Multichannel sound recording · ceived, for the four channel system, purely as an extension and improvement of the original Quadraphonic Recording/Reproduction System, and it seems,

1

Multichannel Microphone Array Design (MMAD)

Multichannel sound recordingMultichannel Microphone Array Design (MMAD)

By Michael Williams1

and Guillaume Le Dû 2

1 - Independent Audio Consultant, Paris, France2 – Radio France, Paris, France

ing perfect reproduction of the sound environ-ment. We must also be able to use this type ofsound recording system in the recording of con-temporary music which will often make use ofthe surrounding sound space, and of coursesound effects and ambience where soundshould be localised correctly within the total sur-rounding sound field.

But don’t be deceived, the achievement of thisaim needs very careful adjustment of the pa-rameters of the microphone array, and a clearunderstanding of the fundamental characteris-tics of each segment of the array. It must alsonever be forgotten that, as with stereo, the mi-crophone array will pick up not only sound in thehorizontal plane around the array, but bothabove and below the system (12). This soundwill obviously be reproduced only in the horizon-tal plane of the reproduction system. We musttherefore not loose sight of the fact that evenmultichannel surround sound recording and re-production systems still have some limitationsto the impression given of natural reproductionof the total sound field.

Segmentation of the Sound FieldAt the 91st AES Convention (New York - 1991)in a paper entitled "Microphone Arrays for Natu-ral Multiphony" (5), the author (MichaelWilliams) described the development of threemicrophone arrays suitable for recording/re-production systems using four, five or six chan-nels. At this time the author considered the as-sociated loudspeaker layout to be perfectly sym-metrical, the reproduced surround sound fieldbeing generated by loudspeakers placed on acircle divided into equal segments in the horizon-tal plane. Also the recording/reproduction sys-tem was based on a univalent or one to one mi-crophone/loudspeaker relationship. The charac-teristics of a specific microphone array coveringa given number of segments, was determined

IntroductionMultichannel sound recording systems are as muchsubject to the basic rules of sound recording as anyother recording system, be it monophonic or stereo-phonic. The microphone or microphone array willneed to be nearer the sound source to obtain thesame sound perspective compared with perceptionby normal hearing of the actual sound source. Theperception of distance is as usual a function of directto reverberant sound and therefore directly influ-enced by the directivity of the microphones used inthe system. The advantage of the multichannel sur-round sound environment is that we are able to« envelop » the listener as much as we consider de-sirable. In achieving this aim we are not limited to thefront facing triplet of microphones and loudspeakersto create our main sound stage, for this would createlimits to the main sound stage and be little betterthan the traditional stereophonic reproduction. How-ever this front perception zone of about 60° seemsto satisfy approximately our needs as to the angularsize of the main sound stage, but with multichannelcontinuous sound field recording and reproduction,we have the freedom to be able to widen this soundstage if we feel the need.

The stereophonic listening configuration severely lim-its the sound recording engineer in the reproductionof early reflections and of course in the reproductionof the surrounding reverberant sound environment(4). The process of MMAD shows how it is possibleto design multichannel arrays that are capable of giv-ing complete freedom for the sound engineer tospread the main sound stage over any desired angle,and to integrate both first reflections and reverbera-tion and whilst maintaining their natural acousticstructure. This is another major step towards achiev-

2

from the intersection of the physics of the micro-phone array with the psychoacoustics of the lis-tening configuration. This approach was con-ceived, for the four channel system, purely as anextension and improvement of the originalQuadraphonic Recording/Reproduction System,and it seems, was somewhat in advance of themultichannel systems now required.

Although a similar approach by segmentation ofthe sound field is used in Multichannel Micro-phone Array Design (MMAD) (8), almost com-plete freedom can be obtained in the relation tothe physical angle between the microphones withrespect to the corresponding segment coverageangle. A detailed analysis will show how to “link”segments without overlap, and obtain a smoothcontinuous reproduction of the total sound field.

The standardisation in the context of multichan-nel reproduction in relation to surround sound foraudio-visual applications, has since modified con-siderably the layout of the sound field segments.This, together with an ever increasing interest inthe use of multichannel systems for the repro-duction of music as a purely audio media, hasshown the need for a clear and unambiguousanalysis of the psychoacoustics of reproduction ina multichannel environment, in order to help inthe development of microphone arrays that canmeet the need for specific sound field segmentpick-up and reproduction.

In Stereophony, considerable use has been madeof Intensity Difference, Time Difference and com-bined Intensity/Time Difference systems to pro-duce reasonably satisfactory microphone sys-tems for the recording and reproduction of thesound field. In previous AES papers (1) (2) (3) (4)(5) (6) the author has shown how a multiplicity ofdual microphone arrays can be developed to en-able the sound recording engineer to choose thebest configuration for optimum stereophonic re-sults in a given set of conditions. This same ap-proach has been applied to microphone arraysfor multichannel systems.

It is a convention established in a previous AESConvention papers (1) (2) that the StereophonicRecording Angle is always specified for instanceas +/- 50° to avoid confusion with the angle be-

tween the microphones. However in the case ofMultichannel Microphone Arrays we no longerhave a symmetrical recording/reproductionsound stage in relation to each of the array seg-ments, so it would seem necessary to changeterminology and use the idea of Coverage Angleto mean the total coverage of the StereophonicRecording Angle of a single pair of microphonesmaking up part of the total Multichannel Array. Inthe above example the SRA of +/- 50° becomesa Coverage Angle of 100°.

Back to BasicsIt should not come as any surprise to the initiatedthat we must first deepen our knowledge of thecharacteristics of a simple pair of microphones,reproduced evidently by two loudspeakers. In thecontext of stereophonic reproduction, left/rightsymmetry of the reproduced sound field is, ingeneral, obviously desirable. The sound field re-produced by the loudspeakers can be eithersmaller or larger than the physical angle be-tween the axis of directivity of the microphones,and disposed usually symmetrically on each sideof the physical axis of the pair of microphones.

However in the design of multichannel micro-phone arrays we must have complete controlover the angular offset of the reproduced seg-ments of the sound field in relation to the axis ofsymmetry of the microphone pair covering eachspecific segment. In the design of the front facingtriplet of microphones covering the left and rightfront segments, we must be able to offset thereproduced sound field segments so that theside limits of the left and right sound fields corre-spond to the axis of the front facing centre micro-phone. For the lateral segments however weneed to be able to rotate freely the coverage ofthe side segments, without any coincidence be-tween the physical axis of the microphones andthe limits of each of the sound field segments.And finally, for the back segment we return to thesimple symmetrical segment as used instereophony.

Offset and LinkingThe application of Offset is illustrated in Figure 1to 6.• Figure 1 shows the standard Stereophonic


3

Recording Angle (SRA) where it is less than thephysical angle between the microphones.• Figures 2 & 3 show how this same sound field

coverage can be rotated in an anti-clockwise di-rection, that is using an negative angular offset.

• Figure 4 shows an SRA that is larger than theangle between the microphones.

• Figures 5 & 6 show the same principle of offset,but using, this time, a clockwise rotation or posi-tive angular offset.

In these Figures (1 to 6) the Stereophonic Record-ing Angle has been drawn with its true origin on aline between the two microphone capsules. In

practice this is an approximation that is justifiedbecause the sound source is usually much furtheraway from the microphone system than the dis-tance between the microphones. If the distance ofthe sound source is relatively near the micro-phone source then the analysis becomes muchmore complicated as the microphone system isno longer within the zone of plane wave sound fieldpropagation.

In Figure 2 and in Figure 5 we see that the left lim-its to the SRA are aligned (or parallel) with the axisof the left microphone.

Figure 1Stereophonic Recording Angle (SRA) lessthan angle between microphones

Figure 2Negative Offset of 15°: left limit ofSRA aligned with microphone axis

Figure 3Negative Offset of 35°: axis of leftmicrophone now within the SRA

Figure 4SRA is greater than anglebetween microphones


4

of linking that is an essential characteristic in thedesign of a multichannel microphone array forsmooth and continuous coverage.

Figure 7 shows the front triplet of a MultichannelMicrophone Array. We can see that we have mi-crophones facing both towards the left-hand sideand right-hand side, forming left facing and rightfacing pairs by sharing the centre microphone.

Figure 5Positive Offset of 20°: left limit ofSRA aligned with microphone axis

Figure 6Orientation of SRA withPositive Angular Offset of45°

The process of linking is represented in Figures 7to 10. As the sound source is relatively far awaycompared with the distance between the micro-phones, it is an acceptable approximation fromnow on, to draw the graphical origin of the SRAcoverage at the intersection between the micro-phone axes. This enables us in Figure 8 & 9, andthereafter, to illustrate more clearly the process

Figure 7The Front Facing Triplet


5

In Figure 8 & 9 we show how to link the SRAs ofeach pair to produce continuous coverage of thefront sound stage. Figure 8 uses the link betweentwo SRAs that are each smaller than the physicalangle between the microphones that make uptheir respective pairs. We need to use Positive An-gular Offset on the left segment and Negative An-gular Offset on the right segment to critically linkthe two segments. However in Figure 9 the link is

between the two SRAs that are larger than theangle between the microphones, we thereforeuse Negative Angular Offset on the left segmentand Positive Angular Offset on the right segmentfor correct linking. By this process we can cre-ate a front triplet with any desired coverage an-gle, thereby giving more flexibility to the soundrecording engineer in setting up the coverage ofthe main front sound stage.

Figure 8Critical Linking of Front Left Segment(FLS) and Front Right Segment (FRS)(left & right Coverage Angles < anglebetween microphones)

Figure 9Critical Linking of Front Left Segment(FLS) and Front Right Segment (FRS)(left & right Coverage Angles > anglebetween microphones)


6

Intensity and Time OffsetGenerationThe method used to produce this offset techniqueis in fact remarkably simple, both in theory andpractice. The difficulty comes in the choice of spe-cific offsets so that « critical linking » is obtained,and smooth and continuous sound field reproduc-tion is achieved. Basically there are four differenttypes of offset that we can apply :

• Electronic Intensity Offset – addition of a con-stant Intensity Difference to the Intensity/TimeDifference function between two microphones

• Electronic Time Offset – addition of a con-stant Time Difference to the Intensity/TimeDifference function between two microphones

• Microphone Position Intensity Offset• Microphone Position Time Offset

These last two offsets are one and the same,as it is just a question of one type of offset rela-tive to the other. By creating a Microphone Posi-tion Intensity Offset in one direction, it is equiva-lent to creating a Microphone Position Time Off-set in the other direction.


Figure 10Critical linking between all segments of aMultichannel Microphone Array

Figure 10 shows how this process of linking is ap-plied for the complete Multichannel MicrophoneArray. Only the front facing centre microphone isaligned with the linking of two segments i.e. the

link between the Front Left Segment (FLS) and theFront Right Segment (FRS). The Back Segment(BS) is usually symmetrical with respect to theback axis of the system.

7

Figure 11 illustrates the analysis of this Micro-phone Position Offset, but seen from two differentpoints of view :

• Using the microphone diaphragm as the centreof rotation for each microphone, we can considerthat both microphones have been rotated 15° inan anticlockwise direction (Figure 11b), therebyalso rotating the Intensity Axis (DI=0) anticlock-wise by the same amount i.e. Negative Micro-phone Position Intensity Offset.

• Or we can think of the right hand microphone as

having been moved backwards whilst maintain-ing the same distance between the micro-phones (Figure 11c) - the time axis (DT=0) hastherefore been rotated 15° in a clockwise direc-tion i.e. Positive Microphone Position Time Off-set.

It is agreed that the trigonometric analysis ofthese two interpretations is somewhat compli-cated, but the practical end result is exactly thesame i.e. the orientation of the Intensity axis rela-tive to the Time axis is the same.


Fig 11a No offset

Fig 11b Negative Intensity Offset

Fig 11c Positive Time Offset

8

There is however a subtle difference between Elec-tronic Offset and Microphone Position Offset.• Electronic Offset is simply the addition of a con-

stant value of Intensity Difference or Time Differ-ence to the Intensity/Time function of a pair ofmicrophones covering a particular segment.

• Microphone Position Offset is created by chang-ing the physical position of the microphones form-ing the pair, thereby creating an angular differ-ence between the Intensity and Time axes.

Let us be clear on the convention that has beenadopted for Positive and Negative Offsets :

Positive Offset is defined as that offset whichproduces a rotation of the Coverage Angle in aclockwise direction.

• This is obtained in the case of a Positive Elec-tronic Time Offset by introducing a delay in theright hand microphone in relation to the left handmicrophone.

• Similarly with Positive Electronic Intensity Offsetthe right hand microphone is attenuated in rela-tion to the left.

• With Positive Microphone Position Time Offsetthe right hand microphone is rotated clockwise(i.e. backwards) using the centre of the left micro-phone diaphragm as the centre of rotation whilstmaintaining the same distance between the mi-crophones and angle between the axes of the mi-crophones.

Each of these different positive offsets will producea shift in their respective Intensity/Time functiongraphs either to the left for Positive Time Offsets,or downwards for Positive Intensity Offsets. Thiscan be a little disconcerting at first, however thesound position co-ordinates show the relative posi-tion of the Coverage Angle and this leaves no roomfor ambiguity

Further detailed analysis of different offset func-tions can be found in AES preprint 4997 (8).

The process of Multichannel Microphone Array De-sign is highly complex, due to the interaction of themany parameters that are involved. It would seemalmost impossible to envisage undertaking thecomplete multichannel design process just before

the beginning of a recording session. Unfortu-nately this is the moment when the soundrecording engineer is more or less in posses-sion of the complete range of characteristics ofthe sound field to be recorded. This rest of thisdocument will present a quick reference guideto some different Multichannel Microphone Ar-ray configurations to assist the sound engineerin quickly choosing a system adapted to eachparticular circumstance. These Multichannel Mi-crophone Array configurations were first pre-sented in AES preprint (11).

All systems described present good Critical Link-ing; i.e. the reproduction of the sound field iscontinuous without any « holes » or « overlaps ».However as a complete description of each sys-tem would require as many pages as there aresystems, it was considered more reasonable topresent the characteristics of each of the ar-rays grouped within a table covering each spe-cific value of Front Triplet Coverage. Withineach of these Front Triplet Coverage values arange of possible combinations are suggested.In general Microphone Position Offset is used toobtain Critical Linking of the Front Triplet Seg-ments whereas Electronic Time Offset is nor-mally required to obtain Critical Linking of theLateral Segments. Microphone Position Offset,as its name implies, is obtained purely by thecorrect position and orientation of the micro-phones in the array, however, Electrical TimeOffset needs specific time delays to be intro-duced into certain channels. The sound record-ing engineer does not always have the possibilityto adjust Electrical Time Offset with the equip-ment available for a recording session. For thisreason a few selected arrays have been de-scribed which have Natural Critical Linking, i.e.the parameters of the array have been chosenso that no electrical offset is necessary to ob-tain Critical Linking. The orientation of each mi-crophone and their corresponding « X » and« Y » co-ordinates are the only parameters thathave to be set up to obtain Critical Linking withthese specific arrays.

The concept of Critical Linking means that theimpression of a continuous sound field is repro-duced no matter what the orientation of the lis-


9

tener within the loudspeaker set-up. This allows acertain freedom of rotational movement of thehead, but does not obviously change the need forthe listener to be in the centre of the loudspeakersystem, commonly called « the sweet spot ». But« Oh! How sweet » is this experience with a goodrecording made with a correctly designed Multi-channel Microphone Array!

MMA design procedure

General Overall StrategyAs with the choice of a dual microphone array forstereophonic sound recording, the first stage of ar-ray design is to determine the probable position ofthe microphone system, and therefore to measurethe angle of the sound source as « seen » by themicrophone array. Details for the construction of asuitable measuring instrument are given in the doc-ument “St Zoom” on this CD-ROM. With a stereo-phonic microphone array the Stereophonic Record-ing Angle (SRA) is chosen according to the amountof « side room » to be left on either side of thesound source. In general a small musical soundsource such as a quartet needs an SRA that isabout 10% wider than the actual sound source(the extra 10% on each side is called the « sideroom »). However for a wider sound source, suchas a Symphony Orchestra, the SRA is usually about10% smaller than the total sound stage. This en-ables better resolution in the centre part of the or-chestra, but obviously to the detriment of the ex-tremities of the orchestra, which will becomesomewhat crushed onto the left and right loud-speakers in stereophonic reproduction.

With the Multichannel Microphone Array we havethe added advantage of a wider reproduction stage.We are therefore not limited to the 60° of stereo-phonic reproduction. Although the complete use ofthe side segments for direct sound is not advisabledue to the progressively worsening localisationcharacteristics on the side, it is nevertheless possi-ble to exploit the sound reproduction further thanthe « 30° + 30° » defined by the front three loud-speakers. A 10° or 20° spread of the sound sourceinto the lateral segments is possible, with theadded advantage of a better feeling of envelopment

of the listener and, of course, considerably betterreproduction of early reflection groups. However itremains to be seen as to precisely how muchmore spread is possible with a wide sound sourceand also whether we still need to stay within thefront 60° with the smaller sources. In practicethis obviously means that we need to use a Multi-channel Microphone Array that allows us to fullyexploit the recording and reproduction character-istics of the lateral segments.

In the design of a dual microphone pair for stereo-phonic sound recording complete freedom existsto choose any combination of distance and angle,for the desired Coverage. This allows the soundengineer to choose any system from coincidentdirectional microphones through hybrid combina-tions of distance and angle to the purely time de-pendent spaced pair of omnidirectional micro-phones. In the multichannel array design a purelyintensity dependent system is not feasible if weare looking for a Critically Linked surround soundsystem. However careful examination of the multi-channel microphone array configurations pre-sented in the tables will demonstrate that it ispossible to select systems that have either Timeor Intensity Dominance according to the optionschosen in design. It remains to be seen whether aTime Dominant, or Intensity Dominant system, ispreferable for better localisation in the lateral seg-ments.

The completely «surround sound» characteristicsof natural reverberation mean that a continuoussound field pick-up around the array becomes anecessity. Attention to Critical Linking thereforebecomes of paramount importance and musttherefore be integrated into the total process ofMultichannel Microphone Array Design presentedin this document.

The majority of musical sound recording situa-tions that we encounter concern the reproductionof a limited sound stage as seen in the concerthall or theatre. The research for a satisfactory re-production of the completely surrounding directsound source, although technically very challeng-ing, is not part of our daily bread.


10

In general the Multichannel Microphone Array hasa considerable wing span somewhat like the prover-bial « albatross ». However careful analysis of themany arrays presented in this paper will show thatthis is not always the case. There are also somecircumstances, for instance when a minimum ofphysical size is important, in which case it is neces-sary to neglect the reproduction characteristics ofthe lateral segments. But we are perforce limitedto the 60° of the front sound stage (as with thestandard stereophonic sound stage) created by thefront three loudspeakers, and the far from satisfac-tory 140° of the back segment reproduction. How-ever in this document we are only concerned withgood continuous reproduction of the total surroundsound environment.

This still leaves us with the first major decision asto what coverage angle to adopt relative to the di-rect sound from the sound source. There are norules as to how to go about setting up a micro-phone array for multichannel sound recording sys-tem. Each sound recording engineer will have hisown ideas and preferences and this is of coursepart of the art of sound recording. However hereare some guidelines as to a possible procedure :

a) Determine the probable position of the micro-phone system

b) Decide on the required reproduction angle ofthe direct sound source. This will determine theCoverage Angle of the Front Triplet Array. Re-member that the Front Triplet Coverage will onlybe reproduced between the front three loud-speakers. Further “spread” will concern the sidesegments. If the Front Triplet Coverage angle issmaller than the sound source, then in repro-duction there will be spread into the side seg-ments. If the Front Triplet Coverage is widerthan the sound source, then the reproducedsound source will be within the front soundstage created by the front three loudspeakers.

c) Decide on the relative balance between BackSegment and Lateral Segment Coverage. Thiswill determine the Back Pair Coverage charac-teristics.

d) Choose your own preference concerning Inten-sity/Hybrid/Time Dominance. It is obviouslyphysically impossible to have Intensity or Time

Dominance for both the Back Segment andthe Lateral Segments; they will usually becomplimentary. Both can of course be Hybridcombination (equal dominance between Timeand Intensity).

e) Introduce, if needed, the correct ElectronicOffset values between the front Triplet andthe Back Pair.

f) Listen to the result! and start again until sat-isfied that you have achieved the optimum re-sult in the circumstances.

Critical Linking with Electronic Off-set versus Natural Critical LinkingThe various techniques needed to obtain CriticalLinking have been described in previous AESpreprints (9) (11) under chapters concerningMicrophone Position Time Offset (MPTO), Elec-tronic Time Offset (ETO) and Electronic IntensityOffset (EIO). Of these different types of offset,MPTO is of course the easiest to generate, as itis obtained simply by the physical position andorientation of the microphone. ETO however re-quires the fine adjustment of Time Delay be-tween the Front Triplet and the Back Pair, a facil-ity that is not always available on the standardmixing desk or even in post production with anAudio Workstation. On the other hand EIO is justa matter of introducing the correct constant In-tensity Difference, into the microphone arraywithin the Front Triplet or between the FrontTriplet and Back Pair (EIO can be used to gener-ate Critical Linking either in the design of theFront Triplet Coverage or in the Lateral SegmentCoverage). However it must be said that therange of Critical Linking that can be obtainedwith EIO processing is very much more limitedthan with ETO. Also the difference in level be-tween each component of the array may pro-duce an undesirable imbalance between differ-ent parts of the Array Coverage. Detailed consid-eration of the design of this type of array usingEIO is outside the scope of this paper.

Although « Natural Critical Linking » considerablylimits the choice in array design, it has a consid-erable advantage in that no electronic manipula-tion of Segment Coverage is required. CriticalLinking is obtained purely by microphone posi-


11

tion and orientation, not only for the FrontTriplet, but also between the Front Triplet, theLateral Segment and the Back Pair Coverage.This type of array is operational without any fur-ther signal processing. Microphone signals cantherefore be recorded direct to any 5 channelrecording system without the intervention of amore complex sound mixing or signal processingstage. This is obviously an important factor whendoing « on-location » recording with limited studiofacilities.

Tables 1, 2 & 3 show three sets of operationalMultichannel Microphone Arrays where no electrical offset is needed . The combination of « Table1a line 3 » together with « Table 1b line 3 » hasalready been presented at a paper given at the91st AES Convention in New York (5) some 10years ago! But this was before the seeds of themultichannel market were sown.

Table presentationSeven table groups are presented covering over220 possible microphone array configurations.The Front Triplet design is shown in eachTable « a », whilst Table « b » describes the cor-responding Back Pair and Lateral Segment Cover-age parameters. It is a remarkable feature of thedesign process that within the same table number , a specific combination chosen from Table« b » will automatically be Critically Linked withany combination chosen from Table « a » on con-dition that the correct ETO has been applied. TheLinking function depends on the Coverage Angleand not on the angle between the microphones.Variations of distance and angle used to obtainthe same Coverage Angle will automatically Criti-cally Link to neighbouring segments. Each « Table-set : a & b » will therefore cover between 30 and36 possible microphones arrays (except for table3 where Critical Linking is very limited and onlyone Back Pair configuration is shown). Two exam-ples of microphone lay-out are given in the figuresimmediately after each table-set.

MMAD – First stage - front tripletdesignThe first stage in design (as explained above) is


the choice of microphone array position, andthereafter the Front Triplet Coverage. A widerange of Front Triplet Coverage values are pre-sented in Tables 4 to 8, allowing the sound engi-neer to adjust the Front Triplet Coverage from amaximum of « 90° + 90° » to a minimum of« 50° + 50° ». Critical Linking is almost impossiblefor smaller values of Front Triplet Coverage.

MMAD – Second stage – back pairThe second stage in design is to determine the de-sired Back Segment Coverage. Again a widechoice of Coverage has been presented, from 90°to 40°, allowing the sound engineer to optimisethe reproduction of the back segment that usuallycovers the reverberant field. In most cases it isbetter not to overload back segment coverage asthis can be somewhat disconcerting when repro-duced in the segment behind the listener. Thismeans that in practice it is better to choose thelower values of Back Segment Coverage that willbe spread out over the 140° of back segment re-production and of course includes considerableAngular Distortion (10).

MMAD – Third stage – lateral seg-mentsThe third stage concerning the Lateral SegmentCoverage is usually considered to be the Coverageof the remaining surround sound-field angle, andwill be conditioned by the distance between theFront Triplet and the Back Pair. Critical Linking willbe obtained by the introduction of the correctamount of Electronic Time Offset, unless of coursea combination producing Natural Critical Linkinghas been selected.

Other possible configurationsFor the sake of simplicity and quick reference, onlycertain specific configurations have been pre-sented in the tables, but of course there are amultitude of other configurations possible withinthe continuum of possible microphone distanceand angle. However from the operational point ofview, the large range of choice, presented in ta-bles 1 to 8, should be enough to satisfy mostsound recording situations.

12

Microphone directivityAs with the design of a dual microphone stereo-phonic array, we are not restricted to the use ofcardioid microphones. It is the intention of the au-thors of this document to present, in future AES

papers, not only a choice of microphone arraysusing other first order directivity patterns, butalso to suggest a wide range of possible hybridarrays using different directivity patterns for eachpart of the array structure.

From the above table (Front Triplet Coverage of72° + 72°) note « the microphone orientation »,« X » co-ordinates and « Y1 » co-ordinates forthe desired configuration of the Front Triplet.Then choose any combination « distance /angle » from the table below and note the dis-tance between the left and right Back Pair of

TABLE 1aFRONT TRIPLET COVERAGE 72° + 72°

Microphoneorientation

Distancebetween

microphones

X co-ordinates Y1 co-ordinates (forboth left and right

microphones)

Microphone PositionTime Offset (MPTO)

90° (R) + 30.5 cm270° (L) 35 cm - 30.5 cm 17 cm -15.6°80° (R) + 31 cm280° (L) 37 cm - 31 cm 20.5 cm -6°72° (R) + 31.5 cm288° (L) 39 cm - 31.5 cm 23 cm No offset60° (R) + 33 cm300° (L) 42.5 cm - 33 cm 26.5 cm +9°50° (R) + 34 cm310° (L) 45 cm - 34 cm 29.5 cm +15.5°40° (R) + 36.5 cm320° (L) 48.5 cm - 36.5 cm 31.5 cm +20.9°

TABLE 1bLATERAL PAIRS BACK PAIR

LateralSegmentCoverage

ElectronicTime Offset

(ETO)

BackSegmentCoverage

Micorientation

Anglebetween

mics

Distancebetween mics

Y2 co-ord

160°(R)72° No offset 72° 200°(L) 40° 48 cm 58.5 cm

155°(R)72° No offset 72° 205°(L) 50° 45 cm 59 cm

144°(R)72° No offset 72° 216°(L) 72° 39 cm 60 cm

135°(R)72° No offset 72° 225°(L) 90° 34.5 cm 62 cm

130°(R)72° No offset 72° 230°(L) 100° 32 cm 63.5 cm

microphones and their orientation, and the« Y2 » co-ordinate of the back pair (with respectto the front facing centre microphone). This is allthe information that you need to set up this « nooffset » array which covers the complete sur-round sound field in five equal segments of 72°each.


13


ms

ms

Figure 8Front Triplet -Table 1a / line 1Back Pair -Table 1b / line 1


14



Distancebetween

microphones


microphones)


90° (R) + 42.5 cm270° (L) 46 cm - 42.5 cm 17 cm -23°80° (R) + 43.5 cm

280° (L) 48 cm - 43.5 cm 20.5 cm -15°72° (R) + 44.5 cm288° (L) 50.2 cm - 44.5 cm 23 cm - 7°60° (R) - 46 cm300° (L) 53 cm - 46 cm 26.5 cm No Offset50° (R) + 47.5 cm310° (L) 56 cm - 47.5 cm 29.5 cm +7°40° (R) + 50 cm320° (L) 59 cm - 50 cm 31.5 cm +12°

From the above table (Front Triplet Coverage of60° + 60°) note « the microphone orientation »,« X » co-ordinates and « Y1 » co-ordinates for thedesired configuration of the Front Triplet. Thenchoose the combination « distance / angle » fromthe table below and note the distance betweenthe left and right Back Pair microphone and their

orientation, and the « Y2 » co-ordinate of the backpair (with respect to the front facing centre micro-phone). This is all the information that you need toset up this « no offset » array which covers the com-plete surround sound field in five segments (the Leftand Right Front Segments of 60° each, two sidesegments of 97.5° and a back segment of 45°).




(ETO)

Back SegmentCoverage

Micorientation

Anglebetween mics


Y2 co-ord

148°(R)97.5° No offset 45° 212°(L) 64° 73 cm 46.5 cm

138°(R)97.5° No offset 45° 222°(L) 84° 67.5 cm 48 cm

128°(R)97.5° No offset 45° 232°(L) 104° 62 cm 49.5 cm

118°(R)97.5° No offset 45° 242°(L) 124° 58 cm 51 cm

130°(R)98° No offset 44° 230°(L) 144° 54 cm 53 cm


15

ms

ms




16




(ETO)

BackSegmentCoverage

Micorientation

Anglebetween mics


Y2 co-ord

140°(R)114° No offset 32° 220°(L) 80° 103 cm 44 cm

No other solutions exist for values of Back SegmentCoverage, these being the only values where inter-

section between the physical parameters of themicrophone array and the psychoacoustical limitsare possible.




Distancebetween

microphones


microphones)


90° (R) + 58.5 cm270° (L) 61 cm - 58.5 cm 17 cm -28.5°80° (R) + 58 cm

280° (L) 62 cm - 58 cm 21 cm -20°72° (R) + 59 cm

288° (L) 63.9 cm - 59 cm 24 cm -1360° (R) + 61 cm

300° (L) 66.5 cm - 61 cm 27 cm -6°50° (R) + 63 cm

310° (L) 69.5 cm - 63 cm 29.5 cm No Offset40° (R) + 65 cm

320° (L) 72.5 cm - 65 cm 32 cm +6°

From the above table (Front Triplet Coverage of50° + 50°) note « the microphone orientation »,« X » coordinates and « Y1 » coordinates for thedesired configuration of the Front Triplet. Thenchoose the combination « distance / angle » fromthe table below and note the distance betweenthe left and right Back Pair microphones, and their

orientation, and the « Y2 » coordinate of theBack Pair (with respect to the front facing centremicrophone). This is all the information that youneed to set up this « no offset » array which cov-ers the complete surround sound field in five seg-ments (the Left and Right Front Segments of 50°each, two side segments of 114° and a backsegment of 32°.

17

ms

ms




18



Distancebetween

microphones


microphones)


90° (R) + 17.3 cm

270° (L) 24.5 cm - 17.3 cm 17.3 cm No Offset

80° (R) + 17.5 cm

280° (L) 27 cm - 17.5 cm 20.5 cm +9.2°

72° (R) + 18 cm

288° (L) 29.5 cm - 18 cm 23.5 cm +18°

60° (R) + 19 cm

300° (L) 32.5 cm - 19 cm 26.5 cm +24.5°

50° (R) + 20 cm

310° (L) 35.5 cm - 20 cm 29 cm +30.5°

40° (R) + 22 cm

320° (L) 38.5 cm - 22 cm 31.5 cm +35.5°

From the above table (Front Triplet Coverage of90° + 90°) note « the microphone orientation »,« X » co-ordinates and « Y1 » co-ordinates for thedesired configuration of the Front Triplet. Thenchoose the combination « distance / angle » fromthe table below and note the distance betweenthe left and right Back Pair microphones, and their

orientation, and the « Y2 » co-ordinate of thethe Back Pair (with respect to the front facingcentre microphone) and corresponding ETO(Electronic Time Offset). Please note that a nega-tive ETO means that the Back Pair will be delayedwith respect to the Front Triplet by the requisiteamount. This is all the information that you needto set up the array.




(ETO)

BackSegmentCoverage

Micorientation

Anglebetween mics


Y2 co-ord

140°(R)45° - 0.98 ms 90° 220°(L) 80° 27 cm 103 cm

140°(R)50° - 0.9 ms 80° 220°(L) 80° 32 cm 94 cm

135°(R)55° - 0.94 ms 70° 225°(L) 90° 36 cm 90 cm

135°(R)60° - 1.05 ms 60° 225°(L) 90° 45 cm 86.5 cm

130°(R)65° - 1.23 ms 50° 230°(L) 100° 55 cm 87.5 cm

120°(R)70° - 1.5 ms 40° 240°(L) 120° 68 cm 98 cm


19


ms


ms


20

From the above table (Front Triplet Coverage of 80°+ 80°) note « the microphone orientation », « X »co-ordinates and « Y1 » co-ordinates for the de-sired configuration of the Front Triplet. Then choosethe combination « distance / angle » from the tablebelow and note the distance between the left andright Back Pair microphones, and their orientation,

and the « Y2 » co-ordinate of the the Back Pair(with respect to the front facing centre micro-phone) and corresponding ETO (Electronic TimeOffset). Please note that a negative ETO meansthat the Back Pair will be delayed with respect tothe Front Triplet by the requisite amount. This isall the information that you need to set up thearray.


Lateral SegmentCoverage


(ETO)

BackSegmentCoverage

Micorientation

Anglebetween mics


Y2 co-ord

140°(R)55° - 0.75 ms 90° 220°(L) 80° 27 cm 85.3 cm

140°(R)60° - 0.7 ms 80° 220°(L) 80° 32 cm 80 cm

135°(R)65° - 0.7 ms 70° 225°(L) 90° 36 cm 76.5 cm

135°(R)70° - 0.6 ms 60° 225°(L) 90° 45 cm 71 cm

130°(R)75° - 0 .7 ms 50° 230°(L) 100° 55 cm 71 cm

120°(R)80° - 0.76 ms 40° 240°(L) 120° 68 cm 76 cm




Distance betweenmicrophones


microphones)


90° (R) + 24 cm270° (L) 29.5 cm - 24 cm 17.5 cm -9°

80° (R) + 24.5 cm280° (L) 32 cm - 24.5 cm 20.5 cm No Offset

72° (R) + 25 cm288° (L) 34.5 cm - 25 cm 23.5 cm +8°

60° (R) + 26.5 cm300° (L) 3.5 cm - 26.5 cm 26.5 cm +15°

50° (R) + 27.5 cm310° (L) 40 cm - 27.5 cm 29.5 cm +22°

40° (R) + 29.5 cm320° (L) 43 cm - 29.5 cm 31.5 cm +27°

21

ms

ms

Figure 16Table 5a / line 1Table 5b / line 1



22



Distancebetween

microphones


microphones)


90° (R) + 30.5 cm

270° (L) 35 cm - 30.5 cm 17 cm -15.6°

80° (R) + 31 cm

280° (L) 37 cm - 31 cm 20.5 cm -6°

72° (R) + 31.5 cm

288° (L) 39 cm - 31.5 cm 23 cm No offset

60° (R) + 33 cm

300° (L) 42.5 cm - 33 cm 26.5 cm +9°

50° (R) + 34 cm

310° (L) 45 cm - 34 cm 29.5 cm +15.5°

40° (R) + 36.5 cm

320° (L) 48.5 cm - 36.5 cm 31.5 cm +20.9°

From the above table (Front Triplet Coverage of72° + 72°) note « the microphone orientation »,« X » co-ordinates and « Y1 » co-ordinates for thedesired configuration of the Front Triplet. Thenchoose the combination « distance / angle » fromthe table below and note the distance between theleft and right Back Pair microphones, and their ori-entation, and the « Y2 » co-ordinate of the Back

Pair (with respect to the front facing centre mi-crophone) and corresponding ETO (ElectronicTime Offset). Please note that a negative ETOmeans that the Back Pair will be delayed with re-spect to the Front Triplet by the requisite amount.And inversely, a positive ETO means that the FrontTriplet is delayed with respect to the Back Pair.This is all the information that you need to set upthe array.


TABLE 6bLATERALPAIRS

BACK PAIR



(ETO)

BackSegmentCoverage

Micorientation

Anglebetween mics


Y2 co-ord

140°(R)63° + 0.2 ms 90° 220°(L) 80° 27 cm 67.5 cm

140°(R)68° + 0.07 ms 80° 220°(L) 80° 32 cm 63.5 cm

135°(R)72° No offset 70° 225°(L) 90° 39 cm 60 cm

135°(R)78° - 0.19 ms 60° 225°(L) 90° 45 cm 60.1 cm

130°(R)83° - 0 .36 ms 50° 230°(L) 100° 55 cm 61.1 cm

120°(R)88° - 0.43 ms 40° 240°(L) 120° 68 cm 62 cm

23

ms

ms




24

TABLE 7aFRONT TRIPLET COVERAGE 60° + 60°Microphoneorientation

Distancebetween

microphones

X co-ordinates Y1 co-ordinates(for both left and

right micro-phones)

MicrophonePosition TimeOffset (MPTO)

90° (R) + 42.5 cm270° (L) 46 cm - 42.5 cm 17 cm -23°80° (R) + 43.5 cm280° (L) 48 cm - 43.5 cm 20.5 cm -15°72° (R) + 44.5 cm288° (L) 50.2 cm - 44.5 cm 23 cm - 7°60° (R) + 46 cm300° (L) 53 cm - 46 cm 26.5 cm No Offset50° (R) + 47.5 cm310° (L) 56 cm - 47.5 cm 29.5 cm +7°40° (R) + 50 cm320° (L) 59 cm - 50 cm 31.5 cm +12°

From the above table (Front Triplet Coverage of 60°+ 60°) note « the microphone orientation », « X »co-ordinates and « Y1 » co-ordinates for the de-sired configuration of the Front Triplet. Then choosethe combination « distance / angle » from the tablebelow and note the distance between the left andright Back Pair microphones, and their orientation,and the « Y2 » co-ordinate of the Back Pair (with

respect to the front facing centre microphone)and corresponding ETO (Electronic Time Offset).Please note that a negative ETO means that theBack Pair will be delayed with respect to theFront Triplet by the requisite amount. And in-versely, a positive ETO means that the FrontTriplet is delayed with respect to the Back Pair.This is all the information that you need to set upthe array.




(ETO)

BackSegmentCoverage

Micorientation

Angle between mics


Y2 co-ord

140°(R)75° + 0.69 ms 90° 220°(L) 80° 27 cm 57 cm

140°(R)80° + 0.56 ms 80° 220°(L) 80° 32 cm 53 cm

135°(R)85° + 0.42 ms 70° 225°(L) 90° 36 cm 51 cm

135°(R)90° + 0.28 ms 60° 225°(L) 90° 45 cm 49 cm

130°(R)95° + 0.1 ms 50° 230°(L) 100° 55 cm 49 cm

120°(R)100° - 0.1 ms 40° 240°(L) 120° 68 cm 52 cm


25

ms

ms




26

From the above table (Front Triplet Coverage of50° + 50°) note « the microphone orientation »,« X » co-ordinates and « Y1 » co-ordinates for thedesired configuration of the Front Triplet. Thenchoose the combination « distance / angle » fromthe table below and note the distance betweenthe left and right Back Pair microphones, and their

orientation, and the « Y2 » co-ordinate of thethe Back Pair (with respect to the front facingcentre microphone) and corresponding ETO(Electronic Time Offset). Please note that a posi-tive ETO means that the Front Triplet is delayedwith respect to the Back Pair by the requisiteamount.. This is all the information that you needto set up the array




(ETO)

BackSegmentCoverage

Micorientation

Anglebetween mics

Distancebetween

mics

Y2 coord

140°(R)90° No solution 80° 220°(L) 80° 32 cm -

135°(R)95° No solution 70° 225°(L) 90° 36 cm -

135°(R)100° No solution 60° 225°(L) 90° 45 cm -

130°(R)105° + 0.5 ms 50° 230°(L) 100° 55 cm 42.5 cm

120°(R)110° + 0.28 ms 40° 240°(L) 120° 68 cm 45.5 cm

140°(R)114° No Offset 32° 220°(L) 120° 103 cm 52.5 cm


TABLE 8aFRONT TRIPLET COVERAGE 50° + 50°Microphoneorientation

Distancebetween

microphones

X co-ordinates Y1 co-ordinates(for both left and

right microphones)

MicrophonePosition TimeOffset (MPTO)

90° (R) + 58.5 cm270° (L) 61 cm - 58.5 cm 17 cm -28.5°

80° (R) + 58 cm280° (L) 62 cm - 58 cm 21 cm -20°

70° (R) + 59 cm288° (L) 63.9 cm - 59 cm 24 cm -13

60° (R) + 61 cm300° (L) 66.5 cm - 61 cm 27 cm -6°

50° (R) + 63 cm310° (L) 69.5 cm - 63 cm 29.5 cm No Offset

40° (R) + 65 cm320° (L) 72.5 cm - 65 cm 32 cm +6°

27

ms

ms




28

This version of MMAD©2002 Rycote Microphone Windshields Ltdand Human Computer Interface

References1 1984 : 74th AES Convention in Paris –preprint 2072« The Stereophonic Zoom, A Practical Ap-proach to determining the Characteristics of aSpaced Pair of Microphones » by MichaelWilliams.

2 1987 : 82nd AES Convention in London –preprint 2466« Unified Theory of Microphone Systems forStereophonic Sound Recording »by Michael Williams

3 1990 : 88th AES Convention in Montreux –preprint 2931« Operational Limits of the Variable M/SStereophonic Microphone System »by Michael Williams

4 1991 : 91st AES Convention in New York –preprint 3155« Early Reflections and Reverberant Field Dis-tribution in Dual Microphone StereophonicSound Recording Systems » by MichaelWilliams

5 1991 : 91st AES Convention in New York –preprint 3157« Microphone Arrays for Natural Multiphony »by Michael Williams.

6 1992 : 92nd AES Convention in Vienna –preprint 3252« Frequency Dependent Hybrid Microphone Ar-rays for Stereophonic Recording »by Michael Williams.7 « The Stereophonic Zoom » by MichaelWilliams

8 1999 : 107th AES Convention in New York :Preprint 4997« Microphone Array Analysis for MultichannelSound Recording »by Michael Williams and Guillaume Le Dû

9 2000 : 108th AES Convention in Paris :Preprint 5157« Multichannel Microphone Array Design » byMichael Williams and Guillaume Le Dû

10 2000 : VDT in Hannover« Loudspeaker Configuration and ChannelCrosstalk in Multichannel Microphone Array De-sign » by Michael Williams and Guillaume Le Dû

11 2001 : 110th AES Convention in Amsterdam :preprint 5336« The Quick Reference Guide to Multichannel Mi-crophone Arrays , Part 1: using Cardioid Micro-phones » by Michael Williams and Guillaume Le Dû

12 2002 : 112th AES Convention in Munich :preprint 5567« Multichannel Microphone Array Design: Seg-ment Coverage Analysis above and below the Hori-zontal Reference Plane » by Michael Williams

Michael Williams started his professional career atthe BBC Television Studios in London in 1960. In1965 he moved to France to work for ''Societe Au-dax'' in Paris developing loudspeakers for profes-sional sound and television broadcasting, and laterworked for the ''Conservatoire National des Arts etMetiers'' in the adult education television service. In1980 he became a free-lance instructor in AudioEngineering and Sound Recording Practice, workingfor most of the major French national television andsound broadcasting companies, as well as manytraining schools and institutions. He is an activemember of the Audio Engineering Society, and haspublished many papers on Stereo and MultichannelRecording Systems over the past twenty years. Heis at present the AES Publications Sales Represen-tative in Europe.


Documents

Multichannel sound recording · ceived, for the four channel system, purely as an extension and improvement of the original Quadraphonic Recording/Reproduction System, and it seems,