
pizz
mp
mp
95
234
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
fp
mp
mp
fp
mp
mp
fp
mp
mp
fp
mp
mp
fp
mp
mp
fp
mp
mp
B§
Bb B§
Bb
96
238
K
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
fp
f
fp
f
fp
fp
f
fp
f
fp
mf
mf
B§
Bb
LMLOLMLM
A§
3 3
3 3
mf espress.
arco
mf
97
241L
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mf espress.
3
3
mp
mp
mf
mp
mf espress.
5
mp
mp
MMLOMMMM
mf
Db
B§
pp
3
pp
3
pp
pp
pp
98
245
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
3
mf
mf
3
mf
mf
3
mf
mp
mp
f
3
mf
mf
3 mf
fp
mf
3 mf
fp
mf
fp
mf
fp
D§
mp
Gb
Ab
Eb
mp
mp
mp
mp
mp
99
250
rit.
M Meno mosso (q. = 68)
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
p
mf
p
3
mf
p
3
p
f
p
f
p
mf
fp
p
mute
open
mf
fp
p
mute
open
f
f
p
B§G§
MMMOMMMM
A§D§ mp B§
E§
Eb E§
Ab
Bb
Eb
nails
A§
E§G#
G§
3
3
mf
fp
fp
pp
pizz
mf
fp
fp
pp
pizz p
mf
fp
fp
pp
pizz div.
p
mp
mf
fp
fp
pp
pizz p
mf
fp
fp
pp
pizz p
100
258N Tempo (q.=76)
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
ff
ff
ff
ff
ff
GbDb
Eb
Ab
nails
A§D§
Cb
ff
G§ Db
C§
arco
ff
arco
ff
arco
ff
mp
arco
ff
arco
ff
101
264
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mute
open
mp
mf
mute
open
AbB§
Gb
D#E§
G§C#
§b
DbEb
D§
Bb
G#C#
C§A§
102
269O
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mp
mp
mf
G§ Fb
F§
G#Db
GbD§
Ab
A§
B§G§
3
p
mf
p
mf
p
div.
mf
p
pizz
p
pizz
103
275P
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mf
mp
mf
mf
mf
mf
p
Ab
Bb A§
Db
b
Gb
D§
MMLOLMMM Ab
C#
C§E§
3
pizz
p
arco
mp
pizz
f
mf
pizz
p
arco
mp
pizz
f
mf
pizz
p
arco
mp
pizz
f
arco
mf
pizz
f
mf
arco
mf
pizz
f
104
282
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
p
p
p
p
C#
§ A§
G#
G§
Ab
A§
gli
ss.
arco
pp
ff
arco
pp
ff
arcopp
ff
arco sul pont.
mf
mp
nat.
ff
arco sul pont.
mp
mp
nat.
ff
105
288
Q
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
mp
f
mp
f
mp
mf
mp
mf
f
mp
mp
f
mp
MMMOLMML
ff
Bb mf
B§
E§
ff
ff
ff
pizz
arco
ff
pizz
p
pizz
arco
ff
106
295
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mf
mp
mf
mf
MMMOMMMM
f
pizz
mf
arco
p
pizz
mf
p
(pizz)
4 4
pizz
mf
arco
p
mf
arco
p
pizz
4 4
mf
p
107
301
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
Eb Ab
E§
ff
3
4 arco
ff
4 3
ff
3
4 arco
ff
4 3
ff
3
108
307
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Timp.
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
f
f
f
f
f
ff
f
p
ff
f
p
f
f
p
f
f
p
ff
f
p
ff
f
p
mf ff
3
ff
f
f
LMLOLMMM
Bb DbE§ A§
109
R
e = q313rit.
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Glock.
Xyl.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
p
ff
ff
ff
p
ff
f
f
f
f
f
f
p
p
ff
Ab
CbGb
B§
G§
D§
p
ff
p
ff
p
ff
p
ff
p
ff
110
S Meno mosso
319
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Tub. B.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
f
ff
ff
f
f
f
f
fffGb
MLLONMNL
BbE# �
G#
ff
ff
ff
ff
ff
111
321
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Tub. B.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
MMMONMMM
112
T Tempo
323
Fl. 1
Fl. 2
Ob.
Cl. 1
Cl. 2
Bsn.
Hn. 1
Hn. 2
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
LLMOLLLN Cb Db
Eb Fb Gb
ff
3
3
3
3
fff
3
3
3
3
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
113
PRELUDE, INTERMEZZO & FINALE for Orchestra
Leif Sundstrup
2002
Submitted in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
2009
Instrumentation
2 Flutes
1 Oboe
1 Clarinets
1 Bass Clarinet
1 Bassoon
4 Horns
2 Trumpets
3 Trombones
1 Tuba
Harp
Violin 1
Violin 2
Viola
Cello
Contrabass
Percussion
Timpani
Bass Drum
Snare Drum
Clash Cymbals
Suspended Cymbals
Tambourine
Tam-tam
Bongos
Glockenspiel
Tubular Bells
Note on Trills
Oboe
Half- tone trill
Oboe
Whole-tone trill
Transposing score
Prelude (q=152)
Prelude, Intermezzo, & Finale
Leif Sundstrup
Copyright © 2002 by Leif Sundstrup
Flute 1
Flute 2
Oboe 1
Oboe 2
Cor Anglais
Clarinet 1 in Bb
Clarinet 2 in Bb
Bass Clarinetin Bb
Bassoon 1
Bassoon 2
Horn 1in F
Horn 2in F
Horn 3in F
Horn 4in F
Trumpet 1 in Bb
Trumpet 2in Bb
Trumpet 3 in Bb
Trombone 1
Trombone 2
BassTrombone
Tuba
Timpani
Bass Drum
Tubular Bells
Harp
Violin 1
Violin 2
Viola
Violoncello
Contrabass
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
mp
ff
ff
ff
ff
ff
ff
ff
ff
3 3
ff
3 3
ff
3 3
ff
3 3
ff
3 3
ff
3 3
ffp
ffp
hard timpani mallets
mp
ff
6 6 6 6
ff
f
f
ff
ff
ff
ff
ff
4
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
ff
3 3 3 3 3 3 3
mp
ff
3 3 3
3 3 3
3
mp
ff
3 3 3
3 3 3
3
mp
ff
3 3 3
3 3 3
3
mp
ff
3 3
3 3 3 3 3
mp
ff
3 3 3
3 3 3 3
mp
ff
3 3 3 3 3 3 3
mp
ff
3 3 3
3 3 3
3
fff
prominantly
3
fff
prominantly
3
ff
3
3
3 3 3
ff
prominantly
3
ff
3
3 3 3 3
ff
prominantly
3
f
3 3 3 3 3 3
f
3
3
3 3 3 3
f
3
3
3 3 3 3
fff
prominantly
3
fff
prominantly
3
fff
prominantly
3
fff
prominantly
3
mp
ff
6 6 6 6
fff
3 3 3
3 3 3 3
fff
3 3 3
3 3 3 3
fff
3 3
3 3 3 3 3
fff
prominantly
3
fff
prominantly
3
2
A
7
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Cym.
S. D.
Tamb.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ffp
ff
ffp
ff ffp
ff
f
3 3 3 3 3 3
ff
f
3 3 3 3 3 3
ff
f
3 3 3 3 3 3
ff
f 3 3 3 3 3 3
ff
fp
ff
fp
ff
fp
f
f
f
f
f
mf
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
f
ff
ff
A#
A§
A#
f
sempre
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
f
sempre
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
f
sempre
3 3 3 3 3 3 3 3 3 3 3 3
3
3 3
3 3 3 3 3 3 3 3 3 3 3 3 3
3 3
f
pizz
f
sempre
3 3 3 3
3 3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
3
11
Broadly
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Cym.
S. D.
B. D.
Tamb.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
fff
ff
fff
ff
fff
ff
fff
ff
fff
ff
fff
ff
fff
ff
fff
ff
ff
ff
ff
ff
fff
3 3 3 3
ff
fff
3 3 3 3
ff
fff
3 3 3 3
ff
fff
3 3 3 3
ff
mf
fff
ff
mf
fff
ff
mf
fff
mf
fff
mf
fff
mf
fff
mf
fff
f
ff
fff
3 3 3
ff
3
3
3
3
3
3
3
3
f
f
C#
D#
Bb
ff
fff
pizz
arco
div.
mf
fff
pp
ff
fff
pizz
arco
div.
mf
fff
pp
ff
fff
pizz
arco
div.
mf
fff
pp
fff
arco
fp
fp
fff
fff
pizz
arco
fp
fp
fff
3 3 3 3 3 3
4
B Mysteriously (q=68)17
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Cym.
T.-t.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
p
mp
pp
p
mp
pp
p
mp
pp
p
mp
pp
pp
pp
p
mp
pp
p
mp
pp
p
mp
pp
p
mp
pp
p
mp
pp
p
mp
pp
p
p
p
p
pp
pp
mf
mp
mf
mp
mf
Db
div.
pp
div.
pp
div.
pp
div.
pp
pp
5
C29 D
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Tamb.
T.-t.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
espress.
5
7
5
mf
espress.
7
5
3
mp
f
espress.
5
mf
p
pp
mf
p
pp
mf
p
pp
mf
p
pp
mf
p
pp
mp
mp
mp
mp
mp
mp
mp
mp
mp
mp
mp
mp
pp
mp
pp
pp
mp
pp
pp
mp
pp
pp
mp
pp
mp
pp
mp
pp
mp
pp
mp
pp
mp
mp
p
thumb roll
p
mf D§
mp
p
pizz
mp
mp
p
pizz
mp
p
pizz
mp
pizz
mp
pizz
mp
6
42
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Sus. Cym.
B. D.
Tamb.
T.-t.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
f ff
mp
f ff
mp
f
ff
mp
f ff
mp
ff
3
mf
solo
espress.
f
mp
f ff
9 5
mp
f ff
mp
f ff
ff
3
ff
3
mp
ff
fff
mp
ff
fff
mp
ff
fff
mp
ff
fff
f ff
f ff
f ff
f
ff
f
ff
f
ff
mp
ff
3
p
p
f
sfz
7 7 7
arco
pp
f
ff
arco
pp
f
ff
arco
pp
f
ff
arco
pp
pizz
mf cresc.
sfz
arco
f
ff
3
cresc.
sfz
arco
f
ff
3
7
E With energy and optimisim ( q q q q = 144 )
54
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Sus. Cym.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
6
3
mp
mf
ff
3 3
6
3 mp
mf
ff
6
3 mp
mf
ff
3
3
3
3
6
f
mf
6
3
3
6
f
mf
6
3 3 6 f
mf
6
3
3
mf
f
3 3
3
3
mf
f
3 3
sfz
mf
f
ff
sfz
mf
f
ff
sfz
mf
f
ff
sfz
mf
f
ff
sfz
ff
sfz
sfz
sfz
sfz
sfz
3
ff
f
6
3
pizz
mf
3
arco
mf
f
6
3
pizz
mf
3
mf
arco
f
6
3 pizz
mf
3
mf
arco
f
3
3
6
pizz
mf
3
arco
mf
f
6
3
3
6
pizz
mf
3
arco
mf
f
6
8
62
F
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
S. D.
B. D.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
3
3
ff
ff
3
3
ff
3
3
ff
3
3
ff
3
3
3
3
ff
3
3
ff
3
3
ff
ff
ff
ff
ff
ff
ff
ff
ff
3
ff
3
3
ff
3
ff
3
3
ff
3
ff
3
3
ff
3
ff
3
3
ff
ff
3
3
ff
3
3
ff
3
3
f
3
ff
3
3
f
3
ff
3
3
f
3
ff
3
3
f
3
rim shot f
3
3
ff
3
3
ff
3
6
ff
3
6
ff
3
3
ff 3
3
ff
3
3
9
G70
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Cym.
S. D.
B. D.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
ff
f
ff
f
ff
f
ff
mp
f
ff
mp
f
ff
mp
f
fp
fp
mp
f
fp
fp
mp
ff
ff
ff
ff
ff
ff
ff
f
ff
f
ff
f
ff
f
f
f
f
ff
f
mp
pizz
6 6
pizz
6 6
pizz
6
6
fp
fp
mp
fp
fp
mp
10
77
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Cym.
S. D.
B. D.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
6
3
ff
3
3
3
ff
3
ff
3
ff
f
3
3
6
f
3
3
5
f
3
3
5
f
3
f
3
ff
6
3
ff
6
3
ff
3
3
ff
3
3
3
mf
3
mf
3
mf
f
sfp
ff
f
ff
f
sfp
ff
f
ff
f
sfp
ff
f
ff
ff
ff
ff
ff
f
ff
rim shot
ff
ff
arco
f
ff
3
ff
6
arco
f
ff
3
ff
5
arco
f
ff
3
ff
5
3
fff
sfp
ff
3
3
3
fff
sfp
ff
3
3
11
H85
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Bongos
Tamb.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
f cantabilé
mp
f cantabilé
mp
f cantabilé
mp
ff
mp
ff
mp
ff
mp
ff
mp
mp
mp
mp
mp
mpmp
mp
mp
medium mallets
mp
mp
mf
Eb Ab
ff
GbDb
p
mf cantabile
mp
p
mf cantabile
mp
p
mf cantabilé
p
pizz
mf
p
pizz
mf
12
92
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Bongos
Cym.
Tamb.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
mp
mp
mp
mf
7
7
7
mf
mf
ff
boldly
7
ff
boldly
ff
boldly
ff
boldly
ff
boldly
mp
mp
mp
mf
mp
mf
mp
mf
mf
Sus. Cym.
mp
G§D§
GbDb
G§
mf
7
mf
7
7
arco
arco
13
97
rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Bongos
Sus. Cym.
Tamb.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
ff
f
ff
3
3
3
6
f
ff
3
3
3
6
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
3
3
f
ff
3
3
f
ff
3
3
f
ff
3
3
f
ff
3
3
f
ff
3
3
3
f
ff
3
3
3
f
f
ff
f
ff
f
ff
3
3
3
f
ff
3
3
3
6
f
ff
3
3
3
6
14
I Heroically ( q = 148 )
105rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
S. D.
B. D.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
ff
3
sfz
3
mf
ff
sfzsfz
3 3
mf
ff
sfz
3 3
ff
p
f
dim.
5
3 3
3 3
ff
p
f
dim.
5 3 3 3
3
ff
p
f
dim.
5
3 3
3 3
ff
p
f
dim.
5 3 3 3
3
ff
3
mute
open
3
ff
3
mute
open
3
ff
3
mute
open
3
ff
3
3
3
ff
3
3
3
ff
3
3
3
mf
ff
3 3
fff
3
3
3
3
3 3
3
3
ff
3
3
3 3
ff
pizz
f
3arco
ff
sfz
3 3 3
pizz
f
3arco
ff
sfz
3 3 3
pizz
f
3
arco
ff
sfz
3
3 3
ff
f
ff
sfz
f
dim.
3 3
ff
f
ff
sfz
f
dim.
3 3
15
J Andante (q = 88)115
rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
pp
p
fp
fp
pp
p
fp
fp
fp
pp
p
fp
fp
pp
p
fp
fp
fp
pp
p
fp
fp
fp
pp
p
fp
fp
pp
p
fp
fp
fp
pp
p
fp
fp
fp
fp
fp
fp
fp
fp
mp
mp
mp
mp
mp
fp
fp
mp
fp
fp
fp
mp
mp
fp
fp
mp
fp
fp
fp
mp
p dolce
ff
f
p dolce
ff
f
p dolce
ff
f
p
ff
f
p
ff
f
16
K poco meno mosso122 rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
Solo
mf espress.
7 7
Solo
mf espress.
p
mp
p
mp
p
p
p
mp
p
p
mp
p
p
mp
p
p
mp
f
Gb
Cb
C§ B§
pp
mp
pp
mp
pp
mp
p
pizz
arco
p
p
pizz
arco
p
17
L Intermezzo (q = 72)132
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
T.-t.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
dolce
3
f cantabile
f cantabile
mp
dolce
3
mp
mp
dolce
3
mp
espress.
3
mf cantabile
mf cantabile
mpespress.
3
p
mp
p
mp
p
mp
p
f
Bb § b
pp
p espress.
3
p
pp
p espress.
3
p
pp
p
p
pp
p
pizz mp
arco
espress.
3 p
pp
p
pizz arco
mp
p
18
146
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
B. D.
T.-t.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
p
p
p
p
p
mp
p
mp
p
mp
p
mp
p
mp
p
mp
mp
mp
pp
6 6 6 6 6 6 6 6 6 6 6 6 6 6
pp
6 6 6 6 6 6 6 6 6 6 6 6 6 6
pp
6 6 6 6 6 6 6 6 6 6 6 6 6 6
pp
pp
19
150
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
T.-t.
T.B.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
pp
mp
f
ff
f
ff
pp
mp
fp
fp
fp
fp
ff
3 3 3 3 3 3 3 3 3
pp
mp
f
ff
3 3 3 3 3 3 3 3 3
pp
mp
fp
fp
fp
fp
ff
pp
mp
p
fp
fp
fp
fp
ff
3 3 3 3 3 3 3 3 3
pp
mp
p
fp
fp
fp
fp
ff
3 3 3 3 3 3 3 3 3
pp
mp
p
fp
fp
fp
ff
fp
fp
fp
fp
ff
mf
ff
mp
p cresc.
f
ff
mp
p cresc.
f
ff
mp
p cresc.
f
ff
mp
p cresc.
f
ff
fp
fp
fp
fp
ff
fp
fp
fp
fp
ff
fp
fp
fp
ff
mf
fp
fp
fp
fp
ff
mf
fp
fp
fp
fp
ff
mf
f
ff
f
ff
soft timpani mallets
pp
medium timpani mallets
ff
hard chime
hammer
ff
ff
pp cresc.
f
ff
6 6 6 6 6 6 6 6 6
pp cresc.
f
ff
6 6 6 6 6 6 6 6 6
pp cresc.
f
ff
6 6 6 6 6 6 6 6 6
mp
f
ff
mp
f
ff
20
M162
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Sus. Cym.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp espress.
mf
rit.
fp
p
7 6
mp
mp
pp dolce
p
pp
mp
pp dolce
p
pp
mp
pp dolce
p
pp
mp
pp dolce
p
pp
mp
pp
p
pp
mp
21
173
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Sus. Cym.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f espress.
9
mp
mf
ppp
mf
3 3 3
mf
mf
mf espress.
7
pp
dim.
pp
dim.
pp
dim.
pp
dim.
pp
dim.
22
N183 rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
B. D.
T.B.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
molto espress.
mp
molto espress.
pp
p
ff
pp
p
ff
pp
p
ff
pp
mp
molto espress.
p
ff
p
ff
p
ff
p
ff
p
ff
p
ff
p
ff
p
soft beater
mp
soft chime
hammer
p
p
p
23
O Finale (q = 60) growing in volume and pressing forward (little by little)195
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
passionately
3 3
mp
passionately
3 3
pp
sost.
3
3
3
3
3
3
pp
sost.
3
3
3
3
3
3
pp sost.
3 3 3 3
p
3
3
p
3
3
p
3
3
p
3
3
p
3
3
p
3
3
sof mallets
mp
mf
Gb
Cb
pp
sost.
3
3
3
3
3
3
pp
sost.
3
3
3
3
3
3
pp sost.
3
3
3 3
pp
pp
24
203
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mf
3
3
3 3
mf
3
3
3
3
3
3
p
3
3 3 3
3
3
p
3
3 3 3
3 3
p
3
3
3
3
3
3
3
3
3
3
mp
C§
mp
A§
AbDb
3
3
3
3
3
3
3
3
3
3
3
3
3
3 3 3
25
P211
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
f
ff
f
f ff
7
f
f ff
7
f
f ff
7
f
f
f
f
f
f
f
f
f
f
f
f
f
3
f
3
f
f
3
f
f
f
3
f
f
3
f
f
3
f
f
f
f
26
Q With brilliance (q = 148)
217
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
S. D.
B. D.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
ff
ff
ff
ff
ff
f
3
f
3
ff
3
3
ff
3
3
ff
3
3
ff
3
3
f
3
ff
f
3
ff
ff
ff
3
3
ff
3
3
ff
3
3
ff
3
3
f
3
3
f
3
3
ff
ff
ff
ff
3
3
ff
3
3
27
R poco crescendo223
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
T.B.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
6
mf
6
mp
5 6
mf
6
mf
pp
pp
mp
mf
pp
mp
mf
mp
mp
mp
mp
mp poco cresc.
mp poco cresc.
mp poco cresc.
mp poco cresc.
p poco cresc.
pp poco cresc.
p cresc.
F# only
pp
6
pp
mf
6
mp
pizz
pp
pizz
pp
pizz
28
S234
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
B. D.
T.-t.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
f
ff
f
ff
f
2
2
2 ff
f
2
2
2 ff
f
ff
f
ff
f
ff
f
ff
f
ff
3
3
77
f
ff
3
3 77
f
ff
f
ff
f
ff
f
ff
f
2
2
2
ff
2
2
2
2
f
2
2
2
ff
2
2
2
2
f
2
2 2 ff
2
2
2
2
f
2
2
2
ff
2
2
2
2
f
ff
2
2
2
2
f
ff
f
ff
3
3
ff
ff
ff
ff
fff
gliss.
gliss.
gliss.
gliss.
f
2
2
2 ff
6 6
f
2
2
2 ff
6 6
f
arco 2
2
2 ff
6 6
f
arco
ff
3
3
77
arco
f
ff
3
3
77
29
244
rit.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
2
2
2
2
fff
fff
fff
fp
3
3
fff
fp
3
3
fff
fp
3 3
fff
fp
3 3
2
2 mf
ff mf
ff
3 3 3
2
2 mf
ff mf
ff
3 3 3
2
2 mf
ff mf
ff
3 3
3
2
2 ff
fp
3 3
2
2 ff
fp
3 3
ff
fp
3 3
ff
fff
gliss.
BbEb
Ab
gliss.
fp
6
3 3 3
fp
63 3
3
fff
fp
6
3 3 3
fff
7
fff
7
30
T Meno mosso ( q = 76 )
252
molto accel. Presto ( q = 180 )
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Sus. Cym.
S. D.
B. D.
T.B.
Glock.
Hp.
Vln. 1
Vln. 2
Vla.
Vc.
Cb.
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
fff
mp
cresc.
ff
ff
mp
cresc.
ff
ff
f
cresc.
f ff
f
cresc.
f ff
f
cresc.
f ff
f
cresc.
f ff
fp
ff
fp
ff
fp
ff
fp
ff
fp
ff
fp
ff
ff
hard beater
fff
fff
mp
cresc.
f
fff
fff
p
3
3
cresc.
3
3
3
3
3
3
3
3 ff
p
cresc.
ff
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
p
cresc.
ff
fff
p cresc.
ff
fff
p cresc.
ff
fff
p cresc.
ff
fff
fp
fp
fp
ff
fff
fp
fp
fp
ff
fff
31
FOUR BAGATELLES for Orchestra
Leif Sundstrup
2008
Submitted in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
2009
Instrumentation
2 Flutes
1 Oboe
1 Clarinets
1 Bass Clarinet
1 Bassoon
4 Horns
2 Trumpets
2 Trombones
1 Bass Trombone
1 Tuba
Timpani
Glockenspiel
Vibraphone
Tubular Bells
Violin 1
Violin 2
Viola
Cello
Contrabass
Note on Trills
Oboe
Half- tone trill
Oboe
Whole-tone trill
Transposing score
Lento (h = 48)
Copyright © 2008 by Leif Sundstrup
Leif Sundstrup
1.
Four Bagatelles
Flute 1
Flute 2
Oboe
Clarinetin Bb
Bass Clarinetin Bb
Bassoon
Horn 1 in F
Horn 2 in F
Horn 3 in F
Horn 4 in F
Trumpet 1 in C
Trumpet 2 in C
Trombone 1
Trombone 2
Bass Trombone
Tuba
Timpani
Tubular Bells
Glockenspiel
Violin I
Violin II
Viola
Violoncello
Contrabass
ff
p
ff
pp
mf ff
p
ff
pp
mf ff
p
ff
pp
6 6
ff
p
ff
pp
ff
p
ff
pp
mf ff
p
ff
pp
6 6
ff
fp
mf
ff
pp
mp
fp
ff
fp
mf
ff
pp
mp
fp
ff
fp
mf
ff
pp
mp
fp
ff
fp
mf
ff
pp
mp fp
ff
fp
mp
f
ff
fp
mp
f
ff
fp
mp
f
ff
fp
mp
f
ff
fp
mp
f
ff
fp
mp
f
ff
pp
ff
mf
3
f
div.
f
mf
ff
p
div.
f
mf
ff
p
f
div.
ff
mp
ff
f
3
ff
mp
ff
f
5
ff
mp
ff
5
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
ff
p
f
ff
mf ff
p
f
ff
mf ff
p
f
ff
6 6
ff
p
f
ff
ff
p
f
ff
mf ff
p
f
ff
6
6
f
f
f
f
f
f
f
f
mute
f
open
f
mute
f
open
f
mp
f sost.
mp
f sost.
f
mp
f sost.
f
mp
f sost.
mp dolce
div.
mf
p
ff sost.
mp dolce
div.
mf
p
ff sost.
p
div.
mp dolce
mf
p
ff sost.
p
mp dolce
mf
p
div.
ff sost.
p
mp dolce
mf
p
ff sost.
2
9
molto rit.
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
mp
ff
mp
ff
3
mp
ff
6 6 6
3
mp
ff
mp
ff
5
mp
ff
6 6 6
sfz
mp
ff
p
sfz
mp
ff
p
sfz
mp
ff
p
sfz
mp
ff
sfz
p
sfz
sfz
mp
ff
p
sfz
mp
ff
p
sfz
mp
ff
p
sfz
mp
ff
p
ff
mf
3 3
ff
div.
sffp
mf
ff
3
div.
sffp
div.
sffp
3 div.
sffp
f
p
5
sffp
f
p
3
A Andante (q = 86)13
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
mf
ff
p
ff
p
6
6
ff
p
6 6
mf ff
p
mf
mf ff
p
mp
f
mf
mp
f
mf
mp
f
mf
mp
f
mf
mute
mf
mute
mf
mute
mf
mute
mf
mute
mf
mf
f
mf
mp
pizz div.
3
mf
mp
pizz div.
3
pizz div.
mp
f
col legnoff
pp
pizz
mp
ff
3
pizz div.
mp
f
col legno
ff
pp
pizz
mp
ff
3
pizz
mp
f
col legno
ff
pp
mp
pizz
ff
3
4
17
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
f
mp
f
flutter
3 3
f
mp
f
flutter
3 3
f
3
mp
f
flutter
3 3
f
mp
f
flutter
3 3
f
3
mp
f
flutter
3 3
f
mp
f
flutter
3 3
mf
mf
mf
mf
open
f
ff
open
open
mf
f
ff
open
open
mf
f
ff
mf
mf
arco
f
ff
mf
3
mf
arco
f
ff
mf
3
mf
arco
f
ff
col legno
ff
pp
3
mf
arco
f
ff
col legno
ff
pp
3
mf
arco
f
ff
col legno
ff
pp
3
5
22
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
f
ff
ff
mf
fp
sfp
ff
f
espress.
mf
fp
sfp
ff
f
espress.
mf
fp
sfp
ff
f
espress.
mf
fp
sfp
ff
f
espress.
ff
ff
mf
fp
fp
ff
f
espress.
mf
fp
fp
ff
f
espress.
mf
fp
fp
ff
f espress.
mf
fp
fp
ff
f
espress.
f
pizz
arco
ff
espress.
pizz
f
arco
ff
espress.
f
pizz
ff
arco
espress.
pizz
f
ff
arco
espress.
pizz
f
ff
arco
espress.
6
29
B
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
mp espress.
Solo
mf
5 7 3
mf
3
pp
pp
pp
pp
pp
pp
pp
pp
motor slow
mp espress.
sfz
7
sfz
6
sfz
pizz
mp
7
sfz
div.
mp espress.
6
sfz
pizz
mp
7
7
35
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
mp
mf
3
p
5 6 3 3 3
mf
f
mf
p
3
f
f
p
mf
p
5
6 3
mf
f
mf
3
p
3
3 3
p
mf
p
mf
p
mp
f
p sub.
mf
p
mp
f
p sub.
mf
p
mp
f
p sub.
mf
p
mp
f
p sub.
mf
p
mf
f
mf
3
p
3 3 3
mf
f
mf
3
p
3
3 3
p
mf
p
p
mf
p
mf
p
p
mf
p
mp espress.
f p sub.
mf
3
p
3 3
mp espress.
f
p sub.
mf
p
div. arco
mp espress.
f
p sub.
mf
p
f
p sub.
mf
p
div. arco
p
f
p sub.
mf
p
8
C43
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Tub. B.
Vln I
Vln II
Vla
Vc.
Cb.
ff
poco a poco dim.
p
ff
poco a poco dim.
p
ff
poco a poco dim.
p
ff
poco a poco dim.
p
ff
poco a poco dim.
p
ff
poco a poco dim.
p
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
mp
mf
ff
mp
mf
ff
poco a poco dim.
ff
poco a poco dim.
div.
f
poco a poco dim.
div.
f
poco a poco dim.
ff marcato
poco a poco dim.
ff marcato
poco a poco dim.
7 7 7 7 7
ff marcato
poco a poco dim.
7 7 7 7 7
9
49
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Tub. B.
Vln I
Vln II
Vla
Vc.
Cb.
pp cresc.
molto
pp cresc.
molto
pp cresc.
molto
pp cresc.
molto
pp cresc.
molto
pp cresc.
molto
ff
ff
ff
ff
mp
mp
mp
mp
mp
mp
p
molto
ppp
ppp
ppp
ppp
7 7 7
ppp
7 7 7
10
Vivace (q.=92)55
2.
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mf
mf
mf
mf
fp
mf
mf
fp
mf
mf
fp
mf
mf
fp
mp
f
mf
fp
mp
f
mf
fp
mp
f
mf
fp
mp
f
fp
fp
fp
fp
mp
fp
fp
fp
fp
mp
fp
fp
fp
fp
mp
fp
fp
fp
fp
mp
fp
fp
fp
fp
fp
fp
fp
fp
mp
f
mp
f
mp
f
mp
f
f
pizz
11
D
71
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mf
f
mf
f
mf
f
f
mf
f
mf
f
mf
f
fp
mf
f
fp
mf
f
fp
mf
f
fp
mf
f
fp
f
fp
mf
f
fp
f
fp
mf
f
fp
f
fp
fp
fp
fp
fp
fp
fp
fp
fp
mf
fp
fp
fp
fp
mf
fp
fp
fp
fp
arco
mf
fp
fp
fp
fp
12
E
78
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
mp
mp
mp
mf
mp
mf
mp
mp
mp
mp
mp
mf
mp
mf
mp
mf
mf
div.
mp
mp
mp
div.
13
86
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
mp
mf
mf
mp
mf
mf
mf
mf
mp
mf
mp
mf
mf
mf
mf
14
F
94
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mf
mf
mf espress.
mf
mf
ff
ff
f
f
f
f
f
f
pizz
15
102
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Tub. B.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
fp
f
f
ff
fp
f
f
ff
fp
f
f
ff
fp
f
f
f
f
f
f
f
mp
f
f
16
105
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Tub. B.
Vln I
Vln II
Vla
Vc.
Cb.
ff
fff
ff
fff
ff
fff
ff
fff
ff
ff
fff
ff
ff
ff
col legno div.
ff
ff
col legno div.
ff
17
Allegro (q.=88)
3.
110
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
mf
f
ff
mf
mp
mf
f
ff
f
mp
f
f
mp
f
ff
p
mp
p
ff
p
mp
p
3
ff
p
mp
p
ff
p
mp
p
mute
mp
f
mp
mute
mp
f
mp
f
f
f
ff
sul pont.
p
mp
p
f
ff
sul pont.
p
mp
p
f
ff
sul pont.
p
mp
p
f
ff
sul pont.
p
mp
p
ff
18
126
G
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
mf espress.
f
mf
f
mf
mp
f
mf
mp
f
mf
mp
f
mf
mp
f
f
f
f
fp
f
fp
f
fp
f
nat.
ff
mf
mp
nat.
ff
mf
mp
nat.
ff
mf
mp
nat.
ff
mf
mp
f
mp
f
19
H140
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mf
3 3 3 3
mf
3 3 3 3
mf
3 3 3 3
mf
3 3 3 3
mf
3 3 3 3
mf
3 3
3 3
mf
f
mf
f
mf
f
mf
f
(mute)
mp
(mute)
mp
mp
mf
mf cantabilé
mp
mf
mf cantabilé
mp
mf cresc.
mf cresc.
f cantabilé
3
mp
f cantabilé
3
mp
mp
f cantabilé
mf
pizz
mf cresc.
div. arco
mf
pizz
mf cresc.
arco
20
146
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mp dolce
mp dolce
ff
fp
ff
fp
ff
Bells up
fff
ff
Bells up
fff
ff
Bells up
fff
ff
Bells up
fff
mf
f
open
mf
f
open
ff
sfz
ff
sfz
ff
sfz
ff
sfz
mp
mp
ff
mp
ff
mp
ff
21
152
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
pp
pp
mf
f
pp
mf
f
pp
mf
f
pp
mf
f
pp
mute
f
pp
pp
pp
pp
22
I Vivace (q.=132)
158
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
f
ff
ff
ff
fp
ff
ff
fp
ff
ff
fp
ff
ff
fp
ff
mute
f
open
pp
open
ff
fp
ff
ff
ff
fp
ff
ff
ff
fp
ff
ff
ff
ff
pizz
f
pizz
f
2
pizz
f
2
pizz
f
2 ff
pizz
f
2 ff
23
J Tempo
166
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f cantabilé
3
f cantabilé
3
f cantabilé
3
f cantabilé
3
mp
f
mp
f
mp
f
mp
f
mf
mf cantabilé
mf
mf cantabilé
mf
mf
arco
mp
mp
3 3 3 3
arco
mp
mp
3 3 3 3
arco
mp
3 3 3 3
arco
mp
3 3 3 3
arco
mf
24
172
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mf
f
f
ff
mf
ff
p
ff
p
ff
p
f
fp
f
fp
ff
p
ff
p
ff
p
ff
p
f
mf
ff
f
mf
ff
ff
div.
ff
ff
25
178rit. K Poco meno mosso rit.
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
p
p
mf
mp
pp
p
fp
pp
mf
pp
p
fp
pp
mf
fp
p
mf
fp
p
ff
mp
mf fp
pp
mf
fp
p
ff
mp
mf
fp
pp
mf
fp
p
ff
mp
mf fp
pp
mf
fp
p
ff
mp
mf
fp
pp
mf
fp
p
p
p
p
p
ff
p
f
mp
fp
pp
p
ff
p
fp
fp
pp
p
ff
p
fp
fp
pp
p
ff
p
fp
fp
pp
p
f
mp
pp
p
f
pp
p
f
mp
fp
pp
div.
p
fp
fp
pp
p
fp
fp
pp
p
26
Andante (h=62)
4.
188
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
p
p
fp
p
fp
p
fp
p
fp
p
mp
f
mp
f
mp
f
mp
f
ff
mp
pp
f
p
pp
f
p
pp
f
p
pp
f
p
27
L195
M
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Glock.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
fp
fp
mf
mf
mf
fp
fp
fp
fp
fp
fp
fp
fp
mf
mf
fp
pp
mute
fp mute
pp
fp
pp
mute
fp
pp
mute
pizz
mf
f
28
204
N
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
f
fp
f
fp
f
fp
f
fp
mf espress.
f
mp
mute
mf
3 f
mute
mf
3 f
mp
f
mute
mf
3 f
mp
f
mute
mf
3 f
mp
f
mp
f
mf
p
nat.
nat.
nat.
pizz
mf
p
nat.
pizz
mf
p
mp
arco
f
29
212O
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Tub. B.
Glock.
Vln I
Vln II
Vla
Vc.
Cb.
fp
fp
fp
fp
ff
ff
ff
ff
pp
open
pp
open
pp
open
pp
open
ff
mp
ff
ff
mp
ff
p
p
pp
arco
pp
pp
arco
pp
arco
pp
arco
30
222
P
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Tub. B.
Glock.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
mp
f
fp
mf
3
mp
f
fp
mf
3
mp
mf
3
mp
f
fp
mf
3
mf
fp
mf
3
mp
mf
3
mp
mp
mp
mp
mf
fp
mf
mf
mp
mf
mp
mf
3
pp
pp
pp
arco
mf
fp
31
Q232
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
mp dolce
mp dolce
fp
fp
mf
mf
mf
mf
mf
mf
mf
mf
mp
fp
mp
fp
mp
fp
mp
fp
mp
fp
mp
fp
arco
mp
arco
mp
arco
mp
arco
mp
mp
32
241
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
f
f
f
f
mf
f
mf
f
mf
f
mf
f
mp
f
mp
f
mp
f
mp
f
mp
f
mp
f
ppp
f
7
ppp
f
7
ppp
f
7
ppp
f
7
ppp
33
R249
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Glock.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
f
f
ff
f
ff
f
ff
f
ff
f
fp
mp
fp
mp
fp
mp
fp
mp
mf
p
fp
mp
fp
mp
ff
fp
fp
ff
fp
ff
fp
pizz
mf
p
34
258
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
3
ff
3
ff
3
ff
3
mf
mp
ff
fp
3
ff
fp
3
ff
fp
3
ff
fp
3
ff
ff
ff
ff
ff
ff
f
f
pp
7
f
pp
7
f
pp
7
f
pp
7
pizz
f
pizz
mf
mp
35
S264
Fl. 1
Fl. 2
Ob.
Cl.
B. Cl.
Bsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Vib.
Vln I
Vln II
Vla
Vc.
Cb.
mf espress.
f
mp
dim.
ppp
5
mp
mp
mp
mp
mp
pizz
mp
pizz
mp
36
THEME & VARIATIONS for Eb Clarinet, Bb Clarinet, Bass Clarinet & Harp
Leif Sundstrup
2009
Submitted in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
2009
Instrumentation
1 Eb Clarinet
1 Bb Clarinet
1 Bass Clarinet
1 Harp
Note on Trills
Oboe
Half- tone trill
Oboe
Whole-tone trill
Transposing Score
Theme (q = 98)
Copyright © 2009 by Leif Sundstrup
Theme and Variationsfor Eb Clarinet, Bb Clarinet, Bass Clarinet & Harp
To Vanessa Ann Sundstrup
Leif Sundstrup
Clarinet in Eb
Clarinet in Bb
Bass Clarinetin Bb
Harp
mp
mf
mp
fLLLOLMLL
A§C§
G§D§
AbCb
GbDb
G§
5
Eb Cl.
Cl.
B. Cl.
Hp.
mp
mp
p
mf
p5 5 5
p
mf
5
p
Gb §
b
D§
Db
5
10 A
Eb Cl.
Cl.
B. Cl.
Hp.
p
mf
p
f
5
D§
G§
f
E§
6
GbDb
G§
15
Eb Cl.
Cl.
B. Cl.
Hp.
f
f
5
f
Eb
E§
Eb
mf
E§
Eb
E§
6
6
65
20B
Eb Cl.
Cl.
B. Cl.
Hp.
mf
pp
3
mf
pp
3
mf
pp
3
Eb
C§ f
Fb
GbCb
F§
26
Eb Cl.
Cl.
B. Cl.
Hp.
mf
E§
Eb
cresc.
3
2
31
Eb Cl.
Cl.
B. Cl.
Hp.
f
f
f
f
ff
E#
F#
F§
Eb
33C
Eb Cl.
Cl.
B. Cl.
Hp.
p
p
p
F#D#
F§Db
C§
mp
E§
Cb
Eb
3
38
Eb Cl.
Cl.
B. Cl.
Hp.
f
3
mf
f
E§
Eb3
3
41
rit.
Eb Cl.
Cl.
B. Cl.
Hp.
mf
pp
f
mf
pp
pp
mp
C§ D§
Variation 1 (q. = 92)45
Eb Cl.
Cl.
B. Cl.
Hp.
mf
fp
mp
mf
fp
mp
mf
fp
mf
f
mp
G§
60
Eb Cl.
Cl.
B. Cl.
Hp.
fp
fp
fp ff
ff
fp
fp
fp
ff
ff
fp
fp
fp
ff
ff
Gb
E§
G§ A§
nails
ff
4
72
Eb Cl.
Cl.
B. Cl.
Hp.
Ab Fb EbDb
D81
Eb Cl.
Cl.
B. Cl.
Hp.
mf
fp
2 2 2
fp
fp
fp
fp
fp
fp
fp
fp
f
fp
ff
Cb
91
Eb Cl.
Cl.
B. Cl.
Hp.
p fp
p fp
fp
p
fp
p
fp
fp
p
fp
p
fp
fp
mf F§
5
106E
Eb Cl.
Cl.
B. Cl.
Hp.
f
mf
f
mf
f
mf
Gb
G§D§ mf
C§
118
Eb Cl.
Cl.
B. Cl.
Hp.
fp
mp
fp
fp
fp ff
fp
mp
fp
fp
fp
ff
fp
mp
fp
fp
fp
ff
f
Gb
132
Eb Cl.
Cl.
B. Cl.
Hp.
f
f
f
Gb AbDb
G§D§
B§
Bb
3 3
6
142F
Eb Cl.
Cl.
B. Cl.
Hp.
ff
ff
f
ff
mp
f
mp
3
155
Eb Cl.
Cl.
B. Cl.
Hp.
mp
f
mp
mp
mp
GbDb
G§B§
169
Eb Cl.
Cl.
B. Cl.
Hp.
f
ff
flutter
mp
p
mp
f
ff
flutter
mp
p
f
ff
flutter
mp
p
D§
Db f
D§
AbDb
7
Variation 2 (q = 60)182
(q. = 88)
Eb Cl.
Cl.
B. Cl.
Hp.
mp
2
LMLOLMML
mp
mf
3
186
Eb Cl.
Cl.
B. Cl.
Hp.
mp
2
mp
5
Gb
3
190G
Eb Cl.
Cl.
B. Cl.
Hp.
mp espress.
mf
2 2 2
mf
mf
p
mf
G§ mp
Gb mf
3 3
8
195H
Eb Cl.
Cl.
B. Cl.
Hp.
mf
mp espress.
mf
2
2
2
mf
f
B§
mp
Bb
201I
Eb Cl.
Cl.
B. Cl.
Hp.
mp
pp
mp
pp
mp
pp
mf
3 3
205
Eb Cl.
Cl.
B. Cl.
Hp.
fp
fp
fp
Gb B§ Bb
ff
mf3
3 3 3 3 3 3
9
208
J
Eb Cl.
Cl.
B. Cl.
Hp.
ffp
f
5 5
ffp
f
3
ffp
f
ffCb
3
215
Eb Cl.
Cl.
B. Cl.
Hp.
fp
p
fp
p
3
3
fp
p
fp
p
3 3
fp
p
fp
p
3 3
f
E§
G§
Eb
E§
3 3
3 3
K226
Eb Cl.
Cl.
B. Cl.
Hp.
mp
2
mp
2
mfEb
C§
3 3
10
230
Eb Cl.
Cl.
B. Cl.
Hp.
mf
mp
5
Gb
3 3
234
Eb Cl.
Cl.
B. Cl.
Hp.
mp
ppp
mp
ppp
mp
ppp
G§
3 3
Finale (h = 48)238
Eb Cl.
Cl.
B. Cl.
Hp.
mp
MMLOLMML
6 6 6 6 6 6 6 6 6 6 6 6
11
240
Eb Cl.
Cl.
B. Cl.
Hp.
Gb Db
A§
6 6 6 6 6 6 6 6 6 6 6 6
L242
Eb Cl.
Cl.
B. Cl.
Hp.
pp espress.
3
3
pp espress.
3
3
pp espress.
3
3
G§D§
Gb
Db
6 6 6 6 6 6 6 6 6 6 6 6
244
Eb Cl.
Cl.
B. Cl.
Hp.
3
3
3
3
3
3
mf
G§D§
Gb
Db
6 6 6 6 6 6 6 6 6 6 6 6
12
246
Eb Cl.
Cl.
B. Cl.
Hp.
D§��
Db
D§
6 6 6 6 6 6 6 6 6 6 6 6
Ab
rit. 248
M Freely
Eb Cl.
Cl.
B. Cl.
Hp.
p espress.
p espress.
3
p espress.
3
LMLOLMML
p
252
N Tempo
Eb Cl.
Cl.
B. Cl.
Hp.
mp
MMLOLMML
mp
6 6 6 6 6 6
13
254
Eb Cl.
Cl.
B. Cl.
Hp.
Gb
6 6 6 6 6 6 6 6 6 6 6 6
256O
Eb Cl.
Cl.
B. Cl.
Hp.
pp espress.
3
3
pp espress.
3
3
pp espress.
3
3
Db
A§
G§
D§
6 6 6 6 6 6 6 6 6 6 6 6
258
Eb Cl.
Cl.
B. Cl.
Hp.
3
3
3
3
3
3
6 6 6 6 6 6 6 6 6 6 6 6
14
260
Eb Cl.
Cl.
B. Cl.
Hp.
Gb
Ab
6 6 6 6 6 6 6 6 6 6 6 6
Db
262P
Eb Cl.
Cl.
B. Cl.
Hp.
pp espress.
3
3
pp espress.
3
3
pp espress.
3
3 3
D§
G§
6 6 6 6 6 6 6 6 6 6 6 6
264
Eb Cl.
Cl.
B. Cl.
Hp.
3
3
3
3
3
3 3
Gb
Db
G§
Gb
6 6 6 6 6 6 6 6 6 6 6 6
15
266
Eb Cl.
Cl.
B. Cl.
Hp.
D§
Db
D§
6 6 6 6 6 6 6 6 6
268Q
Eb Cl.
Cl.
B. Cl.
Hp.
pp
pp
pp
Db
6 6 6 6 6
273
Eb Cl.
Cl.
B. Cl.
Hp.
ppp
ppp
ppp
6 6 6 6 6 6
16
275
Eb Cl.
Cl.
B. Cl.
Hp.
G§
Gb
6 6 6 6 6 6
276
Eb Cl.
Cl.
B. Cl.
Hp.
pp
f
mp
pp
f
mp
pp
f
mp
A§ D§f
cresc.
6 6 6 6 6 6
6ff
3
17
VOYAGE for Orchestra
Leif Sundstrup
2004
Submitted in partial fulfilment of the requirements
for the award of the degree
Doctor of Creative Arts
from
University of Wollongong
2009
Instrumentation
1 Piccolo Flute
2 Flutes
2 Oboes
1 Cor Anglais
2 Clarinets
1 Bass Clarinet
2 Bassoons
1 Contrabassoon
4 Horns
3 Trumpets
2 Trombones
1 Bass Trombone
1 Tuba
Harp
Violin 1
Violin 2
Viola
Cello
Contrabass
Percussion
Timpani
Bass Drum
Snare Drum
Toms
Tam-tam
Clash Cymbals
Suspended Cymbal
Sizzle Cymbal
Tambourine
Triangle
Bongos
Wind Machine
Ratchet
Whip
Glockenspiel
Xylophone
Vibraphone
Tubular Bells
Transposing score
Leif Sundstrup
Copyright © 2004 by Leif Sundstrup
Allegro assai ( q = 160 )
VOYAGE
Variations for Orchestra Piccolo
Flute 1
Flute 2
Oboe 1
Oboe 2
Cor Anglais
Clarinet 1in Bb
Clarinet in 2in Bb
Bass Clarinetin Bb
Bassoon 1
Bassoon 2
Contrabassoon
Horn 1in F
Horn 2in F
Horn 3in F
Horn 4in F
Trumpet 1in C
Trumpet 2in C
Trumpet 3in C
Trombone 1
Trombone 2
Bass Trombone
Tuba
Timpani
Percussion 1
Percussion 2
Percussion 3
Harp
Violin I
Violin II
Viola
Violoncello
Contrabass
mf
6 7
mf
5 6 7
mf
5 6
heroically
ff
mf
5 7
heroically
ff
mf
56
heroically
ff
mp
6 6
6
mp cresc.
5
6
mf
5 6 7
p cresc.
3
heroically
ff
p
mf
3
7
heroically
ff
p cresc.
mf
3
3
mf
3
3
heroically
ff
heroically
ff
heroically
ff
heroically
ff
ff
ffp
f
3
ff
ffp
f
3
3
ff
ffp
f
3 3 3
f
ff
sfp
3
f
ff
sfp
3
f
ff sfp
3
f
ff
hard mallets
f
fff
S.D.
(snares off)
f
snares on
mp
ff
B.D.
f
ff
Glock.
Vibes-medium motor
(hard mallets)
ff
EbbbbF####G§§§§AbbbbBbbbbCbbbbDbbbb
ff
f ff
pizz
f
ff
(vib.)
arco
mp
3
f ff
pizz
f
ff
(vib.)
3
f
ff
gliss. pizz
f
ff
(vib.)
mf
arco
3
f
ff
gliss. pizz
f
ff
(vib.)
3
f
ff
pizz
f
ff
(vib.)
f
arco
f
ff
pizz
f
ff
(vib.)
f
ff
pizz
f
ff
(vib.)
arco
ff
3
3
f
pizz
arco
ff
3
A h = 80
8
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
ff
ff
mf
espress.
f
ff
ff
ff
mp
mp
mp
mp
ff
ff
ff
ff
ff
ff
mp
sfz
Sus. Cym. (timp mallets)
p
p
soft mallets
E####F §§§§ A§§§§ sffz
ff
pizz.
mp
arco
ppp
arco
ppp
ff
pizz.
mp
arco
ppp
arco
ppp
ff
pizz.
mp
pp
sfz
ff
pizz.
mp
pp
sfz
arco (div.)
pp
2
20
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
5 6
5
5
7
ff
5 6 5 5 7
ff
5 6
5
5
7
ff
5 5 5 6
ff
5 5 5
5 6
mf
espress.
f
ff
5 5 5
7
ff
5 6 55
7
ff
5 5 5 5 6
ff
ff
ff
ff
ff prominantly
3
ff prominantly
3
ff prominantly
3
ff prominantly
3
f
mute
f
mute
f
mute
ff
f
ff
f
ff
ff
medium mallets
f
Clash. Cym.
f
S.D.
sfz
mp
arco (non div.)
sfz f
3
mp
arco (non div.)
sfz f
3
mp
arco (non div.)
sfz f
3
f
3
B27
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
cantabilé
f
3
f
marcato
f
marcato
f
f
f marcato
f
f marcato
f
f marcato
f
3
marcato
ff marcato
marcato
ff marcato
ff marcato
f
mf
f
ff
cuivré
f
mf
f
ff
cuivré
f
mf
f
ff
cuivré
f
mf
f
ff
cuivré
ff
open
3
ff
open
3
ff
open
3
f marcato
f marcato
f marcato
f
ff marcato
ff
3
ff
3
f
Glock.
F §§§§Bbbbb
EbbbbFbbbbGbbbbA§§§§B§§§§C§§§§Dbbbb
f
5 5 5 5 5 5 5 5
ff marcato
arco
f cantabilé
3
ff marcato
arco
f cantabilé
3
nondiv.
ff marcato
f marcato
3
nondiv.
ff marcato
div.
ff marcato
f marcato
3
ff marcato
pizz
f
arco
ff
4
39rit.
A tempo
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
3
ff
3
ff
3
ff
ff
ff
mp espress.
ff
3
ff
ff
3
ff
ff
ff
ff
ff
ff
ff
ff
ff
mf
ff
mf
mute
ff
open ff
mf
3
ff
ff
mf
ff
ff
mf
ff
mf
ff
ff
mf
ff
ff
ff
f
f
3 3 3 6
Sus. Cym.
mf
f
Whip
sffz
Xyl.
ff
3
EbbbbFbbbbG####A§§§§BbbbbCbbbbDbbbb
f cresc.
F####
B§§§§
fff
mf cresc.
gliss
.
gliss
.
fff
p
mp
pp
mf cresc.
gliss
.
gliss
.
fff
p
arcosul tasto
mp
pp
mf cresc.
gliss.
fff
sffz
p
arcosul tasto
mp
pp
ff cresc.
fff
sffz
arcosul tasto
p
mp
pp
ff cresc.
fff
sffz
p
arcosul tasto
mp
pp
ff cresc.
fff
sffz
arco
pp
mp
pp
5
C
49
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
3 3 3 3
f
5
f
3 3 3 3
f
5 3 3 3
3
f
3 3 3 3 3
f
3
f
3
f
3
f
3
f
più f
f
più f
f
più f
f
più f
ff
f
ff
f
ff
harmon mute (stem in)
f
open
33 3 3
ff
f
ff
f
ff
f
ff
Glock.
f
f
cresc.
nat.
f
nat.
f
3 5
nat.
f
f
pizz 3
f
f
pizz
3
6
57
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
7
ff
3 7
ff
7
ff
3 7
ff
7
ff
3 7
ff
f ff
5
5
3 7
ff
f ff
5
57
ff
3
ff
3
3
3
3
ff
3
3
3
3
ff
3
3
3
3
ff
3
9
f ff
7
9
ff
3
9
f ff
7
9
p
ff
5 5
p
ff
5
5
ff
5
5
ff
3 3
ff
ff
3
3
3
3
ff
9
ff
mp
ff
3 3 3 3 3 3 3 3 3 3 3 3 3
p
Wind Machine
f
mf
ff
f
B.D.
mp
ff
mf
ff
E§§§§F####G§§§§A####BbbbbC####Dbbbb
7
ff
3 7
ff
3
7
arco
ff
3
3
3
9
arco
ff
3
3
3
9
7
62 D
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
flutter
ff
pp
flutter
ff
pp
flutter
ff
pp
ff
p
ff
p
ff
p
ff
p
ff
p
ff
p
f
f
f
f
f
ff
p
f
mf
mf
mf
Xyl.
mf
f
mf
f
mp
Sus. Cym.
f
mf
Glock.
ff
Vibes - motor off
(hard mallets)
ff
E§§§§FbbbbGbbbbAbbbbB§§§§CbbbbDbbbb
ff
EbbbbA §§§§
ff
EbbbbF§§§§G§§§§AbbbbBbbbbC§§§§Dbbbb
mf
mf
pizz
f
f
sul pont.
ff
pp
pizz
mf
mf
sul pont.
ff
pp
pizz
mp
mp
pizz
p
f
p
f
p
pizzf
p
f
8
77
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
3 3 3 3
f
3 3 3 3
f
3 3 3 3
p ff
p ff
p ff
p ff
mp
9
7 7 7
p ff
p ff
p ff
p ff
mp
9
7 7
7
p ff
p ff
p ff
p ff
mp
9 7 7
7
p ff
p ff
p ff
p ff
mp
9 7 7 7
p ff
p ff
p ff
p ff
mp
9
77 7
p ff
p ff
p ff
p ff
mp
9
7 7 7
p ff
p ff
p ff
p ff
mp
9 7 7 7
p ff
p ff
p ff
p ff
mp
9 7 77
mp
f
metal mute
open
3 3 3 3
metal mute
f
open
3 3 3
3
f
metal mute
open
3 3 3 3
mf
f
Xyl.
mp
ff
mp
Glock.
Tubular Bells
f
ff
gliss.
gliss.
Cbbbb
gliss.
glis
s.
EbbbbF§§§§G§§§§A§§§§BbbbbC§§§§D§§§§
mf cresc.
mf
(pizz)
mf
ff
(pizz)
mf
mf
ff
(pizz)
mf
ff
(pizz)
mf
ff
mf
pizz
ff
9
E
85
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 2
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mf cantabilé
ff
ff
ff
ff
ff
ff
ff
ff
mf cantabilé
mf cantabilé
mf cantabilé
mf cantabilé
mf cantabilé
mute
mp
fp
mute
mp
fp
mute
mp
fp
mute
f
fp
mute
f
fp
mute
f
fp
Sus. Cym.
mp
ff
E§§§§ ff
arco
mf cantabilé
arco
mf cantabilé
arco
mf cantabilé
arco
mf cantabilé
10
F93
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
fff
f
fff
ff
f
fff
f
fff
ff boldly
ff
2 2
f
ff
2 2
ff boldly
ff
2 2
f
ff
2 2
f
open ff
f
open ff
f
open
ff
f
open ff
6 gliss. 7
ff
gliss.
ff
f
open ff
gliss. 6
ff
gliss.
ff
f
open fff
ff
f
fff
mf ff
mf ff
ff
S.D.
mf ff
mf ff
Toms
ff
EbbbbF§§§§GbbbbA§§§§BbbbbCbbbbDbbbb
ff
f
f fff
f fff
saltato
3 3 3 3 3 3
f
f fff
f fff
3 3
3 3 3 3
f
f fff
f fff
saltato
3 3 3 3
f
fff
col legno(battute)
sfz
arco
f
fff
col legno(battute)
sfz
11
102
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
flutter ff
pp
f
flutter
ff
pp
f
flutter ff
pp
f
f
f
f
f
f
f
ff
ff
2
ff
2
ff
2
ff
2
metal mute
f
open
ff
ff
mp
Sus. Cym.
f
ff
EbbbbF#G#AbbbbBCbbbbDbbbb
glis
s.
f
3 3 3
f
ff
arco sul pont.
pp
3 3
3
f
arco sul pont.
ff
pp
3 3 3
f
pizz
f
pizz
12
G
112
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 2
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
solo
f
gliss
.
3 3 55
3 3 3 3
p sust.
mp sust.
mp sust.
mp sust.
mp sust.
Bongos
mf
E§F§§§§G§A§§§§B§C§§§§D§
mp
mp
13
120
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mp sust.
mp sust.
mp sust.
mp sust.
mp sust.
mp sust.
mp sust.
mp sust.
mp sust.
f
mp
f
mp
f
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
mute
mf
ff
mute
mf
ff
solo
f
molto vib.
3 3 3
mp
f
mp
f
ff
Sus. Cym.
(hard sticks)
mp
f
Xyl
mp
EbbbbAbbbb Bbbbb ff
pizz
mf
ff
pizz
mf
ff
pizz
mf
ff
f
f
14
131
accel. h = 120
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
f
7
ff
f
7
ff
mf
5 7
ff
mf
3 7
ff
mf
7
ff
mf
5 7
ff
ff
3
3
fp
ff
f
7
fff
mf cresc.
5 7
7
fff
f
ff
3
3
fp
fff
f
ff
3
3
fp
fff
f
ff
3
3
7 7
ff
ff
fff
bells up
mf
fff
f
ff
ff
fff
bells up
mf
fff
f
ff
ff
fff
bells up
mf
fff
f
ff
ff
fff
bells up
mf
fff
f
(mute)
ff
open 4 4 4 4
(mute)
ff
open
4 4 4 4
ff
6 gliss. 7
ff
gliss.
ff
gliss.6
ff
gliss.
fff
ff
3
3
fp
2 2
2 2
fff
f
ff
3
3
fp
2 2
2 2
mf ff
mf ff
ff
4 4 4 4
mf ff
mf ff
Rachet
ff
Lg. Sus. Cym.
(timp mallets)
p
Xyl.
ff
4 4 4 4
A§§§§
ff
E§§§§F§§§§G§§§§A####B####CbbbbDbbbb
ff
arco
f fff
f fff
sul pont.
f cresc.
nat
ppp
3 3
arco
f fff
f fff
sul pont.
f cresc.
mf
nat.
3
3
7
7
mf cresc.
nat.
3
3
3
3
3
arco
f fff
f fff
sul pont.
f cresc.
nat.
mf cresc.
5 7
7
mf cresc.
nat.
3
3
3
3
3
3
arco
fff
sul pont.
f cresc.
nat.
ff
3
3
arco
fff
col legnobattute
f cresc.
fff
nat.
ff
3
3
3
3
15
H Misterioso ( h = 60 )
145
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
pp
pp
ff
pp
pp
ff
pp
pp
ff
pp
mf
pp
mf
ff
pp
mf
pp
mf
ff
pp
mf
pp
mf
ff
pp
pp
ff
pp
pp
ff
pp
pp
mp
ff
pp
mf
pp
mf
ff
pp
mf
pp
mf
ff
mp
ff
mf
ff
mf
ff
ff
ff
ff
p
6mf
1 molto vib (slide)
ff
p
7mf
molto vib (slide)
ff
p
mf
molto vib (slide)
ff
p
Roll on inverted
cym. upon timp skin
f
gliss. f
gliss.
f
B.D.
p
mf
p
Wind Chimes
mf
Bowed Cym.
sfz
sfz
sfz
Tubular Bells
mp
mp
Vibes
ff
E§§§§FbbbbG§§§§AbbbbBbbbbCbbbbD§§§§
Play near fingerboard
with triangle beater
ff p
ff p
6 6
mf
Ab
§
bgliss.
mf
Ab
§
bgliss.
mf
Ab
§
bgliss.
ppp
f
p
ff
ppp
f
p
sul tasto
ppp
f
p
sul tasto
ff
ppp
f
p
sul pont.
mf
p
mf
p
mf
ff
ppp
f
p
sul pont.
mf
p
mf
p
mf
pizz
ff
pizz
pp
pizz
ff
arco
mp
mp
mp
12 12 12
16
160
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mf dolce
3
mf
mf
sfz
sfz
sfz
sfz
Ab
§
bgliss.
Ab
§
bgliss.
A§
#
§gliss.
Ab
§
bgliss.
mute on
mp dolce
3
p
mf
p
mf
p
mf
p
mf
p
mf
p
mf
mp
mp
mp
mp
12 12 12 12
17
167
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mf dolce
p
p mp
mf
p
p mp
mf
p
mp
mf
dim
3
mp
p
straight mute
mp
p
open
straight mute
mp
p
open
straight mute
mp
p
open
p
sfz
mp
Tamb.
thumb
Vibes
mf
A§
#
§gliss.
mf
3
mute off
mf
mp
3
p
mf
two players (nat.)
p mp dolce
p
mf
two players (nat.)
p mp dolce
mp
(pizz) one player
arco tutti
mp
3
mp
mp
pizz one player
12
18
I177
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mf
mf
mf
mf
mf
mf
p
mf cresc.
mf cresc.
mf
mf cresc.
mp
5
mf cresc.
mp
mf cresc.
mp
3
mf cresc.
cup mute
mp
5
mp
5
open
mf cresc.
cup mute
mp
mp
open
mf cresc.
cup mute
mp
3
mp
3
open
f
cup mute
mp
5
open
cup mute
mp
open
cup mute
mp
3
open
pp
cresc.
p cresc.
Sus. Cym.
p
Vibes
motor slow
E§§§§F§§§§G§§§§A§§§§B####C§§§§D§§§§
pp
cresc.
tutti
pp
cresc.
tutti
pp
cresc.
div
cresc.
arco tutti
pp
cresc.
19
190
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
5
5
flutter
mf cresc.
ff
flutter
mf cresc.
ff
3
3
mf cresc.
f cresc.
mf cresc.
mf cresc.
f
mf cresc.
f
mf cresc.
f
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
f
ff
ff
f
ff
f
3
ff
f
ff
f
3
ff
f
ff
f
3
mf cresc.
mf cresc.
f
f
f
f
f
Tubular Bells
3
3 3
3
f
f
arco
f
arco
mf cresc.
mf cresc.
mf cresc.
mf cresc.
f
f
f
20
194
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mf
mp
flutter
ff
ff
mf
mp
mf
f
mp
mf
3
mp
ff
f
ff
mp
3
f
ff
mp
3
f
ff
mp
3
f
ff
mp
3
ff
f
ff
f
p
ff
f
ff
straight mute
mf
ff
f
ff
mp
p
mf
mp
p
mf
mp
p
mp
mp
p
mp
Tamb.
thumb
3 3
mf
C####
C§§§§
3
pp D string
Freely Ad lib. to K
pp D string
Freely
Ad lib. to K
pp G string
Freely Ad lib. to K
pp G string
Freely
Ad lib. to K
mp
sul tasto (two desks)
mp
sul tasto (two desks)
(two desks)
mp
mp
one
21
J202
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vla
Vc.
Cb.
mf espress.
f
mf espress.
f
mf espress.
p cresc.
p cresc.
p cresc.
B§§§§
mp
tutti
p
tutti
p
tutti
p
tutti
p
22
210 K
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vla
Vc.
Cb.
p
mp
mf espress.
f
p
mp
p
mp
p
mp
p
mp
p
mp
mf espress.
mf espress.
f
p
p
mp
sfz
mp
3 3
3 3
3 3
sfz
mp
3
3
3
3
3
3
sfz
mp
3
3
3
3
3
3
sfz
mf
open
p
p
ff
p
p
p
p
p
mp
pizz
23
219
L
Tempo Giusto (q = q)
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
flutter
f
flutter
f
flutter
f
f
p
f
p
f
p
f
p
straight mute
f
p
straight mute
f
p
straight mute
f
p
p
mp
straight mute
f
p
p
mp
straight mute
f
p
p
mp
straight mute
f
p
p
mp
straight mute
f
p
p
mp
Tamb.
mp
Tri.
p
mp
Vibes
(medium mallets)
mp
D####
EbbbbF####GbbbbAbbbbBbbbbCbbbbD§
Falling Hail
mf
f
mf prés de la table
p
sul pont.
f
nat.
f martelé
p
pizz
p mf
sul pont.
f
sfz
nat.
f martelé
p
pizz
p mf
sfz
sfz
p
pizz nat.
f martelé
p
pizz
p
mf
sfz
p
pizz
sfz
three players arco
pp
p
ff
mf
mp
three players arco
pp
p
24
234
M Agitato
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
p
mf
p
mf
f
ff
5
p
mf
p
mf
f
ff
5
p
mf
p
mf
f
ff
5
p
mf
p
mf
f
ff
3
3
3
p
mf
p
mf
f
ff
3
5
p
mf
p
mf
f
ff
p
mf
p
mf
f
ff
3
3 3
p
mf
p
mf
f
ff
p
mf
p
mf
f
ff
ff
3
5
p
mf
p
mf
f
ff
ff
5
p
mf
p
mf
f
ff
p
mf
p
mf
f
ff
ff
mp
mf
f cresc.
sfz
ff
5
mp
mf
f cresc.
sfz
ff
5
mp
mf
f cresc.
sfz
ff
5
mp
mf
f cresc.
sfz
ff
5
open
mp
mf
f
ff
3
open
mp
mf
f
ff
3
open
p
mp
mf
f
ff
3
open
mp
mf
f
metal mute
ff
open
ff
5
open
mp
mf
f
metal mute
ff
open
ff
4
open
mp
mf
sfz
metal mute
ff
open
ff
5
open
mf
sfz
ff
ff
p
pp
sfz
p
Sus. Cym.
f
pp
ff
B.D.
pp
sfz
C§§§§
G§§§§
f nat.
p mf
arco
p
sfz
ff
3
5 5 5 5 5 5 5 5 5
3
3
p mf
arco
p
sfz
ff
p
mf
arco
p
3
3
3
3
3
3
3
3
3
sfz
ff
3 5
pizz (tutti)
mp
arco
p
mf
sfz
ff
ff
4
pizz (tutti)
mp
arco
p
mf
sfz
ff
ff
25
245
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
p
mp dolce
3
mp
p dolce
3 3
p
mp dolce
3
mp
3 3
p
mp dolce
3
mp
3
3
p
mp dolce
3
mp
3
3
p
mp dolce
3
mp
3 3
p
mp dolce
3
mp
3 3
p
mp dolce
3
mp
3 3
p
mp dolce
3
mp
3 3
3
fff
mp
3
3
fff
mf espress.
3
3
fff
3
3
fff
3
p
pp
p
pp
p
pp
p
pp
fff
fff
fff
fff
3
fff
3
fff
3
3
fff
3
A §§§§
mp
f
GbbbbA Cbbbb
mp mf
E§§§§B C §§§§
mp
3
gliss
.
3
mp dolce
3
pp
p
gliss.
one (arco)
p dolce
3
p
pp
p
gliss.
one (arco)
p dolce
p
pp
p
gliss.
mp dolce
3
pp
p
gliss.
two desksdiv.
p
3
3
fff
p
pp
mp
pizz
3
3
fff
p
pp
mp
pizz
3
26
253
molto rit. Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
ff
ff
ff
ff
p
ff
p dolce
ff
ff
p dolce
ff
p
p dolce
ff
ff
ff
mp
ff
p
ff
ff
ff
ff
ff
ff
ff
ff
ff
p dolce
ff
p dolce
ff
p dolce
ff
harmon mute (stem in)
ff
ff
ff
open
ff
7 gliss.
hormon mute (stem in)
ff
ff
ff
open
ff
7 gliss.
harmon mute (stem in)
ff
ff
ff
open
ff
6
gliss.
ff
ff
ff
ff
Sizzle Cym.
(drum sticks)
ff
f
Vibes - motor off
F §§§§
sffz
Thunder
sffz
sffz
mp
p dolce
ff
tutti (arco)
6
mp
p dolce
ff
tutti (arco)
6
tutti (arco)
ff
6
col legno battute
ff
ff
ff
arco
ff
col legno battute
ff
ff
ff
arco
ff
27
N Scherzando (h. = 60)
260
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
one
p cresc.
mp cresc.
mf cresc.
mf cresc.
one
mp cresc.
one
mf cresc.
28
280 O
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
p
p
p
p
f
sffz
sffz
sffz
sffz
f
sffz
sffz
sffz
f
sffz
sffz
sffz
f
sffz
sffz
Tamb.
p
mp
Tri.
mf
p
mp
f
Tam-tam (scrape with coin)
f cresc.
tutti
ff tenuto
fff
f cresc.
tutti (div.)
ff tenuto
fff
f cresc.
one
tutti (div.)
ff tenuto
fff
f cresc.
tutti (div.)
ff tenuto
fff
f cresc.
one
tutti
ff tenuto
fff
29
300
P e = e (h = 90)
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
cresc.
fff
sffz
p
cresc.
fff
sffz
mf cresc.
fff
sffz
cresc.
fff
sffz
mf cresc.
fff
sffz
p
cresc.
fff
sffz
cresc.
fff
sffz
p
cresc.
fff
sffz
cresc.
fff
sffz
p
cresc.
fff
sffz
mf cresc.
fff
sffz
p
cresc.
fff
sffz
mp
mf
fff
sffz
fff
sffz
mp
mf
fff
sffz
fff
sffz
mp
mf
pp
pp
pp
mp
mf
pp
pp
pp
pp
mp espress.
mf
ff
gliss.
mp espress.
mf
div. ff
gliss.gliss.
mp espress.
mf
div.
ff
gliss.gliss.
mp espress.
mf
div.
ff
gliss.gliss.
mp espress.
mf
ff
30
318
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
ff
stomp on floor
mf
f
gliss.
mf
sfz
bells up
mf
f
mf
sfz
bells up
mf
f
gliss.
mf
sfz
bells up
mf
f
gliss.
mf
sfz
bells up
flutter
ff
mp
f
flutter
ff
mp
f
flutter
ff
mp
f
flutter
ff
mp
f
ff
flutter mp
f
flutter
ff
mp
f
flutter
ff
mp leggiero
pizz
sfz
arco
mp leggiero
pizz
sfz
arco
mp leggiero
pizz
pizz
sfz
arco
mp leggiero
pizz
pizz
sfz
arco
mp leggiero
pizz
pizz
sfz
31
332
Q Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
mp
ff
mp
ff
mp
ff
mp
mp
mp
mp
mp
mp
mp
sfz
sfz
sfz
sfz
sfz
sfz
sfz
sfz
mp
mp
mp
mp
mp
mp
ff
drum stick
mp
Tri.
mp
mp
ff
ff
ff
ff
ff
32
344
R
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
mp cantabilé
pp
2 4
mp cantabilé
pp
2 4 4
mp cantabilé
pp
2 4
mp cantabilé
pp
2 4 4
mp cantabilé
pp
2 4 4
mp cantabilé
pp
2 4
mp cantabilé
pp
2 4 4
mp cantabilé
pp
2 4
mp cantabilé
pp
2 4
mp cantabilé
pp
2 4 4
mp cantabilé
pp
2
4
mp cantabilé
pp
2
4
sfz
p
sfz
p
mf
pp
sfz
p
sfz
p
mf
pp
sfz
p
sfz
p
mf
pp
sfz
p
sfz
p
mf
pp
mf
4
hard mallets
f sffz
S.D. (snares off)
f sffz
E§§§§F§G§A§§§§B§§§§C§§§§D§§§§
ff
mf molto cresc.
p spicc
f
p
f
p
stacc (on string)
molto cresc.
p spicc
f
p
f
p
stacc (on string)
molto cresc.
p spicc
f
p
f
p
stacc (on string)
molto cresc.
p spicc
f
p
f
p
stacc (on string)
molto cresc.
arco
p spicc
f
p
f
p
stacc (on string)
molto cresc.
33
357
S e = e (in 3)
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
p cresc.
f
mp cresc.
f
mp cresc.
f
mp cresc.
f
mp cresc.
f
mp cresc.
f
mp cresc.
mf molto cresc.
f
f espress.
p
mf molto cresc.
f
f espress.
p
mf molto cresc.
f
mp cresc.
Glock.
mf
ff
EbbbbAbbbb
mp cresc.
f espress
mp cresc.
5
f espress
mp cresc.
5
f espress
mp cresc.
5
f
mp cresc.
f
mp cresc.
34
368
rit. T Meno mosso (h = 74) Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
f
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
mp cresc.
ff
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
mp cresc.
ff
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6
ff declaratory
3
ff declaratory
3
ff declaratory
3
ff
3
ff
3
ff
3
ff
3
ff
3 3 3
ff
3 3 3
ff
3
3 3
ff declaratory
3
ff declaratory
3
ff declaratory
3
ff declaratory
3
ff
Tri.
ff
Clash Cym.
ff
ff
gliss.
ff
ff
ff
ff
ff
35
Sostenuto375
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Perc. 1
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
ff
ff
ff
ff
ff
ff
ff
ff
fff molto espress.
ff
ff
fff molto espress.
ff
ff
fff molto espress.
ff
ff
ff
ff
ff
ff
ff
ff
ff
non div.
ff
fff molto espress.
non div.
ff
fff molto espress.
non div.
div.
ff
fff molto espress.
non div.
pizz
ff
arco
fff
pizz
ff
arco
fff
36
384
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
3
3
3
3
3 3
3
3 3
3 3
3
3
3
3
3
3 3
3 3
3 3
medium mallets
f
3 3 3
7
div.
7
div.
7
7
37
Risoluto (q = q)390
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
fff tenuto
7
fff tenuto
7
fff tenuto
fff tenuto
fff tenuto
fff tenuto
fff tenuto
fff tenuto
fff tenuto
fff tenuto
fff
sffz
sffz
fff
sffz
sffz
fff like bells
fff like bells
fff like bells
fff like bells
fff like bells
sffz
sffz
fff like bells
sffz
sffz
fff like bells
sffz
sffz
fff like bells
6
fff like bells
7
fff like bells
fff
sffz
sffz
ff
ff
Clash Cym.
ff
Tam-tam
ff
Tubular Bells
ff
E§§§§A§§§§
ff
fff tenuto
7
fff tenuto
5
fff tenuto
7
fff
fff tenuto
7
fff
sffz
sffz
38
402
Picc.
Fl. 1
Fl. 2
Ob. 1
Ob. 2
C. A.
Cl. 1
Cl. 2
B. Cl.
Bsn. 1
Bsn. 2
Cbsn.
Hn. 1
Hn. 2
Hn. 3
Hn. 4
Tpt. 1
Tpt. 2
Tpt. 3
Tbn. 1
Tbn. 2
B. Tbn.
Tba.
Timp.
Perc. 1
Perc. 2
Perc. 3
Hp.
Vln I
Vln II
Vla
Vc.
Cb.
sffz
sffz
sffz
cresc.
sffz
5 6
sffz
sffz
sffz
cresc.
sffz
5 6
sffz
sffz
sffz
cresc.
sffz
56
sffz
sffz
sffz
cresc.
sffz
5 6
sffz
sffz
sffz
cresc.
sffz
5
6
sffz
sffz
sffz
cresc.
sffz
5 6
sffz
sffz
sffz
cresc.
sffz
56
sffz
sffz
sffz
cresc.
sffz
56
sffz
sffz
sffz
cresc.
sffz
56
sffz
sffz
sffz
cresc.
sffz
5
6
sffz
sffz
sffz
sffp
fff
pp
fff
sffz
sffz
sffz
sffp
fff
pp
fff
sffp
bells up
fff
pp
fff
sffp
fff
bells up
pp
fff
sffp
fff
bells up pp
fff
sffp
fff
bells up
pp
fff
pp
fff
pp
fff
pp
fff
pp
fff
pp
fff
pp
fff
sffz
sffz
sffz
pp
fff
fff
sffz
S.D.
p molto cresc.
sffz
B.D.
p molto cresc.
sffz
ff
4 4 4 4 4 4
fff
non div.
sffz
3 3 3 11
non div.
sffz
3 3 3 11
non div.
sffz
3 3 3 10
non div.
sffz
3 3 3 9
sffz
sffz
sffz
non div.
sffz
3
5
39
Recommended