US2810787A - Compressed frequency communication system - Google Patents

Description

OGL 22, 1957 M. J. Dl TORO ErAL 3 Sheets-Sheet 1 INVENTORS MICHAEL J. l TORO WALTN GRAHAM `9TANLEY M CHEF/NER BY Oct. 22, 1957 M. J. D1 ToRo ETAL 3 Sheets-Sheet 2 INVENTORS MICHAEL u. 0170/?0 wALToN GRAHAM BYSTANLEY M. scHRElNL-'R Oct. 22, 1957 M. J. D1 TORO ETAL 2,810,787

coMPREssEn FREQUENCY commNIcATIoN sysm Filed May 22. 1952 3 Sheets-Sheet 5 f' l @7159, Z @Lg-5 75 A CLOSED c 74 79. X l .spasm vou/ME oPEN INPUT cow/PRE 5ms vo/.TAGE

0 c @i194 ,ao ,al ,a2 ,ai SPEECH AMR NaN LINEAR BPF Tone 7o INPUT w/TH Avc cKI /0- 7o- GEN. REL/Av FROM s n I 85 PEELHAM 84 93 Mu/ Tl- M/xsk wB/eAToR AMR O "a, FROM `SPEECH AMP I /00 /0/ /OZ MULTI- MIXER m m ZERO D/ODE RELAY VIQRATOR AMR COUNTER CL/PPER CONTROL 4 99 96 fu REcr 97.. HPF PHASE 98 80- SPL/TTER INVENTORS ORNE United States Patent CONIPRESSED FREQUENCY COMMUNlCATION SYSTEM Michael I. Di Toro, Bloomfield, N. J., Walton Graham, New York, N. Y., and Stanley M. Schreiner, Ciifton, N. J., assignors to International Telephone and Telegraph Corporation, a corporation of Maryland Application May 22, 1952, Serial No. 289,344

14 Claims. (Cl. 179-15) This invention relates to a system forreducing the bandwidth requirements of a communication system and more particularly to a system for facilitating the transmission of wide frequency band signals, such as speech, within a relatively narrow frequency band.

The information transmitted by speech does not, at any one time, necessarily require all the frequency space allotted to it by the human voice. The lack of utilization of the entire available spectrum all of the time is obvious when it is observed that any one persons speech is mostly a time sequence of voiced (quasi-periodic) and unvoiced (aperiodic) sounds, each having different and restricted frequency spectra. Voiced sounds, including vowels and consonants like L, M, and N, involved the use of the vocal cords, while unvoiced sounds, such as T, K, and P, are produced in the mouth. In general, the significant energy of the voiced sounds occupies the lower portion of the frequency spectrum and contains a definite fundamental frequency while the significant energy of the unvoiced sounds occupies, almost exclusively, the high portion of the audible frequency spectrum and has no fundamental frequency present.

The presence of two almost distinct bands of audible sounds has led to attempts to communicate over a frequency bandwidth corresponding essentially to the width of one of these bands, instead of the combined width of both. ln the past the voiced and unvoived sounds have both been passed through time invariant (static) filters causing the intelligence of the output speech to'vary for different locations of the filters center frequency. Vi/hen the center frequency of the filter was located in the lower portion of the audio spectrum, the voiced sounds were best understood but the majority of the significant energy of the unvoiced sounds was lost, while when the center frequency of the invariant filter was placed in the upper portion of the spectrum, the unvoiced sounds were best understood and the voiced sounds were mutilated.

It is known that the maximum syllable and/ or wor-d articulation for a 100G cycle transmission bandwidth is obtained with an invariant filter when the filter pass band contains the portion of the original speech spectrum lying between 260 and 1200 C. P. S., where most of the voiced sounds lie. This method `effects frequency compression merely by placing the available transmission frequency bandwidth at the place in the original speech spectrum which results in maximum articulation for time stationary filters. However, this method of frequency compression loses the significant energy present in the majority of unvoiced sounds which lie in the spectrum outside the frequency pass band of such an invariant filter. The optimum 1000 cycle pass band for unvoiced sounds is located above 1.8 kc. but such a pass band would mutilate the voiced sounds. Accordingly, one of the principal objects of this invention is to provide an improved compressed frequency communication system utilizing separate time invariant lters for the voicedl and unvoiced sounds of the original speech input.

'ice

Another object of this invention is to achieve frequency compression of speech by placing the available transmission bandwidth at the place in the speech spectrum which results in maximum articulation for the particular sound input for time stationary ilters.

A further object of this invention is to provide a communication system capable of transmitting simultaneously a plurality of voiced currents without mutual interference and within a relatively narrow frequency bandwidth.

A feature of this invention is the use of a voiced-unvoiced sound detector to control the placing of the available transmission bandwidth at the place in the speech spectrum which results in maximum 'articulation for the particular instantaneous speech input, then to relegate the separated passed frequency bands to the available portion of the transmission bandwidth and simultaneously transmit a frequency shift synchronizing signal to enable the receiver to demodulate the transmitted frequency bands of the original speech input.

Another feature of this invention is the multiplexing of a plurality of speech transmission channels by relegating the filtered speech sounds of each channel to a predetermined portion of the transmission bandwidth and simultaneously transmitting a synchronizing signal for each channel to enable the receiver to demodulate the received speech lcurrents of each channel so that only a relatively narrow frequency transmission bandwidth is required to conduct a plurality of independent speech messages.

The above-mentioned and other features and objects of this invention will become more apparent by reference to the following description taken in conjunction with the accompanying drawings, in which:

Fig. lA is a schematic diagram in block form of a transmitter for use in a three-channel compressed frequency communication system;

Fig. 1B is a schematic diagram in block form of a receiver for use with the transmitter `of Fig. 1A;

Fig. 2 is a schematic diagram in block form of one embodiment of a voiced-unvoiced detector for use with the transmitter of Fig. 1A;

Fig. 3 is a graphic illustration helpful in the explanation of the detector of Fig. 2;

Fig. 4 is a schematic diagram in block form of a second embodiment of a voiced-unvoiced detector for use with the transmitter of Fig. 1A;

Fig. 5 'is a schematic diagram partly in block form of a third embodiment of a voiced-unvoiced detector; and

Fig. 6 is a schematic diagram in block form `of an alternate embodiment of the detector of Fig. 5.

Referring to Fig. 1A, a multichannel compressed frequency communication system in accordance with the principles of this invention is shown comprising three speech channels, designated channel #1, channel #2, and channel #3. The original speech input to channel #l is coupled through a speech amplifier 1 to a voicedunvoiced detector 2 which determines if the instantaneous speech input comprises 4a voiced or unvoiced sound so that the available transmission bandwidth might be allocated to the most advantageous portion of the audio spectrum for the particular input sound. The output of the synchronous relay 3, responsive to the determination of the voiced-unvoiced detector 2, is coupled through delay line 4 to control speech relay 5. When the instantaneous speech input to channel #l is voiced, the output of the synchronous relay 3 causes the speech relay armature 6 to be moved to the upper or voiced position as shown. The output of speech amplifier 1 is coupled to a delay circuit 7 which delays the input signal an amount of time sufcientjto allow the output of the voiced-unvoiced detector 2 and associate circuits to control the Y spectrum'allocated to channel #1.'

' pass filter 8 to mixing amplifier 10.

i 3 Y speech relay 5. When the speech relay armature 6 is in the upper position, the output of the delay circuit 7 is fed directly to band pass filter via line 9. Band pass lfilteri. passes signals having Veti-frequency between .2

and V1.2 lic., vthus allowing the significant 'energy of Vthe. voiced sounds Vto be passed and coupled to Ymixing ampliL Y fier 10 over line 11. When the output of the'synchronous relay 3 causes thespeech relay armature 16 -tobe moved to the upper or voiced position, it also causes a frequency shift oscillator 12 to emit a 13,60 C; 13.753. synchroi signal which is passed through a synchronous band pass Vfilter 13 to eliminate any spurious frequencies fromthe output 'of Vthe oscillator 12.

The Vfiltered synchronous signal output of-bavnd pass filter 13 is also coupled tomi.-

ing amplifier 10Y via line 11. *A n When theinstantaneousspeech input to channel #l comprises unvoiced sounds, the Voutputof the voicedun voiced Vdetector 2 causes the synchronous relay 3 output,

- delayedin circuit 4, to movethe speech relay armature 6 to theirlower or unvoiced position. When the armature 6V is in the lower position,.the output of the speech amplifier 1' is coupled through delay circuit. 7 to a band Vpass filter 14'passing signals between 1.8 Yand 2.8 kc.,

which'is advantageous'for passing the significant energy contained inunvoiced sounds. The frequency filtered energy is coupled to modulator 15. 5

A separate unit 16 at the'transm' ter generates the frequency necessary for modulating the sampled unvoiced sounds into the portion of the transmission Vfrequency The separate unit 16 comprises aY 200 C.-P. S. oscillator 17 whose output is distorted in frequency( multiplier 1S to yield higher har- Y monies.

.The 3 kc. Vharmonicfrom frequency multiplier 1SV is coupledlto modulator where it modulates the filtered 1.8 to 2.8 kc. unvoiced sound to yield a VV1.2 Vto .2 kc. modulated unvoiced sound which is coupled through band Whenrthe output of the voiced-unvoiced detector 2 causes the ou'tputofY synchronous relay 3 to move Vthe armature 6 to the lower unvoiced position; the' output of relay causes'th'e frequency shift synchronous oscillator to Vemit Aa 1290 C. P. S. synchronous signalwhich is filtered in synchronous band pass filter 13 which passers frequencies between 1275 and 1 375 kc. and is coupled to mixing Vamplifier 10. Y Y v Y l t 'Y Thus'the Voutput'rof the mixing amplifier 141, when the instantaneous sound input to channel #l is voiced, corn- Iand 'a 1360 VC. P. S. synchronous signal, but if the instantaneous input to channel #l is'unvoiced, the-.output ofV mixing amplifier 10 comprises the 1.8 to2.8 kc; frequency band of unvoiced sounds modulated to the 1.2 to .2 allo- Y' Vcated frequency band of channel #l and a 1290 S.

synchronous signal'. Y' Y Y Y The speech' inputrto channel,#2 is coupled through speech amplifier 19 to a voiced-unvoic'ed detector 20,-y In a manner similar tothe operation in channel #1,V the output V,of VtheVvoViced-unvoicredV detector 2f) isV fed toV a synchronous relay 21V and through a delay line'22 to operate the speech relay 23. The armature 24 of the speech relay 23V is movedV to the upper position when the instantaneous speech input Vto channel #2 isrvoiced and to the lower.

position when the instantaneous speech input is unvoiced.

' When the instantaneous speech Vinput is` unvoiced and the armaturer24 is inthe lower position, the output o f the speech amplifier 19 is coupled through a delay circuit 25, which imposes a delay equal in time to the time necessaryY for thefspeech relay 23 to a Yband pass VfilterY 26, which passes only'the significant energy lying inthel to 2.8 kc. frequency 'band which isthermtirnum` frequency Vpass bandA forV unvoiced sounds. VThis filtered energy is coupled via line 11 to mixing amplifier 10; Y When the instantaneous speech input is composed of un'- voiced'sounds, the output fsynchronous relay cal'lrsezsV e' i YA2,810,713?

Y prises 'the .2 tor1.2 kc. frequency band of the voiced sound Y the synchronous oscillator 27 to emit a 1465 C. P. VS. synchronous signal which is coupled through band pass filter Y 28, passing frequencies between 1450 and 1550 C. P. S., to mixing amplifier 10.

When the instantaneous speech input to channel #2 comprises voiced sounds and the speech relay varmature 24 is caused to move to the upper position, as shown, Yresponsive to the output'of the voiced-unvoiced detector 20 and associate circuits 21 and 22, Vthe output of the speech ampli-V fier 13 is coupled through delay circuit 25 to a low passy lter 29 which passes the optimum frequencies, below 1.2 kc., for voiced sounds. This frequency `filtered energy from filter 29 is modulated by a 3 hc. 'modulating signal obtained from transmitter unit 16 in modulator 30 whose output ctunprisesV the voiced sound significant energy modulated to the 2'.8'to 1.8` kc. allocated'frequency band ofrchannel #2.. The modulated energy is coupled through armature 26 to the band pass filter 26 which eliminates any spurious frequencies before ncoupling the modulated sampled speech sound Vto mixing amplifier 10.

When' the output ofsynchronous relay 21 causes the armature 24 to move to they upper position, it simultaneously causes the synchronous frequency shift oscillator 27 to emit a 1535 C. P. S. synchronous signal which is filtered in band pass filter28 and' coupled to mixing amplifier 10.

na manner similar'to the operation of channels #l and #2, the instantaneous speech input tor'channel #3 is coupled to a speech amplifier 31 whose output is fed to delay circuit 32 and voiced-unvoiced detector .33. y TheV output of synchronous relay 34, responsive to the deter-` mination ofthe speech characteristic by voiced-unvoi'cedV detector 33, irs coupled through a delay line 35 to operate the speech relay 36. The speech inputclelayed in circuit 32 is coupled through a low'pass filter 37 having an upper y pass frequency of 3.2 kc. to a modulator 38.

t When the instantaneous speech input :to channel"#l3 is voiced, as'dete'rmined by voiced-unvoiced detector'33,

the'armature 39 of speech relay 36 is moved to thefupper'V or voiced position which coupled a 4 kc. signal from fre-Y quencymultiplier 13 tormodulator 33.V The 4 kc.V signalV output ofthe frequency multiplier 1S is obtained bydistortionV of the 200 C. P. S. signal from oscillator 17, as was the 3 kc. signal utilize-drin channels #1 and #2. The 4 kc. signal is'ut'ilized inrcircuit 38 to modulate the speech output of filter 37 and the output is coupled through a 2.8 ,to 3.8 kc. band pass filter 40. Since filter 4t) passes only the 2.8 to 3.8 kc. modulator output, only the portionY 'Y of the original speech input lyingV between .2 and 1.2 kc. i

When the output'of the Vvoiced-unvoiced detector 33'V causes armature 39 to move tothe lower or unvoiced position, a 6'kc. modulatingsignal is` coupled to modulator 38. The -6 kc. signal is obtained by doubling the 3'kc.Y output of frequency multiplier 1S in frequency doublery circuit'43 and coupling its output through armaturev 39 to modulator 38. `YThe modulated speech output of modu- Y lator 38 is coupled to band pass .filter'4ti passing fre-V quencie's between V2.8 and-3.8 kc. The 2.8 to 3.8 kc. out-V put of modulator 3S is equal 'to the' frequencies of theV speech input lying Vbetween' 2.2 and-3-2 kc., which is anY advantageous band Vfor the significant energy of unvoiced' sounds, modulated by the 6V kc. signal from frequency Y doubler 43.

When the output of the synchronous relay/V34 causesY the armature 39 to' be moved-to the lower position, it also causes the synchronousY frequency shift oscillator 41 tov emit a 1640 C. P. S. synchronous signal Which's filtered by band pass filter 42 and fed to miXing amplifier 10.

Thus the output of mixing amplifier 10 comprises either the .2 to 1.2 kc. frequency band of voiced sound input to channel #l and a 1360 C. P. S. synchronous signal or the 1.8 to 2.8 kc. frequency band of unvoiced sound input to channel #l relegated to the .2 to 1.2 kc. transmission frequency band allocation of channel #l and a 1290 C. P. S. synchronous signal; and either the 1.8 to 2.8 kc. frequency band or" unvoiced sound input to channel #2 and a 1465 C. P. S. synchronous signal or the .2 to 1.2 lic. frequency band of voiced sound input to channel #2 relegated to the 1.8 to 2.8 kc. transmission frequency band allocation of channel #2 and a 1535 C. P. S. synchronous signal; and either the .2 to 1.2 kc. frequency band of voiced sound input to channel #3 modulated to the 2.8 to 3.8 kc. transmission frequency band allocation of channel #3 and a 1710 C. P. S. synchronous signal or the 2.2 to 3.2 kc. frequency band of unvoiced sound input to channel #3 modulated to the 2.8 to 3.8 kc. transmission frequency band allocation of channel #3 and a 1640 C. P. S. synchronous signal. The 200 C. P. S. basic modulating signal from oscillator 17 is also coupled to mixing amplifier l0. The output from amplifier 10 is coupled to the transmitter #i4 which transmits all three speech channel currents and synchronizing signals within the frequency band of .2 to 3.8 kc.

The transmitted signals are detected in the usual receiver circuitry 45. Band pass filter 46 passes all received signals lying in the frequency band between .2 to 1.2 kc. which contains the transmitted speech currents of channel The synchronous band pass filter 47 passes the synchronous signals of channel #1, either 1360 or 1290 C. P. S., to discriminator 48 whose output controls the movement of the armature 49 of speech relay 50. When the synchronous signal has a frequency of 1360 C. P. S., it indicates a voiced sound is being received in channel #l and the armaturei is moved to the upper or voiced position and couples the output of filter 46 to variable resistor Si. When the synchronous signal has a frequency of 1290 C. P. S., it indicates an unvoiced sound is being received in channel #l and armature 49 is moved to the lower or unvoiced position coupling the output of filter 46 to modulator 52.

rThe 200 C. P. S. basic modulating signal utilized in the transmitter to obtain the modulation signals for each channel is detected in receiver 45 and passed by filter 53 to a frequency multiplier circuit 54 which distorts the filtered basic modulating signal to provide the 3 kc., 4 kc., and 6 kc. harmonics in its output. This procedure provides demodulating frequencies in the receiver from the basic modulating signal and allows the same frequency to be maintained at the modulator and demodulator for each channel in spite of slight drifts in the 200 C. P. S. oscillator 17 of the transmitter.

The 3 kc. frequency output of multiplier 54 is fed t0 modulator 52 Where it is combined with the .2 to 1.2 kc. received signal to demodulate the signal to its original position in the frequency spectrum, e. g. 1.8 to 2.8 kc. rThe demodulator unvoiced sound energy is coupled to band pass filter S to remove any spurious responses due to the demodulating step and the ltered output is coupled to variable resistor 5ft. The output of resistor 51 is coupled through tap 56 to an amplifier 57 whose output is equal to the reconstructed speech input to channel #l of the receiver. The tap 56 of variable resistor 51 is adjusted so that the output of receiver channel itl contains a pleasing balance of voiced to unvoiced sounds.

in a similar manner the 1.8 to 2.8 kc. signals representing the transmitted speech input of channel #2 are filtered by speech band pass filter 58 of receiver channel #2. The synchronous signals of channel #2 are passed by the synchronous band pass filter 59 and when the received synchronous signal is 1535 C. P. S., the output of discriminator 60 causes the speech relay armature 61 to be moved to the voiced sound position where the output of the speech band pass filter 58 is coupled to the modulator 62. The output of the speech band pass filter 58 is beat with the 3 kc. modulating signal obtained from the frequency multiplier 54 and the .2 to 1.2 kc. output of modulator 62 is passed through a low pass filter 63 to the variable resistor 64 of channel #2. When the synchronous band pass filter 59 passes a synchronous signal of 1465 C. P. S., the output of the discriminator 60 causes the speech relay armature 6l to be moved to the lower position where the 1.8 to 2.8 kc. signal is coupled directly to the variable resistor 64 of channel #2. The balanced voiced-unvoiced output of the variable resistor 64 is coupled through tap 65 to amplifier 66 to the output circuitry and comprises the reconstructed original speech input to channel #2.

T he received signals lying in the 2.8 to 3.8 kc. frequency band of channel #3 are filtered by the speech band pass filter 67 and passed to the modulator 68 of channel #3. When the synchronous band pass lter 69 passes a 1710 synchronous signal, the output of the discriminator 70 causes the speech relay armature 71 to be moved to the upper position which couples the 4 kc. output of the frequency multiplier S4 to modulator 68 where it is beat with the 2.8 to 3.8 kc. output of the speech band pass filter 67. The output of modulator 68 for the voiced position of the speech relay armature 7i comprises a .2 to 1.2 kc. dernodulated output which is passed to the low pass filter 72 and coupled to amplifier 73 and thence to the speech output of channel #3. When the synchronous signal for channel #3 comprises a 1640 cycle signal, the output of the discriminator 70 causes the speech relay armature '7i to be moved to the lower or unvoiced position where the 6 kc. output of the frequency multiplier 54 is coupled to the modulator 68 to be beat against the 2.8 to 3.8 kc. output of the speech band pass filter 67. The demodulated 2.2 to 3.2 kc. signal from the modulator 68 is passed through the low pass filter 72 to the amplifier '73. The output of the amplifier 73 is the reconstructed speech input to channel #3 of the transmitter. Thus the output of the receiver circuitry comprises the reconstructed separated speech input to each of the three audio channels of the transmitter.

One type of detector, for use in the communication system of this invention, for determining whether the instantaneous speech input to any channel is voiced or unvoiced, is shown in Fig. 2. The amplified instantaneous speech input of the channel is coupled through a volume compressor 74 to two filters 75 and 76, one band pass at 360 to 900 C. P. S. and the other high pass above 3500 C. P. S. The output of the low pass filter 75 is rectified negatively in rectifier 77 while the output of the high pass ii'ter is rectified positively in rectier 78. The outputs of the rectifiers are added and the sum controls a relay 79.

When the instantaneous speech input is voiced, the low frequency filter '75 will have a much greater output than the high frequency filter 7 6. Since the former is rectified to give a negative voltage and the latter to give a positive voltage, the sum of the outputs of the rectifiers 77 and 78 will be a large negative value which may be used to bias a relay control tube to cutoff and thus open the relay 79. On the other hand, when the sound is unvoiced, the sum of the two rectified voltages will be a large positive value which may be used to drive a relay control tube to saturation thus closing the relay 79.

However, there are some marginal sounds for which the output of the two filters will be approximately equal. The bias voltage fed to the grid of a relay control tube would then fluctuate about some intermediate value causing the relay 79 to chatter between its open and close positions. In order to mitigate this undesirable action, the audio voltage is first fed to a volume compressor 74. The volume compressor 74 is similar to a limitor circuit except that it has a large time constant so that the out- Vin Fig. l is illustrated in Fig.

theV volume compressor74 reduces the amplitude ofthe t fluctuations thereby reducing the tendency ofthe relay 79 to chatter. t Y Y v The chattering of Vrelay 79 canl be further'rednced by means of a property common to all relays, that is the current which just causes-a cle-energized relay to energize is not equal toL .e currentwhich just causes the energized relay to de-energize. By proper design of the relay armature and contacts, this difference can be accentuated so that a considerable current change is required to first open Y andthen close the'relay or vice versa, although the change i needed'to do only'one of these functions may be quite small.

Referring toV Fig. 3, wherein this relay effect is graphicallyV illustrated, the x axis' representing the magnitude of the control bias and the y axis representing relay closed position Vabove the x axis and relay open position below the x axis. AFor a control bias A more negative than point B, the relay will always be open and this open condition will continue as the biasris made less negative until point B is reached. `Bor a bias between points B and C, therelay will still stay open if it is originally open, that is if the region BC was approached from the left. At the bias C the relay closesrand will stay closed for all biasV voltages to theY right of point C. lf the'bias voltagev is now made less positive, the relay will stay closed until the bias B is again reached. Thus, in order to make the relay operate'through a complete cycle, it will require a voltage having a magnitude BC which may be made large to decrease the chattering of the relay.

Another type of voiced-unvoiced detector which may be used with the communication system of this invention is shown in Fig. 4, wherein the instantaneous speech input is coupled to anamplier 89 whose output is fed toa non-linear device 81. The instantaneous audio energy is purposely distorted by some non-linear circuit 81, such as a crystal, in order to emphasize the distinction between voicedV and unvoiced sound characteristics. Voiced sounds being periodic will have a definite fundamental frequency generally above 80 cycles per second. Distortion' of such a sound introduces a D.C. component Valong with components at multiples of the fundamental frequency, but there will be no energy at a frequency between the DQ-C. component and the lfundamental frequency.` Unvoiced sounds being aperiodic when distorted will have a frequency spectrum which is continuous rather than a collection of discrete frequencies as will a distorted voiced sound. Distortion of the unvoiced sound introduces both a D.-C. component and continuous frequency values from D.C. up to high frequencies.

The output of the distortion circuit 83. is fed to a band pass filter 82 which passes the energy contained betweenY l0 and 70 C. P. S. When an unvoiced sound is distorted, energy will be present in this band of frequencies, but a distorted voiced sound, generally having a fundamental frequency above 80 C. P. S., will not have any energy present in this filtered band. This energy output or lack of output may be used to control the presence or absence of a steady tone from a tone generator S3. The tone generator 33 output is used to control the synchronous relay of the communication system shown in Fig. 1.

Still'another type of detector applying the distinctive characteristicsV of voiced and unvoiced sounds to operate the synchronous relay of the communication system shown 5. It is obvious that'theV voiced'sounds having the majority of their signicant energy/'inthe lower frequency band will have a greater Y biguity. Here the D.C. output of 5'5`V Vputs which are Vfed A the gridof a relay control tube.

number orfz'rs thanunvoieedv sounds which lie in the upperl portion'ofthe speech frequency spectrum. In theV voieeduln'viiicedVK detector of Fig. V5, the instantaneous speech o'utput'of thelsp'eech amplifier of each channel is fed to aniixer-aiiiplier 84 which continuously adds a 3 licfsignal of vconstant amplitude obtained from the multivibrator 95 to the speech signal. The gain'control of the mixer-amplier isso adjusted that a minimum amphtude signal'required to operate the synchronous relay is large enough to reduce ther-number of zeros considerably. "the output of the mixer-amplifier 84 is limited in circuitV Se Vuntil the output of adifferentiation circuit 87 will produce; one pulse each timeY thelimited speech waveform crossesthe'zero axis; Each of the positive pulse outputs of the dillierentiation circuit 87'lires agas triode 88 causing acondenser S9, fromV itsplate'to ground, to discharge into afsecond larger condensergll in the cathode circuit of the gas'trlode 8S. The time constant of the condenserV 89 from plate togr'ound andgthe plate load resistor 91 f through which-it is lrecharged is small compared Vto the maximum rate at which'zeros mayjoccur, so thattheV charge added to the condenser 9i) in the cathode per pulse is independentfofthe spacing of the zeros of the speech waveform. The averace voltage acrossthe cathode condenser 'ftl'is then proportional to the number ofzeros per Y second, and *if'the'averagefvoltage output is high enough when clipped inldiode 92, it will bias the grid voltage of a relay control tube 93Ato the point where it will Vclose the synchronous relay. This'pull in voltage is controlled by the 'adjustable' biastresistor 94 inthe Vgrid circuit of theV relay control tube. The number of zeros per second required to obtain this voltage is controlled by the rheostat 95 of the cathodecircuitiin the, gas tube S3. The diode clipper 92 in the grid circuitof the relay control tube 93 places an' upper limit onthe voltage developed by the zero counter-circuit. The purpose of this diode clipper 92 is to minimize the time required for therelay to de-energize" when the input comprises a voiced Vsound following an. t unvoicedsound or no speech sound input.

Referring to' Figi 6,'an improved version' of the voicedunvoiced detector o'f Fig. 5 is shown, whereinthe ,At-CnV output of the zero' counter is used in conjunction withthe D.C. output to 'control the bias of relay control tube`93. The detector of Fig. 5 obtains a D.C. voltage output from the zero counter proportional to the numberof zcrosper unitrtime in the' sound input.V While the majorityfof speech soundsare characterized either voiced or unvoiced with a low andra higli number of zeros, respectively,- 'thereV Y are some marginal sounds which Yare intermediate. These 50 intermediate sounds of the zero'type`4 detector of Fig. 5 do not categorize well. Interpretingthem sometimes as voiced 'and sometimes as unvoiced soundsV causes relay chattering. The detectorV of'Fig. 6*-redu'ces this amvthe -zero counter is only part of the measure of Whether a soundY isV voiced or unvoiced. The A.C. Voutput is utilized by passing it through a low pass lter`96'and then a high pass lter 97, which inconjunction constitute a band pass lilter from approximately 80 to 180 cycles. The filtered output is. coupled to a phase splitter 98 giving two push-pull out-V to a full wave rectier 99, the output of which is anegative lD.C. voltage proportional to the A.C. voltage amplitude output of the zero counter. Pushpull output is utilized because theripple components -in the rectier 99 output are'most easily .ltered The D.-C. negative voltage output of the rectifier 99 isV then added to the original positive voltage at theV zero counter output and the sum isV used to control the'bias on Y This sum is amore' accurate indication of whether a sound is voiced or unvoiced. Y v Y While We haveV described above the principles of our invention in connection with specilic' apparatus, it is to im be clearlyy understood that'this' description is-made only v by Way of example and notY as a limitation to the scope of our invention as set forth in the objects thereof and in the accompanying claims.

We claim:

l. A communication system for the transmission, within a narrow frequency band, of speech signals having a wide frequency bandwidth comprising means to detect in said speech signals the voiced and unvoiced sounds, means to generate an identifying signal of predetermined regular characteristics different for each of said sounds responsive to the output of said detecting means, means to separate the speech signals according to their frequencies into narrow sub-bands of said frequencies different for each of said sounds, means for shifting at least one of said sub-bands of frequencies to bring said sub-bands to the same portion of the frequency spectrum to form a narrow audio frequency band signal retaining the speech character of each of said sub-bands, means for transmitting said narrow audio frequency band signals, without destroying their speech character, together with said identifying signals, a receiver to detect said transmitted signals, and means for transposing said received narrow sub-bands of frequencies to their original frequency allocation in the speech signal spectrum responsive to said identifying signals.

, 2. A communication system for the transmission, within a narrow frequency band, of speech signals having a wide frequency bandwidth comprising means for detecting the voiced and unvoiced component sounds of said speech signals, switching means responsive to the output of said detecting means, means to form a different narrow sub-band of frequencies from each of said speech signal components responsive to said switching means, means to generate an identifying signal responsive to the output of'said detector means, means for relegating said subbands to the same portion of the frequency spectrum to form a narrow frequency band signal, means for transmitting said narrow frequency band signals and said identifying signal, a receiver to detect said transmitted signals, and means for transposing said received narrow sub-bands of frequency to their original frequency allocation in the speech signal spectrum responsive to said identifying signals.

3. A system according to claim 2, wherein said means to form a narrow sub-band of frequencies from each of said speech signal components further includes means to delay the input speech sig-nal to said forming means to allow time for the operation of said switching means responsive to said detector.

4. A system according to claim 2, wherein said means for detecting the voiced and unvoiced components of said speech signals comprises ltering means for the low frequency energy of said input speech signals, means to rectify said filtered low frequency energy, filtering means for said high frequency energy, means to rectify said filtered high frequency to a polarity opposite of said rectified low frequency energy, means to add said rectified energy whereby the polarity of the sum of said rectified energy is indicative of the input speech signal component, and means to reduce the amplitude of said input signal to the two said filtering means inversely to the original amplitude of said input signal.

5. A system according to claim 2, wherein said means for detecting the voiced and unvoiced components of said speech signal comprises means to distort said speech signal input and ltering means to pass a predetermined frequency band less than the fundamental frequency of a voiced sound whereby an output from said filtering means will be indicative of an unvoiced sound component input while lack of an output from said filtering means will be indicative of a voiced sound component input.

6. A communication system for the transmission, within a narrow frequency band, of speech signals having a wide frequency bandwidth comprising means for detecting the voiced and unvoiced components of said speech signal including means to limit the input signal, means to differentiate the output of said limitor means, means to generate a signal responsive to the number of times said differentiator output Waveform crosses the zero amplitude axis whereby the amplitude of said generated signal will be indicative of the characteristics of the speech signal component input, means to generate a relatively high frequency signal, means to mix said generated signal and said speech input, and means to couple the output of said mixing means to said limiting means, switching means responsive to the output of said detecting means, means to form a different narrow sub-band of frequencies from each of said speech signal components responsive to said switching means, means to generate an identifying signal responsive to the output of said detector means, means for relegating said sub-bands to the same portion of the frequency spectrum to form a narrow frequency band signal, means for transmitting said narrow frequency band signals and said identifying signal, a receiver to detect said transmitted signals, and means for transposing said received narrow sub-bands of frequency to their original frequency allocation in the speech signal spectrum responsive to said identifying signals.

7. A communication system for the transmission, within a relatively narrow frequency band, of a plurality of speech signals each having a wide frequency bandwidth comprising means to separate each of said speech signals into its component voiced and unvoiced sounds, means to form different narrow sub-bands of frequencies from each of said component sounds, means for relegating the sub-bands of each speech signal to the same position in the frequency spectrum to form a plurality of narrow frequency band signals, means for relegating said narrow frequency band signals to contiguous positions in the frequency spectrum, means to generate a plurality of identifying signals responsive to the output of said separating means, means to transmit said narrow frequency band signals and said identifying signals, a receiver to detect said transmitted signals, means to separate said narrow band signals, and means responsive to said detected identifying signals to restore the sub-bands of each narrow band signal to its original position in the frequency spectrum.

8. A communication system for transmission, within a relatively narrow frequency band, of a plurality of speech signals each having a wide frequency bandwith comprising means to separate each of said speech signals into its voiced and unvoiced sound components, filtering means responsive to said separating means to separate the significant energy frequencies of each of said components, means to generate a basic modulating frequency, means to distort said basic modulating frequency to provide a plurality of modulating signals, means to modulate at least certain of said significant energy frequencies to transpose said certain significant energy frequencies of each speech signal to the same portion of the frequency spectrum, means to generate a plurality of identifying signals responsive to the output of said separating means, means to transmit said basic modulating frequency and said transposed components and said identifying signals, a receiver to detect said transmitted signals, means to distort said transmitted basic modulating frequency to provide a plurality of demodulating signals having the same frequencies as the said modulating signals, means to separate the transmitted transposed components of each speech signal, and means responsive to said detected identifying signals to restore said transposed components to their original position in the frequency spectrum.

9. A system for transmitting, within a narrow frequency band, speech signals having a wide frequency bandwidth comprising means to detect in said speech signals the voiced and unvoiced sounds, means responsive to the output of said detecting means to generate for each of said sounds a different identifying signal of predetermined regular characteristics, means to separate the speech fri signals according to their frequencies into.Y narrowsub-Y bands of said frequencies dilferent for each of said sounds, means for shifting at least one ofsaid sub-bands of frequencies to bring said sub-bands to the same portion 'of the frequency spectrum to form a narrow audio frequency band signal retaining the speech character of each of s'aid/ sub-bands, and means for transmitting said narrow audio frequency band signals, without destroying their speech character, together with said identifying signals.

' 10. A system for transmitting within a relatively narrow frequency band a plurality of speech signals each having a'wide frequency'bandwidth comprising means to separate each of said speech signals into its voiced and unvoiced sound components, .filtering means responsive to said separating means to separate thek significant energy frequencies of each of said' componentsmeans to generate aY means to transmit said 'basic modulating frequency and` .said transposed components and'said identifying'signals.

11. A device to separate the high frequency and 'lowV frequency components of an input signal'comprising means to. limit the input signal energy, means toY differentiate said vlimited output, VmeanstoV generate a Vsignal Vhaving'an amplitude proportional to the number of Vtimes said differentiator output Ywaveform crosses the zero arnplitude axis, and switching means responsive to said geni erated signal, for effecting separation of said high and low frequency. Y

12. A device according to claim 1l, which further in-V cludes means to generate a constant amplitude signal and L?. Y means. toy mix said generated high frequencyV signal and said input signal whereby the distinct characteristics. of the high frequency and low frequency components of Y said input signal are emphasized. Y 5 13. A device to separate the high frequency and low frequency components of an input signal comprising means tolimit theinput signal energy, means to differ-V entiate said limited output, means to. generate a signal having an amplitude proportional to the number of times said dilferentiator output waveform crosses the zero amplitude axis, including a gas triode space discharge de,-V

entiating means and a condenser coupled to theoutpnt of said gas triode whereby the average voltage acrosssaid condenser is proportional to the Voutput of said'differentiatring circuits, and. switching meansV responsive to said generated signal for low frequency.V Y

14. A device according to-claim 13, which further in.;V

cludes a band pass lilter for the alternating voltage output of said gas triode, means to split the phaseV of saidl filtered alternating voltage,'means to rectify output Vof said vphase splitter, and means to add rectified output to said` voltage across said condenser.

References Cited in the tile of this patent y Y UNrTED STATES PATENTS 2,243,527 Dudley ,May.27, 1941`V` 2,635,146 'Steinberg' Y Apr. 14,11953V YOTHER REFERENCES Journal, Institution of ElectricalEngineers, London, 35, v'ol, 975', part IVII (1947), pp. 3914111 (copy avail., Div.

vice operatively responsive4 to the output of said dilfer-V effecting separation of said highand' Vossberg Nov. 1.7, .1,9'5'3.

Images