La calidad del audio

Enrique Alexandre (@e_alexandre)

13

2.1 BACK TO THE BEGINNING: CAPTURING SOUND QUALITY

In terms of basic sound quality, claims of accurate reproduction began early. Edison, in 1901, claimed that the phonograph had no “tone” of its own. To prove it, he mounted a traveling show in which his phonograph was demon-strated in “tone tests” that consisted of presentations with a live performer. Morton (2000) reports, “Edison carefully chose singers, usually women, who could imitate the sound of their recordings and only allowed musicians to use the limited group of instruments that recorded best for demonstrations” (p. 23). Of a 1916 demonstration in Carnegie Hall before a capacity audience of “musi-cally cultured and musically critical” listeners, the New York Evening Mail reported that “the ear could not tell when it was listening to the phonograph alone, and when to actual voice and reproduction together. Only the eye could discover the truth by noting when the singer’s mouth was open or closed” (quoted in Harvith and Harvith, 1985, p. 12).

Singers had to be careful not to be louder than the machine, to learn to imitate the sound of the machine, and to sing without vibrato, which Edison (apparently a musically uncultured person) did not like. There were other con-sequences of these tests on recordings. The low sensitivity of the mechanical recording device made it necessary for the performers to crowd around the mouth of the horn and fi nd instruments that could play especially loud. Because

Back to the Beginning: Capturing Sound Quality

FIGURE 2.1 Singer Frieda Hempel stages a tone test at the Edison studios in New York City, 1918. Care was taken to ensure that the test was “blind,” but it is amusing to see that some of the blindfolds also cover the ears. Courtesy of Edison National Historic Site, National Park Service, U.S. Department of the Interior.

Evaluación de la calidad

Tests objetivosTests subjetivos Modelos objetivos

Criterios objetivos

• Existen diversos criterios objetivos para medir la calidad de una señal de audio:

• Relación señal a ruido (SNR)

• Relación señal a ruido segmental (segSNR)

• Sin embargo, en audio la evaluación de la calidad es una tarea fundamentalmente subjetiva, aunque puede aproximarse mediante modelos objetivos.

Criterios subjetivos

• La calidad subjetiva se puede obtener mediante experimentos de escucha, por ejemplo:

• Comparar la señal con una referencia de calidad “perfecta”

• Comparar varios sonidos y ordenarlos por orden de preferencia

• Dar una “puntuación” global en una escala numérica

Doble test ciego (A-B-X)

• Se dispone de dos señales A y B, siendo una de ellas la referencia y la otra la señal a testear.

• Al oyente se le presenta una tercera señal, X, y se le pide que identifique si se trata de A o de B.

• El resultado tiene que ser estadísticamente significativo para poder ser tenido en cuenta.

Fiabilidad de los resultados

• Para tener un nivel de confianza del 95%:

Número de ensayos 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25Número de aciertos 9 9 10 10 11 12 12 13 13 14 15 15 16 16 17 18

Test A-B-C

• También llamado doble test ciego con triple estímulo y referencia oculta.

• La señal A siempre es la referencia. Las otras dos (B y C) contienen, aleatoriamente, la señal de referencia y la de test.

• El oyente tiene que puntuar las dos señales B y C de acuerdo a su calidad en una escala de 1 a 5.

• El resultado final es la diferencia de las dos puntuaciones (SDG, Subjective Difference Grade)

Tests subjetivos

A tener en cuenta

• Es IMPRESCINDIBLE:

• Que las señales tengan el mismo nivel

• Que estén sincronizadas en tiempo

• Que no se produzcan chasquidos al conmutar entre una y otra

Limitaciones de las pruebas subjetivas

• Los resultados están muy influenciados por factores humanos (nivel de reproducción, ruido de fondo, etc.)

• La habilidad y los gustos del oyente también influyen

• Son muy caros

Modelos objetivos de calidad percibida

• Existen varios estándares internacionales para evaluar de forma objetiva la calidad percibida de una señal de audio:

• PEAQ (Perceptual Evaluation of Audio Quality)

• PESQ (Perceptual Evaluation of Speech Quality)

• SII (Speech Intelligibility Index)

18

AAC LC 96:

Better than MP2 for no items, worse for no items, equivalent for 10 items.

Better than MP3 for 1 item, worse for no items, equivalent for 9 items.

AAC SSR 128:

Better than MP2 for 1 item, worse for no items, equivalent for 9 items.

Better than MP3 for 2 items, worse for no items, equivalent for 9 items.

Thus, we see that only the Main 96 codec is outperformed by any MP2 or MP3 codec for any of

these examples. For many programme items, an AAC coder gives statistically superior results.

Note that for items Tracy Chapman, Ornette Coleman and Dire Straits there were no significant

differences between codecs – all codecs performed the same on these examples.

10.6. Comparison with MPEG-1 codecs

“Is the performance of AAC codecs at the tested bitrate equal to or better than the performance

of MPEG-1 Layer II and Layer III?” The accumulated results by codec are shown in Figure 5

(note the foreshortened vertical scale).

220220220220220220220N =

CODER

MP3 128

MP2 192

AACSSR 128

AAC LC 96

AAC LC 128

AACMain 96

AACMain 128

.2

0.0

-.2

-.4

-.6

-.8

-1.0

-1.2

-1.4

-1.6

-1.8

-2.0

Diffscores

Figure 5. Overall results (averaged across programme items and position) for each coder.

We see from this figure that overall, AAC Main 128, AAC LC 128, and AAC SSR 128 give

significantly better performance than do MP2 192 or MP3 128. In addition, AAC Main 96

gives better results than MP3 128. There is no statistically significant improvement between

AAC LC 96 and the MPEG-1 codecs.

Within the AAC codec group, AAC Main 128, AAC LC 128, and AAC SSR 128 are all

superior to AAC LC 96. In addition, AAC Main 128 and AAC LC 128 are superior to AAC

Main 96.

Comparativa