US2269295A - Compressed band telephony - Google Patents

Description

Jan. 6, 1942.

C. W. VADERSEN COMPRESSED BAND TELEPHONY Filed Aug.. 21,- 1940 ATTORNEY Patented Jan. 6, 1942 COMPRESSED BAND TELEPHONY Charles W. Vadersen, Near. York, N. Y., assignor to Bell f Telephone Laboratories,

Incorporated,

` New York, N. Y., a corporation of New York Application August 21, 1940, serial No. 353,485

9 Claims. l

The present invention relates to the transmission of speech or similar Waves with reduced frequency range. Such transmission is useful either where the transmission channel is incapable of transmitting normal speech band Width or Where it is desired to conserve frequency space for any reason. v

It has been pro-posed heretofore to conserve band width by taking advantage of the observed fact that vowel sounds and consonant sounds, or more accurately, voiced sounds and unvoiced sounds, do not occur simultaneously in speech but occur in successive 'time intervals; and of the further fact that unvoiced sounds as a rule average higher in the frequency scale than voiced sounds, In such proposal, a method was used for vdepressing the frequenciesof the unvoiced sounds when they occur, so Ithat they occupy the same frequency band as the yvoiced sounds. This method involved `separating the two kinds of sounds and treating them differently in transmission. This required mechanism for performing switching voperations at the frequency of occurrence of the two kinds of sounds, at approximately syllabic frequency.

The present invention, While also shifting the frequencyfof certain elemental :speech compo-v nents to overlap in frequencycertainother components, is based upon a different principle. In

accordance with the present 'invention thespeech is subdivided or chopped at a relatively high rate and the elements that have Vtheir frequencies shifted are chosen without regard 'to' Whether they are voiced or unvoicedsounds. The highv rate of chopping has two very important advantages as regards freedom from distortion, First, the principle of persistence of hearing is used, whereby the ear does not perceive thevreceived fragments of speech separately and individually but they merge into one more or less continuous sensation-analogous to the effect. of persistence of vision. Second, the rate of chopping is not only high but it is chosen to be th'e same as, and varies with, the fundamental or vocal cord frequency of the voiced sounds, which are the high energy sounds. This has been found to eliminate theVV raucous, rasping type of distortion that is observed to result from lchopping speech at any fixed high rate or at any h'igh rate which is different from the -fundamental frequency. Both the articulation andthe naturalness are, therefore, enhanced by the method of the invention as compared with prior art methods.

The nature of the invention 'andits various 5153 plete system from transmitter to receiver em*-v bodying the invention; and

Fig. 2 is a frequency `diagram to be referred to in `the course of the description.

Referring to Fig. l, the speech that is to be 10u transmitted is assumed for illustration to liel within the frequency range of 200 to 3200 cycles per second. This speech is produced by talking into the vtransmitter l. After passing through the voice-operated gain-adjusting device (vogad) 2 lthe speech divides into two paths,

one leading to the delay device 3 and the other leading to the band-pass filter 4. The vogad may be of ythe type disclosedsin Mitchell et al.

n 2,019,577, November v5, 1935, or other suitable zoiv'yp Ther apparatus including 'elements 4 to 9, inclusive, in combination, serve as a fundamental frequency extractor. It takes the vocalharmonies in the region 250 to 1200` cycles and by 2'5"means of a rectifier derives'the difference products which include the fundamental and har- Vmonies thereof. The filter is shown at 4 and the rectifier at 6. The fundamental is the strong- "est component when using a balanced full Wave 1 :awrectier of the type shown lbut in order to strengthen it further relative to its lharmonics an equalizerV 8, which may also contain vamplification is used. This equalizer is a network hav- M ing increasing loss, starting at 80 cycles and'increasing With frequency. High-pass filter 1 With ya cut-olf at 40 cycles is used to attenuate syllabic pulses. desired, Abut is included for illustration for a purpose to be more fully described further on. 4'0This limiter is for the purpose of maintaining the input to the rectifier lconstant for all voice sounds including both voiced and unvoiced sounds.

45, As a result of the action of the elements 4 to at the fundamental or vocal cord rate in the case of voiced sounds and at a random rate in the case of unvoiced sounds and the relay 9 is con- ,structed to vibrate its armatures in synchronism 50 with these pulses. y

The speech which passes through thev delay circuit 3 enters the input side yof the modulatorr I 2 to which there is applied continuous Waves from either the source I 0 The limiter 5 may be omitted, if

inclusive, a succession of pulses is produced' n 4 s n orvthe source ||de; 55 pending upon the position of the armatures of relay 9. Sources IU and II have frequencies that are high compared to speech frequencies and that ydiffer from each other by 1500 cycles. The output of modulator I2 is filtered by a bandpass filter I3 which has a band width of 1500 cycles between frequency limits suitably related to the frequencies of the generators I and Il as indicated in the diagram of Fig. 2. W'hen waves from the generator I0 are applied to modulator I2 the upper side-band is so related to the passband of filter I3 that the upper half of the speech range is transmitted through the iilter. When Waves from generator II are applied to modulator I2 the modulated frequency is 1500 cycles higher than in the first case and the upper sideband is raised sufficiently to permit only th-e lower half of the speech range to lpass through filter I3. As relay 9 vibrates, therefore, the upper half and the lower half of the speech band are transmitted in rapid succession and shifted in frequency to occupy the same frequency band,

The output of the band filter I3 could be passed directly over the line I9 to the receiver in case transmission at a carrier frequency were desired. This can be accomplished in the figure by shifting the D. P. D. T, switches I4 and I5 to their alternate positions so that the transmitting path extends through conductors IB and amplifier Il to line I9. When the D. P. D. T. switches I4 and l5 are in the position shown, the output of band filter I3 is applied to the input of modulator f6 which is supplied with waves from source Il. As a result of this modulation, the 1500-cycle band is shifted downward to occupy the range of 200 to 1700 cycles which is transmitted through amplifier I'l Yto line I9.

The received waves pass through band-pass filter and in part 'through D. P. D. T. switches 2I and 22 which in one position lead through conductors 29, and in the other position lead through demodulator 23, to the fundamental frequency extractor circuit 24 to 28. The path 29 is used if the transmission over the line is in the range 200 to 1700, while the demodulator `23 is used if the transmission over the line is at high frequency. A continuous wave from the source III, which has the same frequency as source II, is applied to the demodulator 23 to step the received waves in the latter case down to the level of 200 to 1700 cycles. The fundamental frequency extractor circuit 24 to 28 may be identical with that at the transmitter, the parts being indicated by the same reference` characters rais-ed by 20. As a result of the action of the fundamental frequency extractor, the relay 30 vibrates in unison with the relay 9 of the transmitter.

A portion of the speech band transmitted through filter 20 leads to the D. P. D. T. switches 3l and 32 which include either the modulator 34 or the conductors 33 in the circuit. Conductors 33 are used if transmission over the line is at high frequency, while modulator 34 is used if th-e band transmitted over 'the line is in the 200to1700cycle range. Modulator 34 has applied to it continuous waves from generator III for the purpose of stepping the frequency of the received band from 200 to 1700 cycles up to the frequency position of the pass-band filter I3 at the transmitter. This band (which is the same for either position of a switch 32 under the above assumptions) after passing through amplifier 35 and ydelay circuit 36 is applied to the input side of demodulator 3l where it modulates either the wave from generator III] or the wave from generator III, depending upon the position of the armatures of relay 30. As a result of these modulations the received waves are alternately stepped down in frequency by different amounts, those corresponding to the upper half of the speech band being shifted 1500 cycles less in frequency than those corresponding to the lower half of the speech band. As a result the two portions are restored to the same positions as in the original speech so that they occupy a region 200 to about 3000 cycles, there being a slight loss in band width due to the switching operations. Low-pass filter 38 is, therefore, indicated as having a cut-off of 3000 cycles, although it could have a higher cut-off, if desirable. Received waves are heard in the receiver 40 which may be any suitable type of soun-d reproducer. As noted above, the listener does not hear all of the spectrum of the signal at one instant but hears different portions of the sound spectrum in such rapid succession as to get an impression of hearing all of the spectrum at once.

If the limiters Sand 25 were omitted, the switching relays 3 and 30 would follow the fundamental or vocal cord frequency of the voiced sounds but would remain unop-erated by unvoiced sounds because of the relatively low energy level of the latter. Both relays would, therefore, release their armatures to the rest contacts during transmission of unvoiced sounds. In this position of the armatures there is applied to the modulators I2 and 3l, the lower of the two waves, i. e., the wave from generator I0 and generator IIO. Reference to Fig. 2 shows that under these conditions the upper portion of the speech band is transmitted. Since most of the intelligence in the unvoiced sounds lies in the upper portion of the speech frequency band, this method of transmission would enable the major part of the unvoiced sound energy to be transmitted. However, 'the use of the limiters 5 and 25 is preferred, since the energy represented by the unvoiced sounds varies sufficiently rapidly to cause actuation of relays 9 and 30 in random manner, resulting in the transmission of both high and low frequencies of unvoiced sounds. A third method would comprise use of a continuous wave of arbitrary frequency for switching oscillators I Il and II and IIO and III when relays 9 and 30 are deenergized similarly to the method disclosed in Dudley application Serial No. 310,527 filed December 22, 1939.

The purpose of the delay circuits 3 and 36 is to permit the fundamental frequency extractor circuit and switching relays to have time to operate before the speech arrives at the modulator circuits.

While the switching devices have been indicated as relays, other devices, such as space discharge or varistor type of switches using no movable parts, may be used.

In a multiplex system Where several channels are on one line the filter pass-bands and the carrier frequencies can be arranged such that the frequency of generator I0 for one channel is the same as the frequency of generator II for the channel just below. This would result in having to supply only (11+ i) carrier frequencies for n channels instead of 2n frequencies.

Instead of transmitting over a line the transmission could take place over any other suitable type .of medium, such as over a radio channel.

substantially two to one without a corresponding loss of quality.

(2) Privacy on account of the dliiiculity of un derstanding the waves as transmitted with any imple type of receiver.

(3) Increased signal-to-noise ratio arising from the fact that the full load on the system may comprise at any one instant the energy of `only one-half of the speech spectrum, the energysent corresponding to the high speech frequencies being equal to that corresponding to the low speech frequencies. In the usual system vthe average energy falls off forhigher speech fre-V quencies. In order to transmit the upper half of the band at relatively high level so asl to load the system to the same extent at the lower half, the voltage of generator l) may be higherthan that of generator i l whereby the output from modulator I2 is greater by the right factor when energized from source lll than when kenergized from source il, for the same signal input. At the receiver, a compensating change is made by making the voltage of source lill smaller than that of source lll by corresponding amount.

(Il) The need for a separate pilot frequency or channel to synchronize switching is eliminated.

The invention is not to be construed as limited to the specific circuit arrangements or numerical values given, since these are to be taken as illustrative rather than as limiting.. The scope of the invention isk dei-ined in the claims which follow.

What is claimed is:

1. The method of transmission of vspeech waves comprising transmitting, atcertain instants of time, only the upper speech frequencies and, at other and intervening instants of time, only the lower speech frequencies, and alternating between said two frequency ranges at the fundamental vocal cord frequency of the transmitted speech.

2. The method of transmission of speechV fundamental voice period, selecting from said speech wave only the upper frequency portion and transmitting the same in the intervening periods each of said same duration, and shifting the frequency position, prior to transmission, of the waves selected from one of said frequency portions so that they overlap in frequency the waves selected from the other of said frequency portions.

4. In the transmission of frequency range, the method comprising transmitting to line in alternate exclusive times in rapid succession the energyk representing the vupper speech frequencies only and the energy representing the lower speech frequencies only,

speech with reduced length and rapidity of the vocal cord fundamental period. y'

v5. The method of conserving frequencies in the transmission of speech comprisingsubdividing the speech waves on a time basis at the frequency of the voice fundamental, and also on a frequency basis such that only the high frequency components are retained in alternate times and onlythe low frequency components are retained in theintervening times, and shifting alternate 'wave portions in vfrequency to overlie the frequency range occupied bythe intervening wave portions. v.

6. In a system for the transmission of speech, a filter having a pass-band, suicient tocover only a lportion of the speech frequency band,

means to shift the frequency level of a speech.V

band such that `only the high Afrequency portion of thespeech bandis passed through said filter in alternate times andr suchv that only the low frequency portion of the speech band is passed through said ilter in intervening times, and means for changing from one of said shifting `operations to the other at a rate corresponding to the voice fundamental frequency. y

'7.l In aspeech transmissionsystem, means to delinea series of time intervals of the frequency kof the Voice fundamental during occurrence of voiced sounds andv of random frequency during occurrence of unvoiced sounds,l meanstotransmit Waves in a .relatively narrow frequency range taken from' only the high frequency portion of the speech band Vin every other time interval,` and means to transmit waves in said narrow frequency range taken from only the low frequency portion ofthe speech band in each of the intervening time intervals.

8. In a speechtransmission system, a source of speech Waves, an'outgoing path, a frequency translating circuit between said source and said path, said circuit having two alternate operating conditions, said circuit in one condition transmit'- ting to said path waves representing only the upper speech frequencies and in the alternate ,condition transmitting to said path Waves representing only the lower speech frequencies, said circuit shifting the frequency of the waves trans frequency shifting circuit between said path and said receiver, means operating in one condition of said circuit to` apply to said receiver waves having the frequency of the transmitted waves of unshifted frequency, means operating in another v,condition of said circuit to restore the waves of shifted frequency tto the original ,frequencies which they had before shifting, and means for commutating said receiving frequency shifting circuit between its two said conditions in synchronism with the commutations of the frequency shifting circuits at the transmitter.

CHARLES W. VADERSEN.

Images