CA1263608A - Microphone actuation control system suitable for teleconference systems - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
A microphone and loudspeaker arrangement for use in a telecon-ference system, wherein a plurality of microphones are held in a fixed relationship to a loudspeaker. The microphones are independently gated ON in response to (1) speech picked up by the microphone, (2) a loudspeaker signal driving the loudspeaker and (3) an electrical signal related to the microphone signals of the other associated microphones. A noise adapting threshold circuit generates a voltage level representative of background noise which is compared with the microphone signal of a respective microphone for determining whether the microphone is receiving speech.
A decisional circuitry monitors the microphone signal of the associated microphone with respect to a MAX bus which carries microphone signals representative of the level of microphone signal at the other microphones.
The decisional circuitry generates a signal indicating that the associated microphone is the first loudest microphone signal.

Description

BACKGROUND OF_'L'IIE INVFNTION
This application i5 a divisional of copendiny Canadian Patent Application Serlal No. ~04,8~2 f:iled on March ~4, 1986 in the name of Shure bro-thers Incorporated.
The present invention relates to an autornatic micro-phone control suited for use in sound reinforcernent, recording, broadcast, teleconference and other applications.
Because of the number of participants involved or the number of locations needing sound pickup, rnultip]e microphones are often used in applications such as churches, courtrooms, broadcasting studios, legislative chambers, and conference rooms, and in particular, in teleconferencing. The outputs of these microphones are usually cornbined in a mixer to feed a sound re-inforcement system, a recording device, or a transmission link to a remote location. If a conventional mixer is used with multiple microphones, the room noise and reverberation pickup is increased as compared to a single microphone; also, the tendency for howl-back is increased, even though typically only one or two micro-phones are receiving useful acoustic input (speech) at any given time. This is discussed extensively in "Direction-Sensitive Gating: A New Approach to Automatic Mixing" by Stephen Julstrom and Thomas Tichy, JAES, ~ol. 32, Nos. 7/8, 1984, July/August.
Automatic mixers have more recently been employed to "gate" ON (pass to the mixer output) only signals from microphones receiving useful acoustic input. The relative effectiveness of mixers is primarily a function of the means used to decide when a ~3~
- la - 613~8-775D

microphone should be yate~ ON. A rnicrophone should gate ON
quickly and independenkly in response ~o valid speech input over a wide dynamic range. Yet it should not respond -to background room noise, nor to -talkers who would be better picked up by another microphone. Additionally, for proper operakion of many teleconference systems, including the system of the aforementioned related application, the room loudspeaker providing audio from the remote location should not trigger microphone yating.
A gating method dependent on a representation of the microphone output level exceeding a fixed threshold level satis-fies none of these criteria. Various prior art references suggest different gating methods.
For example~ in Dugan (United States Patent 3,814,856), the threshold level for each microphone's gating tracks an estimate of the background room noise made from a distant sample taken from a separate noise-sampling microphone or from the averaye of all microphones in the system.
This is done to improve the gating sensitivity while avoiding response to room noise.
In Breeden (United States Patent 3,751,602), a simple microphone is, in essence, gated for use in a speakerphone (a simple teleconference system). Here, the noise re~erence is taken from the single microphone with its level representation processed through a very slow attack, rapid decay circuit. Additional circuitry inhibits microphone gating for loudspeaker sound in most room acoustic environments.
In Maston (United Skates Patent 3,755,625), one and only one of a plurality of microphones is gatecl ON at any time. To gate ON (and thus gate OFF the already ON microphone), a micro-phone's level must exceed a fixed threshold and exceed the level of the already ON microphone by a preselected amount, such as 3 dB, In Kahn (United States Patent 4,099,025), to prevent gating a plurality of microphones for a single source, during the time when a microphone's level exceeds a threshold, triggering of all other microphones is prevented for the duration plus a short additional time roughly corresponding to the transit time for the sound to travel to the farthest microphone in the system.
In Schrader (United States Patent 4,090,032), a pre-selectable, ~ixed threshold is overridden as soon as at least one microphone's level exceeds it and is gated ON. The threshold ~Z63~
-2a-then varies between a high maximum level and approximately the level of the gated ON microphone with the hiyhest level. Gatiny ON of more than a few micxophones simultaneously, even for mul~
tiple sound sources, is stronyly inhibited.
In Anderson et al. ~Uni-ted States Patent 4,489,442, owned by the same entity as the present application), each "microphone" actually consists of an array of typically two unidirectional microphones mounted back-to-back in a common housing whose output levels are compared. When the level of the "front" microphone exceeds the level of the "rear" microphone by a predetermined amount, typically 9.54 dB (indicating the sound source is within an l'acceptance angle"), gating is triggered for the front microphone's signal.
This Anderson arrangement (which also foxms part of the preferred embodiment of the system described in Canadian Patent Application Serial No. 473,243, filed on January 31, 1985 in the name of Stephen D. Julstrom and assigned to Shure Brothers Inc.), has ~Z~3~ 3~

an effective gating threshold which inherently tracks at about 5 dB above the room noise level at the rnicrophone's location. The Anderson arrange~
ment results in direction-sensitive gating which lirnits the number of microphones which gate ON for individual sound sources while not cau.sing the gating of any microphone to inhibit the desired gating of any other microphone for other sound sources. Also, the Anderson arrangement allows positioning a teleconference system loudspeaker in such a way that it will not trigger gating of any microphone and will not significantly inhibit desired gating for local talkers.
However, the operating principle of Anderson requires some care in microphone and loudspeaker placement. Anderson also allows a single sound source to trigger gating in a plurality of microphone~ with over-1QPPjng accept~nce angles. Anderson requires typically two high-quality matched transducer elements for each "microphone" even though only one 15 of the pair is ever heard. Most significantly, Arlderson ean have proper gating inhibited by acoustically renective objects close to the rear OI the microphone or by placement of the microphone too far away from the sound source in relation to the reverberant field of the room, thus preventing the microphone from accurately assessing the direction of the

2~ sound source.
The sound which Q microphone "hears" in a room can be simply described as consisting of two parts: a direct sound which decreases in level 6 dB each time the distance from the source is doubled; and a reverberant field, coming from all directions, which ~tays substantially 25 uniform in level throughout the room as it decays away.
The direction-sensitive gating technique works well unless the micro-phone is so far from the sound source that the reverberant field dominates in the microphone's sound pickup. In larger rooms, this will not occur until the microphone is five feet or more away from a talker. However, at this 30 distance, its pickup would be hollow, "barrelly" and perhaps unintelligible.
In smaller rooms, such as offices and many conference rooms, the reverberant field may dominate at distances of two feet or less, preventing proper gating using the direction-sensitive microphone technique at con-venient taL4er-to-microphone distances. However, in contrast to the larger 35 rooms, in many of ~hese smaller rooms, the sound pickup quality, even if predominantly reverberant, is still intelligible and subjectively acceptable ~3~

due to the quick reverberant decay time. The rnicrophones would be usable if they gated properly.
None of tlle pr~or art cllsclosed in ~he above cited patents fully addresses the goal of ~ttaining ma~imum g~ting serlsitivity in the presence of vArying background room noise, preventing ~ plurality of microphones from gating ON for a single talker, allowing with little mutual inhibition a plurality of talkers to simul~aneously gate ON a plurality of microphones;
and doing all this even when operating in a near totally reverberant (i.e., small room) acoustic environment. Additionally, sound from a telecon-ference system loudspeaker must not gate microphones ON, yet desired gating for simultaneous local speech should not be significQntly inhibited, agflin even in a very reverberant environment.
It is therefore an object of the present invention to provide an automatic microphone gating method which maximizes sensitivity to speech while avoiding sensitivity to varying background room noises.
It is yet another object of the present invention to allow only the single most appropriate microphone in a system to gate ON for a single t~lker.
It is another object of the present invention to allow a plurality of talkers to simultaneou~ly gate ON a plurality of microphones, with minimal mutual inhibition of gating.
It is another object of the present invention to prevent telecon-ference system loudspeaker sound from gating microphones ON, with minimal inhibition of desired microphone gating for locally originating speech.
It is another object of the present invention to enable all the other objects to be met even in near totally reverberant, smaller room acoustic environments.
It is another object of the present inventlon to allow the creation of "dead zones" in a room where sound sources do not trigger any microphone gating.
It is another object of the present invention to provide a variation whereby microphone gating information can be used to control other functions, such as automatic video camera switching.
It is another object of the present invention to link a gating method described herein with an automatic gain adjusting means to maintain 3t;(~

- constant reverberant field pickup flS th~ number of widely sprlced micro-phones gated O;~ vllries above zero.
It is another object of the present invention to provide a variation whereby an automatic gain adjusting means maintains const~nt reverberant S fi~ld pickup as the number of very clo~ely spaced directional microphones gated ON varies above zero.
It is another object of the present invention to use a gating method described herein in the teleconference system of the previously referenced related Julstrom application yielding the benefits described therein.
It is still another object of the present invention to employ such a teleconference system in a combined loudspeaker-microphone arrangement which optimally exploits the characteristics of the gating method, is easy and foolproof to set up, provides improved sound pickup and production through optirnized acoustical and electrical design, works in almost any acoustiGal environment, and is easily expandable.

SUMMARY OF THE INVENTION
These and other ob~ects are achieved in a preferred embodiment of the invention which uses processing circuitry which respond3 to accurate, rapidly varying DC level representations of the frequency equalized output of each microphone and, in a teleconference embodiment uses the elec-trical drive signal to the loudspeaker, similarly equalized.
Each microphone circuit block of the preferred embodiment further processe~ its DC level representation through a very slow attack, immediate decay circuitry, yielding a noise adaptive threshold (NAT). I'his threshold adjusts to the background noise level present during brief pauses in speech, but does not particularly respond to speech waveforms. To satisfy the NAT criterion and potentially initiate g~ting, a microphone's DC level representation must exceed its NAT by an amount, established at 6 dB, just sufficient to guard against ~alse triggerirlg due to random background noise fluctuations.
This enables a high gating sensitivity ~low threshold) to speech, even in the presence of room noise, which is further improved by the use of directional microphones. In the teleconîerence system of the preferred embodiment, the microphones reject room noise pickup typically 5 to 7 dB
a5 compared to direct speech pickup. Thus, direct speech will typically ~263~

gate ON a microphone at about the measured room noise level (as measured by a standard omnidirectionnl noise measurernent device).
The NAT gating criterion al.so works well for microphones located in the predominantly reverberant field of a talker, at least over the range of 5 reverberation buildup and decay r~tes found in acoustically poor, small to medium-sized conference rooms, albeit without the extra sensitivity boost afforded by directional microphones. (The microphones would not be used in the predominantly reverberant field of very large rooms with very long reverberation buildup and decay rates which might cause gating problems, since the sound pickup quality of even a directional microphone used at the implied distance from the source in this environment would be unusable).
To prevent a plurality of microphones from gating ON for a single tallcer, a second criterion (MAX bus) is required to be satisfied in add;tion to the NAT criterion before gating of a microphone i8 triggered. The MAX
15 bus is a single interconnection between the system~s individual microphone circuit blocks which mQintains a varying DC level equal to the maximum of the varying DC level representations of each of the gated ON microphones and the level representations reduced by a fixed amount, established at 6 dB, of each of the gated OFF microphones. To satisfy the MAX bus criterion, a microphone's circuit block must, at least momenterily, be "holding up" (providing the maximum voltage to) the MAX bus.
Only when the MAX bus and NAT criteria are simultaneously satisfied for a microphone will the microphone be gated ON. This simultanaiety results in a trigger signal which is extended by a retrig-S gerable one-shot for ~ "hold time", established at .4 seconds, sufficient to bridge gaps in triggering and pauses between words. A downstream circuit controls the attack and decay rates o~ the audio signal switching to improve the subjective effect of the gating. These trigger extending and switching rate controlling circuits are substantially identical to those disclosed in the previously referenced patent of Anderson et al.
While not necessary in many instances, in the preferred embodiment all microphones are identical and operated at the same relative gains, differing only in their orientation and positioning relative to the various sound sources. A taL~er's speech reaches the microphone closest to him 5 (the one best suited to picking up his voice) before any other microphone and gates it ON. (If the microphones were not identical, were operated at - 7 ~

differing rel~tive gains, or were positioned either poorly or very close together, their relative levels would become rnore importnnt in ad~itiorl to the relative arriv~l times, but even then a microphone well-suited to picking up the talker's voice would gate ON.) The gated ON microphone immediately gain~; Q 6 dl3 advuntage over the gated OFF microphones on the MAX ~us, effectively preventing other microphones from gating ON for that talker. The 6 dB advantage, in combination with the filter time constants and characteristics of the DC
level representation circuits, prevents secondary microphone gating (for 10 that talker) even for impulsive sounds or sounds that terminate abruptly (later at more distflnt microphones~ and for reverberant decays of solmds, which reach all microphones at approximately equal levels. The 6 dB
advantage to the ON microphone insures that the reverberant sound will not trigger gating of an OFF microphone. The filter time constant~, 15 established at 11 milliseconds, are barely slow enough to reliably prevent abruptly terminating sounds from triggering more distant microphones.
Thus, a single talker gates ON only one microphone, ~ut a plurality of taLkers spea}cing normally can reliably gate ON a plurality of microphones, i~ that is what is required to optimally plck up the talkers' voices. The use 20 of precision full wave recti~ication proceeded by attenuation of low frequencies in obtaining the DC level representations of the microphone outputs allows the rapid 11 millisecond îilter time constants. Normal speech patterns have frequent peaks, dips, dropouts, and pauses which are accurately represented in these rapidly varying DC levels. Since a 25 plurality of taLcers cannot synchronize these variations, there are frequent opportunities for each tallcer's closest microphone's circuit to momentarily t'hold up" the MAX bus, resulting in a gating trigger which is extended as described above. The rapid DC level variations from each voice are nor-mally much greater than the differences in average level between 30 individual voices. Thi9 allows a softer talker microphone circuit to hold up the MAX bus to gate ON his microphone even though a louder talker is simult&neously talking ON another microphone. This is not significantly altered by the 6 dB advantage E~iven already ON microphones on the MAX
bus, which is small in proportion to the microphone level vsriations. The 6 35 dB fldvantage does mean, however, that a talker will tend to keep gated ON
an already ON microphone if he is only slightly closer to ~ gated OFF
microphone.

~z~
- ~ -To prevent sourld from a teleconference loudspeaker from gating ON
a microphone, a "speaker inhibit signal" is generateri. The speuker inhibit signE~l is a rapi~ly vurying r~C level representation (substlntially identical in characte~ to those already discussed) of the electrical drive signal to the 5 loudspeaker. This is scaled to represent approximately the worst case ~highest) DC level representation expected from the microphones due to loudspeaker sound.
A speaker inhibit signal such as this could be used in a direct microphone level comparison or applied to the MAX bus to prevent 10 microphone gating from direct loudspeaker to microphone sound. This would have minimal effect on desired microphone gating for simultaneous 1OCQ1 speech (necessery to interrupt the distant talker) due to the rapid time constants involved, as just discussed in connection with multiple microphone gating. However, in most practicai rooms, sufficient energy 15 would be available in the reverberent decay of the loudspeaker sound to trigger microphone gating after the speaker inhibit signal was gone, if applied in this manner. The decay time of the speaker inhibit signal would need to be lengthened to several tenths of a second or more (conceivably much more to allow for particularly poor rooms) to prevent false micro-20 phone gating. This would remove the rapid level variation from the signaland make interruption of an incoming talker by gating ON a microphone very difficult.
Instead, the speaker inhibit signal, with its rapid level variations, is applied in a specific manner to the NAT of each microphone. Each 25 microphone's NAT can be "held up" (with the same meaning as used in connection with the MA~ bus) by either its own action or the speaker inhibit signal, whichever is gre~ter. The speaker inhibit sign~l is scaled appropriately to prevent microphone gating due to the worst case loud-speaker to microphone coupling, and rapid time constants are maintained 30 to minimize inhibition of desired microphone gating for excellent inter ruptability.
The NAT level is not, however, just the maximum of the speaker inhibit and original NAT levels. When a speaker inhibit signal decays down, - the modified NAT level is not what it would have been before the 35 application of the speaker inhibit signal. The modified NAT follows the decaying speaker inhibit level until it reaches the level representation of ~;~63~
- 9 - ~1368-775D

the microphone output, an~ ~rom there follows khe latter le~el in normal N~T action. Tn the absence oE Local speech, t'ne microphone acoust;c input immediately following the decaying away of the speaker inhibit signal is the loudspe~ker .sound's reverberant decay. This does not trigger ~nicrophone ga~ing since the 6 dB
microphone level-to-NAT level criterion is not satiafied, even for uneven reverberant decays. By inherently tracking the actual reverberant decay of a room, the speaker inhibit function done in this manner precludes the need for designed-in-safety margins based on the worst and most unpredictable room acoustics to be encountered, and enables a sensitive local interr~ption capability to be maintained throughout the range of usable room acoustics.
The speaker inhibit function may be beneficially used with a plurality of microphones in conjunction with the NAT and MAX bus, or with a single microphone in conjunction with the NAT without the MAX bus circuitry, as described hereinafter.
While the microphone gating scheme disclosed herein may be used with a variety of additional circuitry to complet~ a tele-conference system, its characteristics are most advantageously exploited when used as part of a system such as disclosed in the previously referenced related ~ulstrom application. This enables the benefits o that system, including total .system gain and feed-back control with sensitive, fast, conversationally-oriented send/receive direction switching, to be achieved in smaller, acoustically poorer rooms and with less restrictive setup require-ments.
In the preferred embodiment disclosed herein, a small ~;~63~
- ga - 6]36~-775D

upward-faclng loudspeaker is comhined in a common low-profile housing, or module, with three outward-fac;ng surface-mounted directional microphones evenly spaced around the circumference.
When used in this manner, the acoustical design can be optimized to yield smooth, wide-range speech pickup and reproduction without the frequency response aberrations commonly encountered when acoustic transducers are used in the vicinity of reflective sur-faces. The loudspeaker distributes its sound uniformly to all conference participants around the module. In combination with the disclosed gating method, the microphones pick up speech substantially uniEormly around the module with a significant reduction in room noise and reverberation pickup as compared to a single omnidirectional microphone. In addition, the directional microphones are placed around the loudspeaker such as to maximalLy reject the direct loudspeaker sound. Even though ~f~6;3~;0~3 - 10 - ~1368-775~

allowance has to be made for strongly reflec-ted loudspeaker sound from closely placed objec-ts, this s-tilL allows some reduction in the speaker inhibit signal level and the feedback con~rolling suppression se-tting (described in the related Julstrom application) as compared to what would otherwise be required with such close transducer spacing.
The loudspeaker-microphone modules can be used singly or "daisy-chained" to give better acoustic coveraye of longer tables or a plurality of tables. Interconnection between modules is made through a single multi-conductor cable which, among other things, links the gated microphone outputs, the MAX bus, and the loudspeaker power amplifier drive signals. The single microphone closest to the talker still gates ON for his speech.
At the user's dis~retion, a module may be "muted", which prevents the audio signals of its microphones from gating ON, but allows its circuitry to act normally in MAX bus in-ter-actions, as if the microphones were gating. This creates a "dead zone" around the muted module where speech and other unwanted sounds (paper rustling, etc.) cannot trigger microphone gating and, therefore, cannot interrupt a teleconference. This muting action can, of course, be applied to single microphones also.
As a variation, applicable to teleconferencing or more general use, instead of spacing microphones apart, a plural-ity of directional microphones may be arrayed very close to each other such that, over a majority of the audio frequency range, sound waves from various directions arrive at the microphones - 10a - 613~8-775D

essentially in the sarne phase. Under this condition, wnen varlous combinations of microphones in the array gate OM, new polar directional patterns are forrned. In essence, -the array forms a single microphone whose polar directional pattern and orientation automatically adjusts to optimize the pickup of the talkers around it.
The invention of this divisional application may be summarized as a microphone system, including: a noise adapting threshold circuitry, said circuitry including:
(i) means for generating an amplitude signal representa tive of the amplitude of a microphone input signal;
(ii) capacitive means for storing charge for generating a threshold voltage level;
(iii) amplifier means responsive to said amplitude signal for charging said capacitive means;
(iv) feedback means for feeding back said threshold voltage to said amplifier for controlling charging of said capac.itive means; and (v) means for developing a signal representative of loudspeaker sound, said developing means applying said signal to said capacitive means and to said input of said amplifier means.
From another aspect~ the invention provides a noisé
adapting threshold circuit, comprising:
(i) means for generating an amplitude signal representa-tive of the amplitude of a microphone input signal;
(ii) capacitive means for storing charge for generating ~Z~36(~
- ]Ob - 6136~3-775D
a threshold voltage level;
(i.ii) amplifier means responsive to said amplitude signal for charging said capacitive means;
(iv) discharge means for discharging said capacitive means at a rate substantially equal to the rate of decrease of said amplitude signal;
(v) scaling means for generating a scaled representation of said amplitude signal; and ~vi) comparison means for comparing said threshold voltage level with said scaled representation of said amplitude signal.
BRIEF DESCRIPTION OF TEE DRAWI~GS
Figure l is a diagrammatic and perspective view of a teleconference system for use with the present invention;
Figure 2 is a top view of a preferred embodiment of a microphone and loudspeaker arrangement of the present inven-tion;
Figure 3 is a side view of the loudspeaker and microphone arrangement of Figure 2;

Fig. 4 is a block diagram of the teleconference system of Fig. l;
Fig. 5 is a block diagram of the Inicl~ophone speaker circuitry of the circuitry of Fig. 4;
Fig. 6 is a block diagram of the microphone gating circuitry of the 5 circuitry of Fig. 5;
Figs. 7-13 are schematic diagrams of the microphone gatinv circuitry of Fig. 6; ~nd ~ ig. 14 is a representation of microphone directional patterns obtained with a variation of the invention.

- 10 DESCRIPTION OF THÆ PREFERRED EMBODIMEMT
Referring to ~ig. 1, an embodiment of a teleconference system is shown. The system includes a microphone/loudspeaker unit 11 approxi-mately one foot in diameter which is positloned in the center of 8 conference table 13. The microphone/loudspeaker unit 11 is connected to a 15 control uni~ 15 of the system by a cable 3S which may pass through holes drilled in the table or may rest on the tsble top.
Control unit 15 is interconnected to a telephone line wall jack l9 by mean~ of conventional telephone wire 31 for conventional "tw~wire"
connections. A telephone 17 is also connected to control unit l~ by 20 telephone wire 33. The user is able to automatically connect either telephone 17 or microphone/loudspeaker unit 11 and associated circuitry in control unit 15 to telephone line wall jack 19.
Although shown connected to a two-wire link (combined send/receive path), the system may also be used with a four-wire link (separate send and \/ 25 receive p~ths) as described in the cited JuLstrom application. The four-wire link may be hard~wired or may be a radio or satellite link, which may include time delay. The radio link may be that of a mobile telephone syst0m, although a more appropriate physical arrangement of the micro-phone(s) and loudspeaker than that to be described hereinafter would be 30 used.
As shown in Figs. 2 and 3, microphone/loudspeaker unit 11 includes three outwardly facing unidirectional (cardioid) microphones 21, 23 and 25 which are equally spaced about the circumference of a base 27 of unit 11.
The base 27 is of a flattened, generally conical shape as shown. A
35 loudspeaker 29 is positioned at the acoustic null of the three microphones :

~Z636(~

21-25 and ~aces upwardly in direction being flushly Inounted in base 27, ~s shown. Base 27 preferably rests on the t~bletop of conference tuble 13.
With such a mounting, microphones 21, 23 and 25 oper~t0 very nenr the surface IRith minimal acoustical interference from the base 27. Also, the loudspeaker 29 oper~tes essentlally as if flush-rnounted in a large, flat surface. As is described in the literature, both microphones and loud-speaker may then have smooth, uniform frequency responses throughout the voice range without the usual deleterious effects of table surface reflections. In addition, the directivity indices of both microphones and loudspeaker sre increased by an amount approaching 3.0 dB (as compared to free space), increasing clarity of sound pickup and production. Hori-zontal dispersion of the loudspeaker sound is achieved primarily through its small size (approximately 2 inch diameter cone), minimizing "beaming".
Each microphone 21-25 reduces undesired sound pickup (room noise, reverberation and loudspeaker sound) by being maintained in a normally O~F condition until needed. A microphone is automatic~lly gated ON only when it is the closest microphone to a speech-like sound source other than the loudspeaker. A microphone will not gate ON for background room noise or for loudspeaker sound.
In the absence of local speech, the microphones are gated OF~ and totally break the acoustic coupling path from loudspeaker to microphone.
No echo of loudspeaker sound will be returned to the far end of the communication link. The microphones will turn ON quickly and reliably in response to speech, without chopping syllables or words. Through their directionality in conjunction with the acoustic arrangement of speaker and microphones as previously described, the microphones minimize acoustic coupling from the loudspeaker and pick up of room noise and reverberation while gated ON.
Several units 11 may be used along the conference table 13. Two or more units 11 may be electrically connected in tandem, as discussed above and further described hereinafter.
Referring to Fi~. 4, microphone/loudspeaker unit 11 houses mike/speaker circuitry 41, while control unit 15 houses control circuitry 43.
Mike/speaker circuitry 41 receives signals from microphones 21-25 and responsiveIy generates gated and non-gated microphone signals along respective conductors 45, 47. The microphone signals are received by .~

~Z63~
- 13 - 6136a-775D

control circui~ry 43 for generating a send .giynal for transmis.sion -to wall lack unit 19. The audio information recei.ved from wall jack 19 is u-tilized by circuitry 43 for generation of a speaker signal along a conductor 49. ~he speaker signal on conductor 49 is received by mike/speaker circuitry 41 for responsively driving loudspeaker 29.
Control circu.i~ry 43 may be constructed utilizing the circuitry of the preferred embodiment described in the cited related Julstrom application. Specifically, substi-tution may be made in Fig. 3~ of the related application for its microphones 11, microphone interface/gating circuitry 353, power amp 321, and loudspeaker 13 by the present application's microphone/loudspeaker unit 11. The gated microphone signal may be applied to the related application's conductor 358 and the non-gated signal to its conductor 1003. The speaker signal may be taken from the output of the related application's mixer/limiter stage 334. As will suggest i.tself, other types of control circuitry can be used as circuitry 43 in order to transmit microphone signals and receive loudspeaker signals along a telephone line or other tele-communi.cation link.
Control circuitry 43, as described in the preferredembodiment of the subject matter of the cited Julstrom application provides suppression (attenuation) of loudspeaker level when local speech interrupts received speech by gating ON a microphone, but only to the degree necessary to reliably maintain feedback stabil-ity. When local speech is interrupted by the far end, the out-~.z636(3~3 - 13a - 61368-775D

going microphone signal will he suppressed and the loudspeaker heard at norma] level. Circuitry of the cited ~ulstrom applica-tion also determines necessary send/receive direction switchi.ng for suppression to occur in an unobtrusive, conversationally oriented manner. When both ends talk simultaneously, priority is given to the interrupting party, maintaining natural interaction.
Either end can always "get through" without yelling.
The combined gated microphone signal from all gated ON
microphones is fed along conductor ~5 to the control circuitry 43 of unit 15. The gated microphone signal may be passed for exam~
ple, through a current control amplifier (as in the cited Julstrom application) prior to entry into a conventional hybrid circuit for placing the microphone signal onto the telephone line via jack 19.
As described in the cited Julstrom application, the control current input to the current controlled amplifier is governed by a send/receive direction decision. Potential feedback loop gain instabil-, ~, ~63~

ities are controlled by inserting suppression in the send pa~h during the receive mode and in the receive path during the send mode.
A number of circuits presently on the market algo may be utilized ~s control circuitry 43 in order to translnit the gated microphone signal to telephone jack 19 for transmission to thc far end of the telecommunication link and to extract the receive signal ~rom the telephone jack 19 for transmission to the loudspeaker 29. The mike/speaker circuitry 41 utili~es a different gating structure than that disclosed in the cited Julstrom ~/ application. This will become evident as circuitry 41 is described here-, 10 inafter.
Referring to Fig. 5, mike/speaker circuitry 41 includes gating circuitry 50 comprised o three like mike gating circuits 51, 53, 56, associated with a respective microphone 21, 23, 25. A conventional power amplifier 59 responds to the speaker signal from control circuitry 43 lS appearing on conductor 49 and generates a similar signal on a conductor 58 of sufficient power level to drive loudspeaker 29. An inhibit signal generator 57 receives the loudspeaker drive signal along conductor 58 and responsively generates an inhibit signal to each gating circuit 51-55. Each gating circuit 51-55 is alike in structure and one will be described with 20 reference to Figs. 6 through 12. A MAX bus conductor 56 interconnects the mike gating circuits, but need not connect to control circuitry 43, as described hereinafter with reference to Figs. 6 and 11.
~eferring to Fig. 6, microphone 21 is electrically connected to a preamplifier/interface circuit 61. After preamplification, the microphone 25 sign~l is output as a nongated microphone signal along conductor 47 and gated throuEFh an opto coupler switch 69 for output as a gated microphone signal along conductor 45.
The remaining circuitry of Fig. 6 controls the gating of opto coupler switch 69 by generation of an opto coupler control signal along a conductor 30 71 to the opto coupler switch. The signal on conductor 71 is devoloped in accordance with (I) the preamplified microphone signal of its associated microphone (appearing on a conductor 73), (2) a signal related to the other microphones (appearing on MAX bus 56) (3) the loudspeaker drive signal (appearing on conductor 68) and (4) a mute logic signal (appearing on 35 conductor 75)-The microphone signal on conductor 73 enters ~ trirn calibrf~tion circuit 81. Circuit 81 is utilized to calibrate its ~ssociated rnike gating circuit Sl (Fig. 5) so that each of the three rnike/gating circuits 51-55 functions identically in its comparison nnalysis, as Nill be understood.
The microphone signal is then fed to a frequency equalization/recti-fication circuit 83 which frequency equali~es the audio signal. Low frequencies, and to a lesser extent high frequencies, are reduced in level relative to mid frequencies. Circuitry 83 also serves to precision full wave rectify the audio signal and filter the resultant. The output of circuitry 83 is a varying DC voltage level signal which carries information of the amplitude and time of occurrence of speech as well as noise in the room as picked up by the associated microphone 21.
The output of circuitry 83 i9 fed to fl noise adapting threshold circuit 85 which gener&tes a threshold voltage level representative of room noise in the vicinity of the micropllone 21. Circuitry 85 generates the threshold voltage level by effectively following the l:)C microphone signal using a very slow attack and immediate decay following. As the DC microphone level signal increases, a capacitor is 910wly charged over a long RC time constant, and then QS the DC microphone level signal is removed, the capacitor is quickly discharged at the same rate as the DC microphone level signal decrease. Because OI the patterns of ordinary speech, the resulting voltage appearing on the capacitor is representative of noise in the room. The noise adaptive threshold voltage will adapt to steady background room noise which does not go below a certain level. Normal speech will not significantly charge the capacitor, which w}ll continually d~scharge to the background noise level during even very brie pauseY in speech~
The DC microphone output of circuitry 83 is also fed to an attenuator circuit 87 where the DC microphone signal is attenuated by 6 dB (a faator of 2). The output of attenu~tor circuit 87 and the output of noise threshold circuit 85 are fed to a voltage comparator 89. Comparator 89 generates an output signal indicating when the rapidly fluctuating speech exceeds by 6 dB the threshold level representative of continuous noise in the room.
Thus, the output of comparator 89 represents an independent decision with respect to one microphone that speech is occurring.

As understood, each rnicrophone gating circuit Sl, 53) 55 tFig~ 5) will make a simila~ decision at its respective cornparator ~9 as speech fro~n a single talker reaches each of the microphones 21, 23, 25. Since it is desired to lilnit the number of rnicrophones gute~ ON for a single sound source, the output of comparator 89 is ANDed at 9I with a second decision signal for determining whether the associated microphone should be gated O~ via opto coupler 69.
The second decisional process determines which of the microphones has received the loudest speech ~irst. The MAX bus 56 receives inputs representative of the other microphone signals for use in the second decisional process. The MAX bus is connected to a decisional circuit 97 where the other microphone signals will be compared to the signal of the associated microphone 21.
The DC microphone SigrUIl from circuitry 83 i~ first attenuated by a 6 dB attenuator 93 prior to input to decisional circuit 97. Attenuator 93, however, is electrically actuflble along a conductor 95 for removing the 6 dB attenuation when the microphone i8 gated ON. The output of the defeatable 6 dB attenuator 93 is coMected to decisional circuitry 97 which compares it to the comparable signals in the other microphone circuits by way of the MAX bus interconnection for determinlng whether it is momentarily the maximum of all such comparable signals. The signal level on MAX bus 56 is controlled by decisional circuitry 97 and the comparable circuits associated with the othet microphones to be equal to such mHximum.
When both the noise adapting threshold criteria is satisfied (i.e., speech is occurring) and the MAX bus criteria is satisfied (i.e., the associated microphone 21 is momentarily receiving the loudest speech as slightly modified by the defeatable 6 dB attenuation of each microphone circuit), an output trigger signal is generated along a conductor 98 for actuating a retriggerable one shot 99. The output of one shot 99 actuates fln opto coupler driver 101 which in turn drives opto coupler switch 69, gating the microphone signal of the associated microphone onto conductor 45. One shot 99 provides a .4 second hold time after each trigger on its input conductor 98. The output of one shot 99 is fed back to the control input of defeatable 6 dB attenuator 93 along conductor 95. Attenuator 93 responds to the HIGH signal from one shot 99, and remove9 the 6 dB
~ attenuation.

3~

The net result of the MAX bus interactionI as described above. is that a single talker gates on only one microphone, but a plurality of talkers speaking normaUy can reliably ~ate ~)N u plurulity of Inicrophon~s.
rhe loudspeaker 29, of course, will present speech signals to each of the microphones 21-25. In order to prevent gatinE~ of the microphone channels by speech from the loudspeaker, a speaker inhibit signal generator 57 is utilized. Generator 57 receives the loudspeaker drive signal 58 for responsively generating a speaker inhibi~ signal along a conductor 105.
C~enerator 57 frequency equalizes~ rectifies, and filters the loudspeaker driving signal for generating a DC output onto conductor 105. The speaker inhibit signal appearing on conductor 105 feeds each of the mike/gating circuits 51-55.
The speaker inhibit sign~l is fed to the noi~e adapting threshold circuit 85 for affecting the noise threshold level in the manner described above and hereinafter with reference to Pig. 10. Microphone gating for loudspeaker sound and its reverberant decay is prevented ~ith minimal inhibition of desired microphone gating for local speeeh.
As will suggest itself, when two or more mic~ophone/speaker units 11 are connected in tandem along the top of a conference table, all MAX
buses are connected together, all loudspeaker power Qmplifier inputs are connected together for being driven by the speaker signal from control unit 15, and all ~ated microphone outputs 45 are connected together as are all nongated outputs 47.
Mute input 75 can be actuated by appropriate logic circuitry to defeat individual microphone gating without altering the MAX bus inter-actions, yielding the results described above.
It will be understood with reference to Figs. 7-13 that all op arnps and comparators are connected to well regulated and filtered, balanced Yoltage supplies of + 15 volts, as is well known in the art. Referring to Fig.
7, microphone 21 is of the electret condenser type. The transducer is interconnected to preamp/interface circuitry 61, including a field effect transistor impedance converter 203 (Sanyo 25K158L), transducer calibration resistor Rl and bias resistors R2, R3, preamplifier components R4, R5, C2, C4 and operational amplifler 207, connected as shown. Resistor Rl is selected to assure uniform sensitivity of each transducer, impedance converter, Rl assembly. Resistor R5 sets the gain of the preamp and is a ~263g~}~ -precision resistor so that Identical acoustical sensitivities frorn each microphone will he presented to the audio mixing buses and to calibration circuit 81. Resis~ors R7 and R8 are also precision resistors.
The microphone signal passes through capacitor C5, photoresistor ~6 (part of opto coupler 69), resistor R7, and onto gated microphone conductor 45. The microphone signal is also passed through capacitor C5 and resistor R8 ontv nongated microphone conductor 47. The microphone signal is also transmitted along a conductor 73 to the trim calibration circuit 81, des-cribed in detail with reference to Fig. 8. The gated ~nd nong~ted buses 45, lD 47 are terminated with resistors to ground, 5.6k, 1.0k respectively, pre-ferentially located in control circuitry 43 so that the values will not ch~nge as a plurality of units 11 are linked. These terminations assure th~t background noise and reverberation pickup remain substantially constant as the number of units 11 used varies and as the number o~ gated ON
Microphones varies above 0. This method was described in Anderson et aL
and discussed in the referenced article.
Referring to Fig. 8, the microphone signal enters trim calibration circuit 81 along conductor 73. Calibration circuit 81 include3 an operational amplifier 211, trim resistcr Rl0, and resistors Rll, R12 connected as shown.
2~ Calibration circuit 81 is used to trim out component tolerance errors to give identical microphone input to MA~ bus gains. In combination with the trimming of resistor Rl and the use of high quality electret transducers, the gating coverage areas of each microphone can be precisely matched and controlled.
Referring to Fig. 9, frequency equalizer/rectifier circuit 83 receives the microphone signal from circuit 8l elong conductor 213. Circuit 83 emphasizes the speech portions of the frequency spectrum and attenuates - - Yery high frequencies somewhat and low frequencies which lie out~ide of the speech band considerably. Also, since there i5 less energy in the high frequency parts of the speech band, for example, "s" sounds, as compared to the energy in the low frequency parts of speech, equalizer/rectifier circuit 83 serves to emphasize the high ~requency portions within the frequency band of speech. OverRll, this greatly reduces the interfering effects of room noise on gating and enables rapid filter time constants to be used.

Circuitry 83 includes an operational amplifier 215, resistors R13-Rl8 and capacitors C7-C10, connected ~s shown. Circuitry ~3 further includes a pair of operationul amplifiers 217, 219 interconnected with diodes Dl-D4, resistors Rl9-R22 and capacitors Cll, C12, connected as shown, for precision full wave rectifying and filtering of the microphone signal.
Attack and decay filter time constants are equal at 11 msec. Circuitry 83 provides accurate level sensing over a wide dynamic range, particularly with speech sounds. An operational amplifier 221 buffers the output of equ~lizer/rectifier circuit 83, appearing on conductor 223.
Referring to Fig. 10, noise adapting threshold circuit 85 i5 illustrated in more detail. The signal appearing on conductor 223 is a linear amplitude representation o~ the frequency equalized microphone signal. This amplitude representation is compared with a voltage appearing on B
capacitor Cl3 which represents room noise. The signal appearing on lS conductor 223 is applied to the noninverting input of ~ field ef~ect transistor input operational amplifler 225 through resistor R88. As the signal at the noninverting input changes, capacitor Cl3 is charged and discharged accordingly. Resistor R27 is of a low value which is u~ed to ~id the stability of operational amplifier 225 and may be neglected in the circuit analysis. Transistor Q102 is interconnected between the output and the inverting input of operational amplifier 225 end is used as a low leakage current diode.
When the noninverting input of op amp 225 is higher in voltage than the capacitor voltage, then op amp 225 holds its inverting input at the same voltage as its noninverting input through transistor Ql02. Resistor R23 charges capacitor Cl3 slowly with a lO second time constant. When the signal on the noninverting input of the operational amplifier attempts to go lower than the voltage on the capacitor, then op amp 225 discharges capacitor Cl3 through diode D5. The operational amplirier pUll~ down at an appropriate rate in accordance with the feedback voltage appearing at the inverting input through resistor n23, which hfls a negligible voltage drop across it due to the low input current of op amp 225 and the low leakage current of Q102. This makes discharge diode D5 appear as a precision diode.
Thus, the noise adaptive threshold voltage on capacitor C13 tracks the voltage on conductor 223 with a slow attack, immediate decay characteristic, seeking the lowest continuous background level.

~63~

The threshold voltage appearing on capacitor C13 is buffered by ~
unity gain field effect transistor input operational amplifier 227. The output of opel ationQl amplifier 227 provides ~ signal representative of room noise.
5~s shown in Fig. 10, 6 dB atter1uutor 87 is constructed b~ resistors R24~ R25 which provide along conductor 229 a 6 dB modification of the input signal appearing on conductor 223. Comparator 83, which is formed from an open collector output LM339, receives the buffered output of the threshold voltage at its inverting input and receives the 6 dB attenuated 10signal at its noninverting input. A resistor R26 provides a small amount OI
hysteresis around comparator 89 for switching stability. Thus, the output of comparator 89 potentially provides a logic high signal along a conductor 98 indiceting that speech is occurring at its associated mlcrophone 21. The voltage level on conductor 98 is also dependent on decisional circuit 97 15(Fig. 6) to be described with reference to ~ig. ll.
Referring to Fig. 11, 6 dB attenuator 93 is constructed by resistors R28, R29 which provide along a conductor 233 a 6 dB attenuation of the input signal appearing on conductor 223. A field effect transistor switch 235 (p-channel, Vp 3 volts~ is connected between resistor R29 and ground 20for effectively removing the 6 dB attenuation when the FET switch 235 is turned off. Turning off of FET 235 effectively removes resistor R29 from the circuit.
A signal appearing on conductor 95 passes through diode D6 for turning FET switch 235 OFF. A resistor R30 is connected between ground 25and the eathode of diode D6 as shown, and a capacitor C14 connects conductor 95 to ground. Resistor R30 ~nd diode D6 help to provide proper control voltages to the gate of FET 235. Capacitor C14 slightly slows the voltage transition on conductor 95 to minimize capacitively coupled noise spikes to the defeatable 6 dB attenuator 93 and other parts of the 30circuitry.
Decisional circuitry 97 includes an op amp 237 having its noninverting input connected to receive the output of attenuator 93 along conductor 233. The inverting input oî op amp 237 is connected to the MAX bus via resistors R31 and R32. Regi8tor R3a is of a low value and aids circuit 35stability. If the noninverting input of op amp 237 attempts to go higher than the arnplitude of the MAX bus, 3iode D7 is forward biased and will J
~'~63~

"hold up" the l~qAX bus to a level equal to the non-inverting input level. If the noninverting input of op amp 237 is lower than the voltage on the MAX
bus, ~iode 1~7 is reverse biased. I)iode D8 prevents excessive negative voltnge excursions at the output of op amp 237. The bias condition of 5 diode D7 is monitored by Q comparator 239 (L~q339), potentially provi :ling a logic high indication at its output along conductor 98 only when diode D7 is forward biased. Resistor R33 provides hysteres~s for stabilizirlg com-parator 239. Resistors R34, R35 are eonnected as shown. At least one resistor R36 should be included to guarantee a load under all conditions for the diode D7 which is holding up the MA~ bus, and thus enable reliable forward bias sensing by comparator 239.
The output of comparator 8g ~Fig. 10) is connected to conductor 98 ~Fig. Il). Comparator 89 and comparator 239 thus serve to provide a hard wired AND (shown symbolic~lly as 91 in Fig. 6) along conductor 98 because of their open collector configuration. Thus, only when both the noise adapting threshold criteria and the MAX bus criteria are satisfied, then retriggerable one shot 99 will be actuated.
Retriggerable one shot 99 includes an open collector output com-parator 241 ~LM339~ interconnected to resistors R37-R41, as shown. Com-parator 241 responds to a logic HIGH input on its inverting input which serves to discharge a capacitor C15. A comparator 243, (LM339) inter-connected to resistors R42-R47, monitors the voltage across capacitor C15 for generating a logic output onto conductor 95. Capacitor Cl5, comparator 243, and associated components provide a .4 second hold time for the output signal appearing on conductor 95. This .4 second time bridges gaps in triggering, as previously described. When triggering stops, cap&citor C15 begins to charge back up, taking .4 second before capacitor ClS reaches a sufficient voltage to change the output of comparator 243.
If triggering recurs before the .4 seconds is finished, the capacitor Cl5 is again discharged and the signal on conductor 95 does not change.
The output of retriggerable one shot 99, which appears on conductor 95 is fed to opto coupler driver lOl, as shown in Fig. 12. DriYer lOI is used to provide a controlled attack and decay time to the opto coupler LED 248 and thus to the resistance change of photoresistor R6 (Fig. 7). The driver is comprised of op amps 247, 249, diodes D9, D10, capacitors Cl6-C18 and resistors R48-R55, connected as shown. The res~llt is that the audio signal ~ z~3~

is gated witll n 4 msec. ~ttack time n11d ~ .3 sec. decny time, yiel~Jing click-free, unobtrusive switching action A mute circuitry cornprised of resistor E~6~ llnd transistor Q11~ serves to prevent the microphone gating signal vithout effecting the gating cont~ol circuitries 85, 87, 89, 93, 97, 99 ~Fig. 6). Transistor Q113 has its collector and ernitter connected Qcross the input to LED driver 101 and ground for shorting the input to ground during muting. ~esistor R48 isolates the shorting signal from the other gate control circuitry. Appro-priate logic circuitry (not shown) may be used to apply a positive voltage to conductor 75, thus turning on Q113 and activating muting. This particular type of muting permits construction of dead zones as previously deseribed.
Referring to Fig. 137 speaker inhibit signal generator 57 (Pig. 6) samples the loudspeaker drive signal along conductor 58. The inhibit generator generates a speaker inhibit signal along its output conductor 105.
Generator 57 serves to frequency equalize, rectiîy and fiiter the lou~
spe~ker driving signal for generating a DC output signal onto conductor 105. The DC output signal is representative of the amplitude level of speech from the loudspeaker. The frequency equalization and filter para-meters are substantially identical to those o~ the similar circuit 83, but the gain is scaled appropriately for the described purpose. Inhibit signal generator 103 is formed from op amps 251, 253, 255, 257, di~des Dll-D14, cap~citors C19-C24 and resistors R56-R65, interconnected as shown.
Referring again to Pig. 10, the speaker inhibit signal appearing along conductor 105 (from Fig. 13) is fed to the non-inverting input of an op amp 259 for modifying the threshold voltage level appearing on c~pacitor C13.
When the voltage level on conductor 105 is at ground (0 volts), or is less than the voltage on capacitor C13, op amp 259 and associated components do not effect NAT circuit 85 operation. Diode D14 prevents excessive negative voltage excursions on the output of op amp 259 whiah would otherwise ocaur due to positive voltage on capacitor C13 feeding back to the inverting input of op amp 259 through buffer op amp 227 and resistor R67. When the voltage level on conductor 105 attempts to exc0ed the voltage on capacitor C13, then op amp 25~ charges capacitor C13 via diode D15 and resistor R68 to maintain the two voltages equal.
The sign&l at the output of op amp 259 is also fed to the noninverting input of op amp 225 via diode D16 in order to override the otherwise ~63~

present input voltage during the tilne when the sp~aker inhibit signal controls the voltage on capacitor C13 to prevent contention ~or the voltage level Qppearing acrosY capacitor C13. The Immediate positive change on the voltage of capQcitor C13, as caused by the speaker inhibit signal, would normaLly cause op amp 225 to pull in the opposite direction through diode 1:)5. By providing the inhibit signal to tlle noninverting input of op amp 225 at a slightly higher voltage level than on eapacitor C13, this problem is avoided. This slight voltage difference is assured by resistor R66 and the higher current level in diode D15 relative to diode D16. As the speaker inhibit signal on conductor 105 decreases from its peak level, capacitor C13 is discharged exactly in step by op amp 225 through diode I)57 controlled by op amp 259 through diode D16.
The voltage on capacitor C13 and the output oî bu~er op amp 227 track exactly the voltage level and rapid attack and decay times of the speaker inhibit signal on conductor 105 until it drops below the microphone level representation on conductor 223, at which point normal NAT action resumes. ~t this time, the microphone signal will probably represent the reverberant decay of the loudspeaker sound, which the voltage on capacitor - C13 will follow, as desired.
The following circuit value~ are given:
Resistors Resistance -R2, R46 8.2K
R3 9.1K
R4, R27 200 R7 5.1K
R8, Rll 11 K
R31, R67, R50, R51, R52, R48100 K
R12, R55, R~0 30 K
30 R13, R56 2.7K
R14, R57 1 K
R15, R16, R20, R22, R47, R58, R59 51 K

~Z63~

R17, R13, R60, R61 510 Rl9, R21, R32, R66, R62, R64 100 R24, R25, R45, R69 10 K
R23, R44 2.2M
R26, R49, R42 1.5!~
R28, R29, E~34, R35, RS8 20 K
R30, R33, R36 1 M
R37, R38 15 K
R39 1.5K

R63, R65 24 K
15 Rl Nominally l.aK

Trimpot 20 R6 Photocell ON
resistance 500 Photocell OF~
resistance 10M

Capa~itors Capacitance C5, C13 4.7 C7, Cg, C10, Clg, C21, C22 .15 C8, C20 .068 30 CU, Cli, C15, C17 .22 ~,z63~W
- ~5 - 61368-775D

Cl4 .Ol Cl6 .1 Cl8 .~)~7 C23, C24 .~7 Microphones may be used in separate enclosures, and directional patterns other than unidirectional may be used. A
speakerphone (simple teleconference s~stem) may be constructed with the control circuitry of the related Julstrom application or other types of control circuitry using just one microphone when this is adequate. The advantage of excellent gatiny performance, even in the presence of incoming speech, would be maintained.
Referring again to Fig. 6, defeatable attenuator 93, decisional circuit 97, AND function 91, and MAX bus 56 would not be needed for a single microphone teleconference system.
Likewise, multiple microphone pickup may be u~ed alone without the use of a teleconference loudspeaker. Referring again to Figs. 5 and 10, power amplifier 59, inhibit signal generator 57, and loudspeaker 29 would, of course, not be needed. Neither would op amp 259, diodes Dl4, 15, 16 and resistors R66, R67, R68, with R68 being replaced by a connection.
It will also be evident that, referring again to Fiy.
11, gating logic signal on conductor 95, for example, may be u~ed to control other related function~, such as overhead loudspeaker muting, automatic video camera switching, or talker indicator lights. An output port 94 is provided to retrieve the logic signal. Accurate microp~one gating coverage areas can be establi-shed with no overlap.

^ .

~Z63~
- 25a - 61368-775D

As a Eurther variation directional microphones may be used in an array spaced very close to each other. ~his would typically be an array of 2 3 or 4 unidirectional ~cardioid) microphones operated at identical gains facing outward in the same plane at equally spaced angles. In this case MAX bus decisions for an individual sound source are based primarily on relative microphone amplitudes rather than relative time-of-arrival of the speech.
The directional pattern in the horizontal plane for an array of three unidirectional microphones is shown in polar co-ordinate form in Fig. 14. Graphs 401 403 and 405 represent by the distance of the line from the center point 411 the relative sensitivities of the three microphones (on a . . ;.~ ~

i3~

linear basis9 not dB) to sound sources coming from any anglc around the array.
Gr~ph 407 represents the direction~l pattern of the combination, or sum, of graphs ~01 und 403. Shnilar graplls rnuy be obtained for the SUMS of the other two pairs. The relMtive sensitivity of graph 407 is scaled to maintain the same total sensitivity as one of the single cardioid rnicro-phones to background room noise and reverberation coming equally from all directions. The directional pattern may be described as a wide-angle cardioid.
Graph 409 represents the sum of graphs 401, 403 and 405, again scaled to maintain the same sensitivity to room noise and reverberation. This directional pattern is omnidirectional. In symmetrical, two or four microphone arrays, opposite pairs of cardioids combining7 or all micro-phones combining will also form an omnidirectional pattern~ In the four microphone array, two or three ad~acent microphones combining will form different wide angle cardioid patterns.
The variety of directional patterns and orientations just described will be obtained using the gating method of the present invention as varying numbers of talkers at various locations around the arrays speak in varying combinations, as understood. One or more of the microphones in an array may be muted to yield direction-sensitive gating sirnilar to that disclosed in Anderson et al. ~U.S. Patent 4,489,442), but relatively less dis-turbed by close reflections and reverberation.
It is generally desirable in an automatic microphone gating system to maintain constant pickup of room noise and reverberation as the number of gated ON microphones varies above zero. This eliminates audible back-ground noise "pumping" and "breathing" effects and maintains constant feedback loop gain, assuming the microphones are in the reverberant field of the loudspeaker. Ordinarily, including in the preferred embodiment teleconference system of the present invention, the microphones are spaced sufîiciently far apart to assume random phase relationships in their response to room noise and reverberation. To maintain constant pickup of these sounds, the gain should be attenuated according to the following formula, which has become standard practice:
Attenuation = 10 logl0 NOM

~Z~3 where attenuation is in dB ~nd NOM is the number of gated 0~ micro-phones. This i9 equal to 3 dB additional attemlation for e~lch doubling of the NOM.
The required atterluation law is different for the closely spaced 5 arrays. To obtain the scalin~ characteristic shown in Pi~. 14 for the three microphone array, requires the relative attenuations:
1 mike OM 0.00 dB
2 mikes ON -5.12 dB
3 mikes ON 8.29 dB
10 To maintain constant reverberant sound pickup in the four microphone array, of which the two microphone array is a subset, requires the relative attenuations:
1 mike ON 0.00 dB
2 opposite mikes ON - 4.77 dB
2 adjacent mikes ON - 5.44 dB
3 mikes ON - 8.45 dB

4 mikes ON -10.79 dB
Referring again to Fig. 7, these attenuation characteristics may be closely Rpproximated using a similar mixing bus arrangement as already employed 20 on gated microphone mixing bus 45. If the bus is terminated with u resistor equal to 4.0 times the sum of the photoresistor R6 ON resistance and resistor R7, then the result is the relative attenuation char~cteristic:
1 mike ON ~ 0.00 dB
2 mikes ON - S.ll dB
3 mikes ON - 8.30 dB
4 mikes ON -10.63 dB
In this case, the bus terminating resistor (not shown~ would be 22K ohms rather than 5.6K ohms.
An array may be used singly or in combination with other micro-30 phones or arrays spaced away. In these instflnces, the bus terminated with the 22K ohm resistor is isolated locally within an array by a bu~fer which feeds the overall system gated microphone mix bu~ through another opto coupler. Associated resistor values are QS originally specified for this bus.
This additional opto coupler is driven on whenever any microphone in its 35 associated array is g~ted on.

.

~.z~36~
- 2~ -The above described option allows what are essentially automaticlllly variable directional charQcteristic, automatically variable oricntlltion microphones to be cre~ted and included in automatic microphone guting system. The option is obviously not limited to unidirectional (cardioid)

5 rnicrophones nor to microphones oriented in the same plane.
It should be understood, of course, that the foregoing description refers to a preferred embodiment of the invention and that modifications or alterations may be made therein without departing from the spirit or scope of the invention as set forth in the appended claims.

:. ., .. ; ~ ~

Claims (2)

1. THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
   PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:
   l. A microphone system, including: a noise adapting threshold circuitry, said circuitry including:
   (i) means for generating an amplitude signal representa-tive of the amplitude of a microphone input signal;
   (ii) capacitive means for storing charge for generating a threshold voltage level;
   (iii) amplifier means responsive to said amplitude signal for charging said capacitive means;
   (iv) feedback means for feeding back said threshold vol-tage to said amplifier for controlling charging of said capacitive means; and (v) means for developing a signal representative of loud-speaker sound, said developing means applying said signal to said capacitive means and to said input of said amplifier means.
2. 2. A noise adapting threshold circuit comprising:
   (i) means for generating an amplitude signal representa-tive of the amplitude of a microphone input signal;
   (ii) capacitive means for storing charge for generating a threshold voltage level;
   (iii) amplifier means responsive to said amplitude signal for charging said capacitive means;
   (iv) discharge means for discharging said capacitive means at a rate substantially equal to the rate of decrease of said amplitude signal;
   
   (v) scaling means for generating a scaled representation of said amplitude signal; and (vi) comparison means for comparing said threshold voltage level with said scaled representation of said amplitude signal.