LOUDNESS COMPENSATION BY MICROPHONE FREQUENCY RESPONSE AND MICROPHONE PLACEMENT

Bruce Bartlett, AES Member, Crown International, Elkhart, Indiana

             AES Working Group SC-04-04

 

 

 INTRODUCTION

 

"Specs don't tell how a microphone sounds in the real world!" This common complaint is being addressed by the AES working group, SC-04-04. The group was formed to propose new microphone specifications that help users better determine whether a microphone is suitable for certain applications. With these new specifications, manufacturers can present information to users that helps them choose a microphone best suited to their needs.

 

Abstract. Rock bands produce high SPLs when heard live. But recordings of these bands are reproduced at much lower SPLs. Consequently, the reproduction may sound weak in the bass and treble compared to the live band. Some microphones compensate for this loss: they boost the lows with proximity effect (and nearfield placement) and boost the highs with a "presence peak." These microphones are often chosen as being the "best sounding" for recording loud instruments. The author suggests a new specification that quantifies this compensation.

Here is a recording situation that suggests the need for a new specification

A typical rock band produces 110 dB SPL at the listener's ears. But a recording of this band is typically played back at 85 dB SPL. As shown by Fletcher & Munson and others, a program heard at a low SPL sounds weaker in the bass and treble than the same program heard at a high SPL. So the reproduction (at 85 dB SPL) sounds deficient in lows and highs compared to the original live event (at 110 dB SPL).

Some microphones partly compensate for this effect by boosting the bass (with proximity effect) and the treble (with a "presence peak"). These microphones are often chosen as being the "best sounding" for recording loud instruments.

The author suggests a new specification that quantifies this compensation. Such a specification may help users determine whether a microphone will be useful to record loud music and reproduce it with a realistic tonal balance at less-than-live SPLs. The new specification is intended to answer the question, "Will this microphone sound good on loud guitar amps and drums?"

The new specification is called "Loudness Compensation." It indicates how well a microphone compensates for the subjective change in tonal balance between loud and quiet music levels. The author suggests that this new specification be included in the revised microphone standard IEC 268-4.

The loudness compensation spec is an overlay of two amplitude / frequency curves: (1) the microphone frequency response at 2 inches (5 cm) from a 12-inch (30.5 cm) diameter loudspeaker (including any proximity effect), and (2) the difference between Robinson-Dadson equal-loudness contours at 110 phons and 80 phons. This preprint explains the need for such a specification, and the reasons behind its parameters.

 

BACKGROUND: EQUAL-LOUDNESS CONTOURS

The proposed specification is based on the equal-loudness contours, which are explained below.

A simple way to quantify the loudness of a sound is to compare it with some standard sound-a 1000 Hz tone. The loudness level of a tone at any other frequency is the SPL of a 1000 Hz tone that sounds as loud as the other tone. The unit of loudness level is the phon [1 ].

For example, if a 100 Hz tone (at any SPL) sounds as loud as a 1 kHz tone at 90 dB SPL, the 100 Hz tone has a loudness level of 90 phons.

Fletcher and Munson, and others, made measurements to determine the loudness levels of pure tones (or narrow bands of noise) as a function of frequency and sound pressure level. The family of curves from these measurements is called the equal-loudness contours. Fletcher and Munson derived the equal-loudness contours in the following way

 [1 ][2]:

A large group of listeners with normal hearing sat one at a time in an anechoic chamber.

2. Each subject listened to test tones from a loudspeaker directly in front

at a distance greater than 1 meter.

3. Loudspeaker playback was set to a certain SPL at 1 kHz.

4. Using a volume control, each subject matched the perceived loudness of the 1 kHz tone with the perceived loudness of tones of other frequencies. The amplitude compensation in dB for each frequency was noted.

The sound pressure levels of the tones were measured with the listener out of the sound field, at a point corresponding to the center of the head.

This procedure was repeated at several phon levels from 0 to 120 phons.

The results for all the listeners were averaged.

The results are shown in Figure 1, Equal=Loudness Contours (Fletcher­Munson, 1933) [3].

Figure 2 shows the same data taken by Robinson and Dadson in 1956 [4]. In 1986 these curves became the International Standard ISO 226.

Figure 3 shows the equal-loudness contours calculated by Moore et al [5], based on models of human hearing.

Each phon value describes a contour, or curve, of equal loudness as a function of frequency. When the SPL at frequencies from 20 Hz to 12.5 kHz is adjusted to follow a phon contour, the typical listener hears all frequencies to be equal in loudness to the 1000 Hz reference tone. At the reference frequency of 1 kHz, the loudness level in phons is the same as the actual value of SPL.

The equal-loudness contours show the subjective frequency response of the human ear at various sound pressure levels.

 

CHANGE IN EQUAL-LOUDNESS CONTOURS AT DIFFERENT SPLs

(SCALE DISTORTION)

 

Note that the contours at different phon levels are not parallel. In other words, the ear is less sensitive to low frequencies and high frequencies (around 4 kHz) at low SPLs than at high SPLs.

This fact has implications for the tonal balance of live vs. reproduced music. As the listening level is decreased, the listener hears less bass and treble. Music played at a low SPL sounds weaker in the bass and treble than the same music played at a high SPL. For example, we hear less bass and treble (around 4 kHz) at home listening levels than at live rock­ music levels.

M.G. Scroggie proposed a term "scale distortion" for the change in subjective "frequency response" (strictly "spectral distribution") that occurs when sound is reproduced at other than its original level [6]. Scale distortion is another term for the normalized difference in SPL vs. frequency between two equal-loudness contours. Subjectively, it is the change in perceived tonal balance at different average SPLs.

Suppose a rock band produces 110 dB SPL average at the listeners' ears. A recording is made of this band using microphones with a flat frequency response. If the recording is played back at 85 dB SPL-atypical home listening level-the reproduction will lack bass and treble compared to the live band. In other words, the reproduction will not sound as "punchy" as the live band, because we hear tonal balance differently at different loudness levels.

That is the theory, and it has also been the author's experience. Rock bands recorded with flat-response microphones sounded dull in the upper midrange and weak in the bass compared to the live band.

For simplicity, the author suggests that the proposed specification be based on just two SPLs or phon levels: 110 dB and 80 dB. The 110 dB contour represents a live rock-band level, and the 80 dB SPL contour represents a typical home playback level.

How were these figures derived? To determine a typical live rock-band level, the author researched various sources and made measurements.

Listed below are some typical SPLs of loud live music: 'Orchestra playing triple forte at 30 feet: 100 to 130 dB SPL [7]. '4 trumpets playing triple forte at 5 feet: 90 to 118 dB SPL [7]. 'Piano at 4 feet: 85 to 110 dB SPL [7]

'Loud parts at a rock concert: 110 d8 SPL A-weighted (distance not given) [8].

'Disco with a decent sound system: 110 dB SPL A-weighted (distance not given) [8].

'Drum set played hard: 108 dB SPL unweighted at 10 feet.

'Small practice guitar amplifier played at full volume: 102 dB SPL unweighted at 10 feet.

The last two values were measured by the author using a Bruel & Kjaer'/V free-field microphone and a Crown TEF-20 sound analyzer running RTA software.

Based on all these measurements, the typical SPL of live rock-band instruments appears to be around 110 dB SPL.

A typical home listening level is 85 dB SPL [9]. Because there is no equal­loudness contour for 85 dB SPL, the author suggests that the 80 dB SPL contour be used to represent a typical home listening level. Indeed, 80 dB SPL appears to be a typical level of an instrument within an 85 dB SPL mix. If a typical home listening level for a band is 85 dB SPL, the level of an instrument in the band must be less than 85 dB SPL-for example, 80 dB SPL. The author's experience with mixing levels confirms this.

So far we have determined two useful phon levels for the loudness compensation specification.

Using the equal-loudness contours, the author calculated the dB difference vs. frequency between the 110 phon contour and the 80 phon contour. This difference curve was normalized at 1 kHz. Results are shown based on the Fletcher-Munson contours (Fig. 4),

The general trend is a low-frequency rise below 100 Hz (+7.5 dB at 30 Hz) and a 4 dB rise centered at 3 to 4 kHz. In theory, that is the equalization required to make a program reproduced at 80 dB SPL match the tonal balance of the same program heard at 110 dB SPL. Note that the difference vs. frequency for the Moore contours shows an upper midrange dip rather than a rise. Since these results do not agree with experience, they should not be included in the proposed specification.

 

 COMPENSATING FOR TONAL DIFFERENCES AT DIFFERENT SPLs

To compensate for the subjective loss in lows and upper-mids at low listening levels, recording engineers may use any or all of the following methods:

'Use equalization to boost low and high frequencies.

'Place each microphone near the sound source so as to pick up stronger lows or highs.

'Choose microphones with boosted lows and highs in their frequency

response.

In the last method, the boosted lows come from proximity effect, and the boosted highs come from a "presence peak" in the microphone's frequency response-typically a rise of several dB around 4 to 6 kHz.

Proximity effect occurs in pressure-gradient (directional) microphones with a "single-D" design. In this design, all the rear sound entries are the same distance from the microphone diaphragm.

Proximity effect is the progressive rise in low-frequency response as the microphone is placed closer to the sound source. For example, most cardioid microphones sound more bassy close to the mouth than far from the mouth.

As the microphone is placed closer to a point source, the wavefront changes gradually from plane to spherical. Proximity effect increases as the wavefront becomes more spherical.

The bass and treble boost of a microphone with proximity effect and a presence peak compensates somewhat for the subjective change in the ears' frequency response when the playback level is less than the original live level. Hence, microphones with proximity effect and a presence peak are very popular on loud instruments such as drums and guitar amplifiers. The microphones have a contoured frequency response that makes the reproduced instruments sound more realistic or natural than would microphones with a flat frequency response.

Let us state this another way. In loud rock music, a typical sound pressure level at the listener is 110 dB SPL. But recordings of this music are typically played back at 85 dB SPL. The difference in equal-loudness contours between 85 and 110 dB SPL is a "smile-shaped" curve: rising low frequencies and rising high frequencies (around 4 kHz). Recording engineers get this response curve when they close-mike with a pressure ­gradient microphone having a presence peak. These microphones with "smile-shaped" frequency-response curves are often picked as the best sounding [10].

To summarize, the effect of scale distortion is a perceived loss in bass and treble frequencies when playback level is less than the original live level. Compensation for scale distortion is a complementary bass and treble boost. In a microphone, this boost comes from proximity effect and a presence peak.

For several years, many consumer preamplifiers included loudness controls which boosted the bass (and sometimes the treble) for low-level listening. A typical boost is 20 dB at 100 Hz relative to 1 kHz [2]. Fig. 7 shows the frequency response of the IRC loudness control Model LC-1 [11 ].

A SUGGGESTED NEW SPECIFICATION

The author suggests that the microphone industry create a new microphone specification-loudness compensation-that indicates how well a microphone compensates for the subjective change in tonal balance between loud and quiet music levels.

Such a specification may suggest how well a microphone can record loud music and reproduce it with a realistic tonal balance at less-than-live SPLs The new specification may answer such questions as, "Will this microphone sound good on loud guitar amps and drums?" "Will the tonal balance conveyed by this microphone at normal listening levels sound like the tonal balance heard at a live concert?"

The current specification for microphone frequency response is not useful for that purpose. Typically, microphone frequency response is measured in (approximately) a plane wave at 2 feet or 1 meter from the sound source. But in real-world studios, the miking distance is usually just a few inches or centimeters, so proximity effect is usually part of the microphone's "sound."

Also, the frequency-response graph in current data sheets does not show the frequency compensation needed to give loud sounds a natural tonal balance at a typical home listening level.

The proposed specification includes three elements

The usual dB vs. frequency grid with a logarithmic frequency scale.

2. The frequency response of the microphone measured at 2 inches [5 cm] from a 12-inch (30.5 cm) diameter loudspeaker in order to include proximity effect.

3. An overlay of the difference between two Robinson-Dadson equal­loudness contours: 110 phons and 80 phons.

Figure 8 is an example of such a specification.

 

4 THE REASONS FOR EACH SPECIFICATION PARAMETER

The suggested specification includes parameters for the mic-to-source distance, sound source, and difference between two equal-loudness contours. This section explains why each parameter was chosen.

4.1 Mic-to-source distance

In a typical popular-music recording, the recording engineer attempts to reduce pickup of room acoustics and leakage. To do this, microphones are placed a few inches or centimeters from their sound sources. Also, the engineer often uses cardioid microphones, most of which have proximity effect (up-close bass boost).

To measure a microphone for this specification, I suggest that the measurement be done with the microphone grille 2 inches (5 cm) from the sound source. 2 inches (5 cm) is a typical mic-to-source distance for recording guitar amplifiers and drums. For example, a microphone is typically placed 2 inches (5 cm) from a guitar-amp loudspeaker (nearly touching the grille cloth), and 2 inches (5 cm) from a drum head.

The proposed specification dictates a standard test distance of 2 inches (5 cm). But the manufacturer also can mention test results for other distances that are closer to the product's design center distance. For example, if a microphone is likely to be used'/, inch (0.6 cm) from the mouth, the manufacturer may want to show the response curve at that distance as well as 2 inches (5 cm).

4.2 Sound Source

An artificial mouth is one simulation of a real-world sound source. Some standard artificial mouths are IEEE 269, CCITT 51.1, and ITU-T. The latte is a 3-inch (7.6 cm) diameter cone speaker in a small sealed housing having a defined geometry and reference mic position. It makes nearly a point source below a few hundred Hz [12]. The IEC P.51 standard specifies an artificial mouth.

In the real world, loud instruments are seldom point sources. Guitar -amplifier loudspeakers, cymbals, and drum heads are approximately pistons, not point sources.  So the author proposes that the sound source for this specification be a loudspeaker rather than an artificial mouth.  

Although it is not a point source, a loudspeaker does produce some proximity effect .To illustrate, Figs. 9 and 10 show the proximity effect  of a microphone near a point source and near a loudspeaker, respectively.

 

Fig.9 shows the frequency response of a cardioid microphone 1.5 inches [~4 cm] from an artificial mouth, referenced to a flat-reponse pressure microphone. Fig. 10 is the same, but 1.5 inches from a 12-inch (30.5 c diameter guitar loudspeaker. Results: The piston-like source produces some proximity effect, but less than the point-like source.

Figs. 9 and 10 mentioned above are for a cardioid dynamic microphone with a bass rolloff in its free-field frequency response. Figs. 11 and 12 the same for a different cardioid dynamic microphone. Figs. 13 and 14 the same for a cardioid condenser microphone with a flat low-frequency response in the free field.

As will be shown later, the proximity effect produced by this loudspeake correlates more closely with scale distortion than the proximity effect produced by the artificial mouth. The artificial mouth produces much m proximity effect than real-world sound sources, such as loudspeakers a drum heads. For this reason, the author suggests that the specification include the amplitude/frequency response of the microphone 2 inches cm) from a 12-inch (30.5 cm) diameter loudspeaker, referenced to the amplitude/frequency response of a flat-response pressure microphone the same position.

This sound source is similar to a typical guitar-amplifier loudspeaker or drum head. It is a repeatable sound source that generates a partly spherical wavefront, which in turn may generate proximity effect in the microphone under test.

 

Because a 30 cm diameter loudspeaker does not reproduce high frequencies well, it should be used only for measurements below 1000 Hz.

 Above 1000 Hz, the measurement can be made using a wide-range  loudspeaker at 2 feet (61 cm) to 1 meter, and the two curves can be spliced together.

 

Musical instruments are not recommended as sound sources for this specification. The radiation characteristics of the source affect the response curve because the amount of proximity effect depends on how curved the wavefronts are. For many musical instruments, wavefront curvature varies substantially with frequency [13J. Because musical­instrument sound sources are not repeatable, they are not recommended.

 

4.3 Overlay of Scale Distortion

So far, this specification includes a grid and a frequency response curve taken at 2 inches (5 cm) from a 12-inch (30.5 cm) diameter loudspeaker. On this grid, overlay the difference in equal-loudness contours (scale distortion) between 110 dB SPL and 80 dB SPL. As stated earlier, 110 dB SPL is a typical level of live rock music at a typical listener's position. 80 dB SPL is a typical home listening level for elements within an 85 dB SPL mix.

The Robinson-Dadson contours should be used because they are the current standard. The differences between the contours derived by Moore et al do not agree with experience.

If the frequency response of the microphone approximates the difference in equal-loudness contours between 110 phons and 80 phons, then the microphone has some loudness compensation. In theory, the microphone would be a good choice for recording guitar amps, drums, or other loud sources.

 

The suggested name of the specification is "Loudness Compensation (110 dB SPU80 dB SPL)." The manufacturer has the option of also presenting the loudness compensation for other SPLs that may be encountered in the intended application.

 

LOUDNESS COMPENSATION OF TYPICAL MICROPHONES

Figs. 15, 16, and 17 show the loudness compensation of three commercially available microphones. Each figure is an overlay of the loudness compensation curve (110 dB/80 dB, Robinson-Dadson) and the amplitude/frequency response of the microphone. These are the microphones tested:

Fig. 15. A cardioid condenser microphone with an essentially flat free-field response at 2 feet (61 cm).

Fig. 16. A cardioid dynamic (moving coil) microphone whose frequency response at 2 feet (61 cm) shows a low-end rolloff and a presence peak

Fig. 17. Another cardioid dynamic microphone with a. fairly similar frequency response as used in Fig. 16.

Below 1000 Hz, the amplitude/frequency response of the three microphones was measured 1.5 inches (4 cm) from the center of a 12-inch (30.5 cm) diameter loudspeaker. For high-frequency accuracy above 1000 Hz, the response was measured 2 feet (61 cm) from a coaxial loudspeaker, approximately in the free field.

Figs. 18, 19, and 20 correspond to 15, 16, and 17, but using the Fletcher­Munson equal-loudness contours.

Results: The cardioid microphones with a presence peak show a bass and treble rise in their frequency response, which is the same trend as the scale distortion. However, the match between the two curves is not exact. The response of the flat-response condenser microphone matches the scale distortion fairly well at low frequencies, but not in the upper midrange. The cardioid dynamic microphones exhibit a bass and treble rise, but the bass rise and treble rise occur at higher frequencies than the scale-distortion rises.

 

From these results, the user can get an idea of how well each microphone would reproduce loud instruments at a normal listening level. The more closely matched the curves are, the more tonally accurate the reproduced instrument should sound.

Another way of showing this data is to difference the microphone frequency response from the scale distortion. If the microphone has perfect loudness compensation, the difference curve would be flat. Such a curve could be called "Subjective frequency response for loud sounds," or "Deviation from loudness compensation (110 dB/80 dB SPL)." If the curve is flat, the subjective tonal balance is natural, i.e., the same as the tonal balance heard at the original SPL.

Figs. 21, 22 and 23 show the difference in dB vs. frequency between the microphone frequency response and scale distortion, for each microphone, normalized at 1 kHz. In other words, the figures show the subjective frequency response for loud sounds (three different microphones). None of these curves is flat, but the deviation from flat is not extreme.

 

LOUDNESS COMPENSATION BY MICROPHONE PLACEMENT

Proximity effect is not the only thing that causes bass boost in a microphone signal. Certain microphone placements can boost the bass as well.

Consider an electric-guitar loudspeaker with an open-back cabinet. At a distance from the loudspeaker, the rear wave from the speaker combines in opposite polarity with the front wave from the speaker, canceling low frequencies. Close to the speaker, the front wave is emphasized at the microphone, so that low frequencies are not canceled. Thus there is an apparent bass boost compared to the spectrum recorded at a distance.

To illustrate, Figure 24 shows the spectrum of a guitar amplifier loudspeaker picked up by a pressure (omnidirectional) microphone at 1 meter. It was fed pink noise and equalized flat. Figure 25 shows the spectrum with the microphone placed 2.54 cm (1 inch) from the grille clc at the center of the loudspeaker cone. There is a sizable increase in bass over that picked up at 1 m. This bass boost is not due to microphone proximity effect since an omnidirectional microphone was used to recon the spectra.

Note that most of the bass boost is below 100 Hz, like the 110 dB/80 dB loudness compensation curve. So the author suggests that this tonal effect, due to microphone placement near an open-back loudspeaker, contributes more to loudness compensation than proximity effect. The necessary bass boost is provided by placement in the speaker's nearfield more so than by proximity effect.

Drums produce different spectral balances at different playing levels. T is a physical effect, not a psychoacoustic one. As a tom-tom is struck progressively harder, its spectrum becomes weaker in the lows relative 1 the highs (Fig. 26). So a lot of bass boost is needed in the microphone signal to make a loudly played tom-tom sound "full." Close-miking a tom- tom creates such a bass boost because the microphone is in the bassy nearfield of the drum head, and also because its partially spherical wavefront produces some proximity effect.

Compared to cymbals heard in front, cymbals miked overhead produce spectrum that is weaker in the highs relative to the lows (Fig. 27). So overhead cymbal microphones probably need some high-frequency boo: to sound as naturally bright as cymbals heard out front. This boost can I in the microphone frequency response or in the channel equalization.

 

As shown, microphone placement has a strong effect on the recorded spectral balance, apart from proximity effect. Instrument radiation patterns and the closeness of the mic to parts of the instrument, create a different tone quality in each microphone position (14). Since this effect varies highly with each musical instrument, it cannot be included in the proposed new microphone specification. But it is important to be aware of the effect.

 

7 CONCLUSION

Equal-loudness contours were derived by Fletcher & Munson, Robinson & Dadson, and Moore et al. As these contours show, we hear less bass and treble in a program reproduced at typical listening levels (say, 85 dB SPL) than in the same program heard live at, say, 110 dB SPL. Microphones with proximity effect and a presence peak compensate approximately for these bass and treble losses. That's one reason why these microphones are popular for recording loud instruments such as guitar amps and drums

Some key findings are listed below:

*The equal-loudness contours calculated by Moore et al do not agree with experience. Although they predict that we hear bass more weakly at low SPLs, they also predict that we hear treble stronger at low SPLs-the opposite of experience.

*A suggested typical SPL for loud rock instruments is 110 dB SPL at the listener. A suggested typical SPL for reproduced rock instruments is 80 d8 SPL at the listener. Reasons are given in this paper.

*The difference in the Robinson-Dadson equal-loudness contours (scale distortion) at 110 phons and 80 phons is a broad rise (up to 4 dB) centered at 3 kHz, plus a low-frequency rise below 100 Hz (+7.5 dB at 30 Hz). In theory, that is the equalization (or the microphone frequency response) needed to make a program reproduced at 80 dB SPL have the same balance as the same program heard live at 110 dB SPL.

*A cardioid microphone measured near an artificial voice shows much more proximity effect than the same microphone measured the same distance from a 12" (30.5 cm) loudspeaker. Because loud instruments (guitar amps, drum heads) are shaped more like loudspeakers than like point sources, a loudspeaker is a more realistic sound source for testing loudness compensation for these loud sources.

*Typical cardioid dynamic mics with proximity effect and a presence peak show the same general response trend as scale distortion. However, in the microphone response, proximity effect near a loudspeaker tends to peak at 120 - 150 Hz, not at 30 Hz as in scale distortion. Also, the presence peak tends to be at 5 kHz, not at 3 kHz as in scale distortion.

*Additional loudness compensation (bass boost) is due to microphone placement in the nearfield of the loudspeaker in an open-back guitar-amp cabinet, and in the nearfield of a drum head. In fact, this microphone placement appears to be the main factor in loudness compensation at low frequencies, more so than proximity effect.

Whether or not a new specification comes out of this research, it is helpful to know exactly what is happening to the tonal balance at different SPLs, and how microphones (or mic placement) may compensate for this effect.

 

SUGGESTIONS FOR FURTHER RESEARCH

1. The equal-loudness contours were derived from pure tones, not music.

The difference in tonal balance between loud and soft music may not the same as the difference in tonal balance between loud and soft tones. Experiments could be done in which listeners hear a piece of music alternately at 110 dB SPL and 85 dB SPL, and equalize the 85 dB program to sound like the 110 dB program. That equalization would be the correct loudness compensation.

2. If a spec results from this research, we need to find the best way to present the spec to the user. Some possibilities are:

"An overlay of two curves (mic response and scale distortion)

'A single curve showing their difference (a flat curve indicates correct compensation).

"A number that indicates the correctness of the compensation (For example, Loudness compensation: 8 on a scale of 10.)

In any case, we need to ask whether a typical user can understand and use such a specification.

3. For an open-back guitar amplifier and a drum head, loudness compensation at low frequencies appears to be mostly due to mic placement in the nearfield of the sound source. Perhaps a more realistic

sound source for the proposed spec would be a loudspeaker in an open­back cabinet of specified dimensions.

19

9 ACKNOWLEDGEMENTS

The author thanks the members of the AES working group SC-04-04 wt contributed many helpful ideas to this paper.

10 REFERENCES

[1] L. Beranek, Acoustics, New York: McGraw Hill Book Company,page 401 (1954).

[2] J. Eargle, Sound Recording, New York: Van Nostrand Reinhold Company, pp. 34-35 (1976).

{3) H. Fletcher and W. A. Munson, "Loudness, Its Definition, Measure and Calculation," J. Acoust. Soc. Am., vol. 5, pp. 82-108 (1933).

[4] D. W. Robinson and R. S. Dadson, "A Re-Determination of the Eq Loudness Relations for Pure Tones," Brit. J. Appl. Phys., vol. 7, pp. 1 181 (1956).

[5] B. J. Moore, B. R. Glasberg, and T. Baer, "A Model for the Predictic Thresholds, Loudness, and Partial Loudness," J. Audio Eng. Soc., vc no. 4, 1997 April.

[6] M.G. Scroggie, "Scale Distortion," Wireless World, Sept. 24, 1937 "Scale Distortion-Again," Wireless World, Vol. LIV, No. 11, p. 392, 1948. Thanks to John Woodgate for supplying this reference.

[7] D. Rosmini, Teac Multitrack Primer, TEAC Corporation, p. 11 (1971

[8] Live Sound! International magazine, July/August 1996

[9] D. Huber and R. Runstein, Modern Recording Techniques, Fourth Indianapolis: Sams Publishing, pp. 49-50 (1995).

20

[10] Email from David Josephson to working group SC-04-04, 1 May 1997.

[11 ] H. Tremaine, Audio Cyclopedia, Indianapolis: Howard W. Sams & Co., pp. 248-249 (1969).

[12] Email from David Josephson to working group SC-04-04, 7 May 1997.

[13] Email from John Woodgate to working group SC-04-04, May 1997. [14] B. Bartlett, "Tonal Effects of Close Microphone Placement," Jour. Aud. Eng. Soc. Vol. 29 no. 10 (October 1981), pp. 726-738.