Hearing Is Not Understanding

 

 Life Safety people are constantly being asked to justify their designs for audio. The wise try to stay safely within the perimeters of the NFPA (National Fire Prevention Association) and cite Code criteria for speaker heights and distances, knowing that outside those palisades lie an evil forest of audio terminology and obscure mathematics. Some get drawn out of cover with subtle appeals to their ego to speak authoritatively on the subject.

 Of course, this is folly, and they suffer the anguish of the damned for their hubris.

 

 What is needed is a simple, reliable and non-subjective means of testing audio systems for intelligibility, one where electronic and acoustic measurements are taken and related to potential system intelligibility. This objective requires an understanding of the way in which humans hear and interpret acoustic speech signals and how sound systems and the acoustic environment effect what we hear and how the chosen objective test signals are affected by the system environment.

 Human speech is a complex auditory stimulus, varying in complexity in functions of both frequency and time. In effect, it is  a modulated waveform - the normal audio range of 125 Hz to 10 kHz is modulated by frequencies from 0.5 Hz to 15 kHz, and no single aspect of the speech wave is essential for speech comprehension. This is prime to understanding speech and its’ potential measurement.

 The way in which speech is processed by the ear-brain mechanism is barely understood. It is clear that the brain treats speech in a different way than other acoustic signals. It is also recognized that the brain treats music differently than speech.

 We can safely state then, that perceptual models related to the way we analyze sound for its’ loudness, duration, rate of change, or spectral and temporal content may not directly apply to speech and even less so to the quality of speech termed Intelligibility. All of these factors affect speech perception to some degree, but the exact way in which each factor synergistically affects the others is a topic of great debate.

 Determining speech intelligibility is akin to hunting down and analyzing a series of clues in a novel. Although it is not always necessary to recognize and understand every single clue, or get the clues in a specific order to reach the right conclusion, enough must be acquired to reach a certain threshold of comprehension. Insufficient auditory clues can lead a listener to the wrong conclusion by misunderstanding or misinterpreting the words or phrase heard. But receiving only limited information can, depending on the circumstances, still lead to deduction of the correct answer.

 The formal analysis of speech intelligibility can be likened to measuring a loss of information. If too much information or certain key elements are lost, then the intelligibility will be affected. Research shows that considerable redundancy is built into normal speech, so it is possible to lose considerable amounts of information before intelligibility is lost. It is this exact combination of potential losses of the highly interactive elements and components of speech that make predicting intelligibility difficult.

 It would seem like an almost impossible task to create an objective measurement system, but in practice surprisingly good results can be achieved. Clear limitations exist, and each of the commonly used methods works only within a narrow window of tolerance. Those limits are constantly overlooked or not understood by some tasked with designing sound systems.

 To understand methods of measuring the potential intelligibility of a sound system, it is important to have a basic knowledge of the main factors affecting intelligibility. From this basis it is then possible to see under which conditions the various methods available may operate effectively.

 Primary factors that affect speech intelligibility:

·         Bandwidth and frequency response.

·         Loudness and the signal-to-noise ratio.

·         Reverberation time and the direct sound field-to-reverberant sound signal ratio.

·         Listener acuity.

·         Talker annunciation and delivery rate.

 Secondary factors:

·         Distortion (THD).

·         Electronic system nonlinearities and compression.

·         Electronic system equalization.

·         Uniformity of sound coverage.

·         Echoes, reflections and the direction of reflections.

·         The direction of the sound source, relative to the listener.

·         The direction of the interfering noise sources.

·         Vocabulary and context of phraseology.

 To accurately predict the intelligibility of speech, the measurement system must not only take each of these factors into account but also assess them in combination, not just as a series of discrete parameters.

 Direct measurement methods

The only truly accurate and direct method of measuring speech intelligibility is to carry out an objectively scored word test. Methods range from the relatively simple modified rhyming test (MRT) to phonemically balance (PB) word tests.

 There are variations in the way each speaker uses a microphone, and it is only the most experienced professional that can control voice inflection and volume changes from minute to minute during the course of articulation of a speech test. The solution is to use a CD that provides repeatable voice consistency and volume. [i]

 Apart from the factors related to the talkers, listeners, and a given sound system, two common factors of all subject-based intelligibility tests have a significant influence on the scores.

The speech material used and the range of materials listeners expect to be presented with during the test, such as the range of possible words (i.e., nouns or verbs).

 Articulation Index

For most sound systems, noise is the primary degrading factor of intelligible speech. One of the first standardized methods established for assessing intelligibility under noisy conditions was the Articulation Index, based on the work of French and Steinberg in 1947 and developed by Kryter, Baranek and others as ANSI Standard S 3.5 (1969).

 The AI concept says that speech intelligibility is proportional to the average difference in decibels between the masking noise level and the long-term average level plus the 12dB of speech signal, taken at either 1/3- or 1/1-octave center frequencies [ii]. The resulting S/N ratios are then weighted and combined to form the Articulation Index. Scores range from 0 to 1; with 0.3 and below rated as Unsatisfactory, from 0.3 to 0.5 as Satisfactory. 0.5 to 0.7 as Good, and greater than 0.7 as very good or Excellent.

 One factor arising from French and Steinberg’s work was the isolation of the relative contributions of each frequency band to intelligibility. The importance of the higher-frequency consonants was realized - the 2 kHz octave band contributes more than 30% to the total score.

 Speech Interference level

This noise-based method uses a simpler procedure than the Articulation Index. The noise level in the 500, 1k, 2k and 4kHz octave bands is measured. The arithmetic mean is taken and related to a table of maximum satisfactory communication distances.

 Percent loss of consonants (%Alcons)

V.M.A. Peutz and W. Klein of Holland first proposed the concept of the percentage loss of consonants in 1971, but a measurement that correlated to %Alcons was not developed until 1986 using the Techron TEF (Time/Energy/Frequency) analyzer.

 The direct-to-reverberant ratio of the sound systems’ transmitted acoustic signal is measured together with the early delay time, which is the first 10dB of the reverberation decay curve.

From these parameters, the TEF computes the %Alcons score based on a set of correlations carried out under the auspices of Syn-Aud-Con [iii] in three different acoustic environments with a total listening panel size of almost 100.

 While the TEF %Alcons method allows the impulse response and the Energy-Time curve to be seen and room reflections evaluated, there are drawbacks. Although the process is semi-automated the algorithm used can be easily fooled, producing misleading results that are not easily identified by inexperienced users. Remember that the %Alcons  measurement is based only upon the 1/3 octave centered at 2 kHz (1.8 kHz to 2.2 kHz) and room behavior at other frequencies is not evaluated.

 While large-format constant-directivity (CD) horns and equivalent devices exhibit a Q-factor and coverage characteristics that don’t vary much with frequency over the upper speech frequency-range, many other speakers do. Measurements at just one frequency produce misleading and overly optimistic results.

 Finally, the %Alcons method does not take into account factors like background noise and the S/N ratio; the frequency spectrum of background noise; the sound system frequency response, bandwidth and equalization; and late, discrete (isolated) reflections and echoes.

 Speech Transmission Index (STI)

The STI family of acoustic measurement is based upon the work of two Dutch researchers, Tammo Houtgast, and Herman Steeneken, who developed and proved the speech transmission index as a measure of speech intelligibility for a number of European languages, (Steeneken, Houtgast, 1983). They found that the reduction in modulation depth and speech intelligibility had a good correlation - again, a measurement of loss of information.

 Although ordinary speech can be used as a test signal, accuracy is better with a specially generated speech-like signal. In its’ analog form, this is seven octave-wide bands of noise from 125 Hz to 8 kHz effectively covering the information-containing frequencies of the human voice.

Each of these bands is amplitude-modulated with 14 sine waves at the low frequencies found in speech. The total number of combinations (7 X 14) forms a matrix of 98 modulation index values. For the full STI calculation, all 98 values are calculated, then averaged for each octave-band and finally weighted and summed to give the STI.

The Rapid Speech Transmission Index (RASTI)  method

The STI dates back to 1971, but did not receive much recognition until Brüel & Kjær (B&K) introduced its’ RASTI meter.

 In 1988 with the PCs then available, the full STI matrix was slow to compute.

It was found that if the reverberation time and system response changed only slightly from octave band to octave band  -  that is, if adjacent octave-bands were found to give similar results, the amount of work could be greatly reduced and the resulting values would still give a viable prediction.

RASTI uses this possibility. It is a coarser sample method than the STI, using only nine combinations - four in the 500 Hz and five in the 2 kHz octave-band are computed.

This short-form is better suited for halls or auditorium environments than in rooms with resonances and echoes. As such, it is not really applicable to amplified systems -  specifically when using significant compression or limiting, because harmonic distortion products falling outside the 500 Hz and 2 kHz octave-bands are ignored in measurement.

 General Calculations for Life Safety Systems

This section gives sensitivity data on a number of audio speakers current to the field and covers eight basic points designers need to handle audio speaker operations in real terms.
An understanding of the use of decibel notation is assumed.

For more information on the subject, the reader is referred to:

Don Davis & Carolyn Davis,

National Environmental Balancing Bureau,

Sound Systems Engineering,

Sound and Vibration in Environmental Systems

2nd Ed., 1994 Howard W. Sams & Co.

1st Ed., 1977 NEBB.

 1.        Rated Speaker Sensitivity (dBA) - reverberant field per U.L. 464

Device

Make / Model

1/8W

1/4W

1/2W

1W

2W

4W

4” Wall speaker

***   ET-1010-R

78

81

84

87

90

93

4” Wall speaker

***   ET-1080-R

78

81

84

87

90

93

4” Wall speaker

***   E-7070-R

---

81

84

87

90

---

4” Wall speaker

***   894B-003

---

---

79

82

85

88

4” Ceiling speaker

***   898B-001

---

---

81

84

87

90

4” Ceiling speaker

***   960B-202

---

82

85

88

91

93

8” Ceiling speaker

***   965A-8R1

---

---

78

81

84

87

 

 

 

 

 

 

 

 

Device

Make / Model

 

0.9W

1.8W

3.8W

7.5W

15W

8” Re-entrant Horn

Atlas  AP-15TU

 

102

105

108

111

114

 

 

 

 

 

 

 

 

2.         Area Ambient Noise Level (dBA)

Area

Low

Ave.

High

Area

Low

Ave.

High

Open Office areas

40

55

 

Subfab-Shell storage

60

62

66

Corridors

40

55

 

Subfab-Exit aisle

70

73

80

Washrooms

40

55

 

Exit aisle-No eqpt.

64

68

73

Cafeterias

45

55

 

Subfab-Very dense

77

80

85

Printing Press plant

40

90

 

Subfab-High density

75

77

82

CUP - Chiller room

 

90

125

Subfab-Low density

65

66

68

CUP - Boiler room

 

92

107

Scrubber trench shell

66

70

74

UPW (Near Motors)

 

89

 

Scrubber trench  #2

73

76

78

UPW (Ctr. aisle)

81

83

87

Scrubber trench  #3

71

76

78

Note: Italics indicate estimated noise levels - not measured, and not considered valid data.

3.         To calculate the expected sound pressure level (SPL) for a given speaker:

            DdB      =    dBdev.  -  dBamb.

where:

dB dev.    is the catalog listed U.L. 464 -rated speaker sensitivity (measured at 10 feet)

dB amb.   is the measured sound pressure level (using the DIN ‘A’-weighted scale)

 4.        Insertion Loss 

People usually calculate speaker loading by just adding up the tap values, but many sound technicians find that they overload amplifiers if they get too close to the amplifier’s maximum rated output (eg. using 235W of a 250W - rated amplifier).

Be aware that there is such a thing on constant voltage (25 v. and 70.7 v.) systems as Insertion Loss. The transformers are not the best quality in the Pro Sound business, and they get very Inductively Reactive at low-end frequencies. To calculate the actual power draw (P actual ) that an amplifier will have to supply:

            P actual      =    P tap ´ 10( I loss/10)

            where

            P actual      is the actual power drawn by the speaker

            P tap          is the expected power value of the speaker tap

            I loss          is the Insertion Loss of the transformer in dB.  (Use a nominal value of 0.6dB)

 eg 4.1:

A 70.7 volt audio circuit has 50 speakers branch-wired to it. They are tapped at 4W each.

For a  250W amplifier, the expected speaker load would be 200 watts - an 80% load factor.

In actuality, insertion losses would make it 230 watts - a 92% load factor. (a 15% discrepancy).

 eg 4.2:

A 70.7 volt audio circuit has 56 speakers branch- wired to it. They are tapped at 4W each.

For a  250W amplifier, the expected speaker load would be 224 watts - an 89.6% load factor.

In actuality, insertion losses would make it 257.6 watts - a 103.4% load.

 5.        Calculating sound pressure level (SPL) any distance from a speaker

            SPLA  = Speaker rating (SPL in dBA)+ 20Log (Rating distance/listener distance)

where

SPLA     is the calculated sound pressure level at listening position ‘A’.

SPLB         is the calculated sound pressure level at listening position ‘B’.



eg 5.1: 

A speaker that is rated at a sound pressure level of 87dB SPL (at 10 feet) is mounted in a ceiling tileseven feet above finished floor. Listeners have a standing hearing height of 6’-0” AFF.

- What is the SPL for listener (A) 1’-0” away? - What is the SPL for listener (B) 23’-0” away?

            SPL for listener ‘A’:              SPLA  = 87 + 20Log (10/1) = 107dB

            SPL for listener ‘B’:              SPLB  = 87 + 20Log (10/23) = 80dB

 

 6.        Combining Decibels

Note that there are two paths from speaker to listener. Note also that the reverberant field is weaker due to distance traveled. The two dissimilar sources combine, but decibels don’t just add, sum, or mean out. They are logarithmic functions, so one needs to convert them into intensity ratios, add them, and reconvert into dB for correct results.

 Summary       

Multiple audio paths are the major problem with good speech intelligibility and message clarity. Further understanding of acoustics is recommended, but the primary indices of intelligibility (RASTI, STI and %ALCONS) all stress the same thing:  If the speech intelligibility factors composing a given environment are already measurably poor, throwing audio power at the situation only makes the problem worse. Power is not a solution to poor acoustic environments - better direct-to-reverberant sound ratios are what is needed. Using lower power taps, enhancing speaker placement, and increasing speaker distribution is the key.

 

A Cleanroom Audio Task

 

 

The Design Requirement:

 

 

 


Design for a voice evacuation system operating clearly and intelligibly in a 360,000 square foot Class-1 fabrication facility cleanroom.

 The client seeks to achieve a cost savings during future cleanroom remodeling activities over the present method of cutting the cleanroom walls and routing conduit for voice evacuation systems during Tool moves.
 In reality, it can be had either cheaper, faster or better  --  pick any two.

 Semiconductor manufacturing requires cleanrooms that offer no quarter to a sound system designer tasked by Fire Code to provide high intelligibility emergency voice evacuation in an H6-rated manufacturing occupancy.

The audio system is required by Code to be UL Listed for Life Safety evacuation applications, and given the priorities, the budget will not allow for elaborate and costly sound reinforcement equipment.

 The Acoustic Environment:

 In a high-efficiency cleanroom, environmental air is routed in through an absolute ceiling of High Efficiency Particulate Air (HEPA) filters and down through a raised, louvered metal grid floor that balances the air for a uniform laminar flow rate of around 90 feet per second. The resulting white noise produces a considerable unwanted sound-masking effect that is usually sought after in open-architecture offices to provide communications privacy.

 Cleanroom aisle and chase walls are prefabricated in designer colors from a honeycomb bonded in an aluminum skin sandwich similar to aircraft wing structures. Cleanroom protocols mandate that nothing that could off-gas or provide a source of free particle contamination be allowed into the environment. The latest protocols allow only one particle below 0.2 micron diameter in one cubic foot of air. In comparison, the air in the pristine Himalayan mountains is rated at 200,000 particles per cubic foot on a good day.

 The Choices:

 The sound system designer is faced with an acoustical environment that has adverse airflow characteristics, almost totally reflective surfaces, a built-in white-noise sound masking system, no possibility for installing speakers in the ceiling, and no allowance for surface projections off the bay walls due to laminar air flow requirements.
The underfloor space is estimated at a value exceeding $2,500 per square foot and not approved for use, since it is reserved for seismically isolated stainless steel and Teflon semiconductor chip manufacturing ‘tools’ costing up to $15 million apiece.

 The market-available audio equipment is generally rated at 5.0% THD (Total Harmonic Distortion) at the unbalanced output of the microphone preamplifier.
 The low-level preamplifier audio line has to be run over a 3,500 foot path using a soul-destroying #16 AWG solid conductor fire-rated cable, because the local AHJ (Authority Having Jurisdiction) interprets the Fire Code as requiring it, which destroys all hope for the designer’s entry into audio engineering heaven.

 The Premises:

 Despite the historical propriety of the fab subfloor space to the factory managers, the argument was tabled and accepted that a cost savings of between $100,000 and $250,000 could be realized over a thirty-six month remodeling scenario by avoiding the labor and material costs associated with relocating the voice evacuation speaker and strobe hardware.
 Due to the focus on audio considerations, only the speaker portion of the system is discussed here.

 The Haas Effect:

 Helmut Haas isolated and analyzed the concept of a single echo and its’ effect on speech, and charted out the results:

 The perceived direction of a sound source was graphed out as the difference between two different sources (in dB) versus time delay (in ms).

The results were that the reflection needed to be about +10 dB above ambient noise before it could overpower the “clamping” effect that a listener performs on the first arriving sound front. This “Haas effect” rises fast for the first 5ms of delay and persists to about 30ms (the Fusion zone) after which it gradually disappears and echoes begin to be perceived (the Transition zone).

 Reflected energy returning to the listener within the following 30ms time period is integrated along with the direct field source, giving the impression of a louder, fuller sound. Successive multiple echoes tend to extend the existing fusion zone already set up by the listener’s brain.

 Localization:

Hans Wallach coined the Precedence Effect from observations that sound from another direction following the first (direct) sound gave the impression that it came from the same direction as the first. This partially explains how the ear can discriminate (localize), or be confused by the direction of a sound source in a reverberant space.[i][i][iv]

 Theory:

Standard acoustic design method focuses upon developing the Direct sound field and eliminating the Reverberant field wherever possible, through wall layout and the use of absorptive materials.

 

 For reasons of cleanroom protocol and expediency of equipment layout, these methods cannot be applied to cleanrooms. However, reasonable quality sound systems can be designed by working with the physics.

 

 With sound traveling at an agreed-upon velocity of 1128 fps at STP (sea-level), we start with a criterion that 40 - 43ms is the worst-case reverberation delay time that we can accept.

 This 43ms is analogous to a 48-foot path of travel for sound reflecting off the sub-floor, honeycomb aluminum laminate walls, process tools, and HEPA ceiling before the power decays below a given intelligibility level of usefulness.

 With speakers located 48-ft apart, the listener would not be more than 24 feet (21.5ms) laterally from any direct sound source.

 Starting with speakers tapped at 7.5W, an initially unquantifiable -6dB loss was factored in to account for theoretical floor grid attenuation and losses due to downward laminar airflow.
 Standard SPL roll-off was also factored, and the direct field power level was calculated to be » 97dB for the listener. Underfloor at the same location it was calculated to be » 103dB, and the collective value of four speakers at that point would be » 109dB.

 Installation:

Installation of speakers was found to be straightforward. Speakers were mounted horizontally in clusters of four on pedestals underfloor, midway between deck and grid and oriented in a (clockwise) NW, NE, SE, SW  cardinal point array on a 48 by 48 foot grid pattern. This created squares across the entire ballroom subfloor, leaving a speaker pointing from each corner into the center of each square.

The cleanroom floor was subdivided into sectors to remain manageable for 1/8-inch scale construction blueprints. On perimeter walls and where sectors adjoined back-to-back, only two speakers were installed on a cluster.

Shielded 16-2 FPL cable was routed up in conduit from the distributed 70.7Vrms power amplifier racks located on the floor below to each cluster junction box. 3/4-inch rain-tight flex conduits were run from each j-box to speaker connectors. EOL devices were installed in the J-box and identified by circuit.

Every cable, junction box and conduit was identified by circuit and raceway ID with a printed, unique-coded, indelible tag.

Speakers were arranged to be epoxy powder coat painted to mitigate off-gassing in a clean environment. Atlas APF-15TU/R folded exponential horn speakers were arranged on fire alarm circuits ten to a circuit (75W max), giving 75% loading on each audio circuit to provide for headroom and future expansion. Only two circuits (150W) were used for each 250 Watt audio amplifier, to provide for headroom and future expansion.

Each speaker circuit was metered for opens and grounds, checked for phasing, then tone-swept for impedance at 400Hz, 1000Hz, and 10kHz respectively and the circuit impedance data recorded and compared to calculated expected values.

Manufacturers’ installation manuals were not precise on audio alignment procedures, leaving much room for debate on set-up procedures. During balancing, a manufacturer’s specified 1kHz 1.0Vrms signal was injected at the amplifier input with amplifier output zeroed, and gain adjustments made for 150 Watt effective power out. That is to say, the gain was adjusted until the voltage on each speaker circuit reached the level required to deliver the full 150W at the speakers as determined from calculations derived from the circuit impedance tests. This correlated to amplifier loading under All-Page conditions. It is noted that subsequent modifications and changes to speaker circuits would necessarily require new measurement of impedance, recalculation and adjustment as with the original procedures.

All data was later copied and delivered to the Owner in the form of Operations and Maintenance manuals prior to system acceptance. Voltage measurements were specifically needed for troubleshooting and to monitor for speaker deterioration during phased and recurring preventative maintenance procedures. The subject of PMI action being afforded during the course of manufacturing operations is not discussed here, and varies on a client-by-client basis.

 Results:

In the open “ballroom” fab environment prior to any walls being installed, some definite echo was perceived from the reverberant field sound pitching off the perimeter walls, doors and windows.

While knowing that speakers were located under the floor on a 48 foot grid interval, localizing on a specific sound source and finding a speaker cluster was an extremely difficult procedure. No sound could be identified as coming directly from a given speaker cluster (direct field sound). Even when looking directly at a cluster it was very difficult to tell if it was actually operating unless confirmed by a sound level meter reading. The sound image had great presence, and appeared to originate from above the HEPA filter ceiling.

When the ballroom was finally sub-divided into smaller manufacturing areas using highly reflective cleanroom aisle and equipment chase walls, the resulting shorter reverberation times appeared to meet calculated sound pressure levels with better-than-expected speech intelligibility.

Actual measurements showed more of a dBA loss than calculated, indicating either that the initial 6dB loss factors for floor grid and laminar airflow needed revision, or that calculations did not account for octave-balanced offset factors when using dB ‘A’ weighting on the meter.

At ear height overall, the measured audio level was a 96 - 98dB. The ambient environmental noise level prior to tool operation was around 65 - 68dB.  After tool qualification (during operations) the ambient sound level was around 70 - 72dB. No TEF data analysis was performed.

The client construction engineers, contractors and fire authorities at the tests indicated that from a purely subjective standpoint, it was the best system that they had heard in any installation to date.

 Summary:

It appears that in spite of all qualified information to the contrary, good paging systems that can be clearly understood by the occupants can be designed for an environment considered absolutely reflective and totally hostile to good speech intelligibility.

The numerous technical disputes waged over a period of twelve months with various client engineers concluded. The professional risk, where all sound engineering texts argued against such a technique, was intimidating. The financial risk was frightening. The design concept was tabled and developed solely on the basis of the math, where no previous audio engineering data for cleanroom environments existed. There is still no comprehensive acoustic analysis being performed in cleanroom environments beyond that developed by cleanroom air systems manufacturers relative to NC criteria.

 Postscript: No microchips were injured during the performance of this project.

 

[i] Suggested audio technology sources:

Syn-Aud-Con Test CD for Sound Reinforcement Systems $45 - Synergistic Audio Concepts,

Syn-Aud-Con Sound System Design Spreadsheet (Excel 5.0 or better) $125,  http://www.synaudcon.com/

Intelligibility and Measurement Test Disc $145 - Prosonus, 111126 Weddington St, N. Hollywood, CA 91601 (800) 999-6191;

JBL Professional Sound System Design Manual (free) ,  http://www.jblpro.com - 559kb *.pdf file

Other more complex TEF electronic test signals sources:

Audio Sciences Corp., PO Box 1189, Eugene, Oregon 97440 (503) 343-9727;

CAVEAT, 770 N. LaSalle St.,Chicago, IL 60610;

E.A.S.I., 1525 Greenleaf, Evanston, IL 60202 (312) 328-2022;

Techron Corp., 1718 Mishawaka Rd., Elkhart, IN 46517 (219) 294-8300;

 

[ii]  ISO Standard 1/1 octave center frequencies are in bold. Others are the 1/3 octave center frequencies.

20, 25, 31.5, 40, 50, 63, 80, 100, 125, 160, 200, 250, 315, 400, 500, 630, 800, 1k, 1.25k, 1.6k, 2k, 2.5k, 3.15k, 4k,

5k, 6.3k, 8k, 10k, 12.5k, 16k, 20k

[iii]  Syn-Aud-Con:  Synergistic Audio Concepts, founded by Don & Carolyn Davis

[iv]  Localization is used extensively when reinforcing weak sound arriving from a stage cluster at the back of an auditorium, by means of overhead fill, giving the effect that the sound is actually coming from off of the stage in front.


Copyright C. Jeppeson  © 1996