US3294918A

US3294918A - Electronic conversions of speech

Info

Publication number: US3294918A
Application number: US195903A
Authority: US
Inventors: Gold Nathan
Original assignee: Polaroid Corp
Current assignee: Polaroid Corp
Priority date: 1962-05-18
Filing date: 1962-05-18
Publication date: 1966-12-27
Anticipated expiration: 1983-12-27

Description

Dec. 27, 1966 N. GOLD 3,294,918

ELECTRONIC CONVERSIONS OF SPEECH Filed May 18, 1962 2 Sheets-Sheet 1 AMPLIFIER AND I PRE-EMPHASIS EE gE EgSE ll CIRCUITRY IO 8 9 I I70 K TRANSMITTER q FREQUENCY MULTIPLIER CIRCUITRY MEL": ILAMPLIFIER LIMlTERJ MIXER FILTERTL 7 7- l7b r"-- LQiAL l OUTPUT 8 cIRcuITRY T 7 OSCILLATOR TRANSMITTER CiRCUlTRY REACTANCE NATHAN GOLD MODULATOR 2 INVENTOR.

FIGZ 13% M7 4/ ATTORNEYS Dec. 27, 1966 N. GOLD ELECTRONIC CONVERSIONS 0F SPEECH Filed May 18, 1962 OUTPUT OF NETWORK OUTPUT OF INFINITE CLIPPER OUTPUT OF FREQUENCY METER HIGH FREQUENCY PRE-EMPHASIS 1 FIG?) ll OOII 4 MILLISECONDS PER CENTIMETER 2 Sheets-Sheet 2 FIG. 5

ll ll "EE" (oo-EE) 4 M\LLISECONDS 20 MILLISECONDS PER CENTIMETER PER CENTIMETER MILLISECONDS MAMQAM 0 4 8 I2 l6 0 2O 4O 6O 60 MILLlSEGONDS MILLISEGONDS INVENTOR.

NATHAN GOLD E4 MEWE ATTORNEYS United States Patent O 3,294,918 ELECTRONIC CONVERSIDNS OF SPEECH Nathan Gold, Sharon, Mass., assignor to Polaroid Corporation, Cambridge, Mass, a corporation of Dela- Ware Filed May 18, 1962, Ser. No. 195,903 8 Claims. (Cl.- 179-15.55)

The present invention relates to improvements in the resolution of intelligibility in speech signals and, in one particular aspect, to novel electronic speech-communications apparatus of simplified low-cost construction having improved power and bandwidth characteristics.

Among the well-known fundamental objectives in the communications art are those of raising efiiciency and effectiveness of transmissions to the highest levels and of reducing the frequency spectrum requirements to a minimum. In the case of speech communications, acceptable quality for purposes of intelligibility'has been found obtainable within but a small band of frequencies in the audio spectrum, and radio transmission of such intelligence has further been aided by exploitation of singleside-band techniques which promote an additional con servation of frequency bands. The latter techniques are based upon the recognition that the desired communication of intelligence may be achieved via transmission of but one component (termed a side-band component) of a modulated wave, with the carrier and other components being suppressed. However, it is a notorious failing of single-side-band communications that the required transmitting equipment is complex, with the expected attendant disadvantages.

In accordance with certain of the present teachings, narrow-spectrum characteristics generally similar to those of singleside-band transmissions may also be realized, albeit without need for suppression of a carrier and sideband, such that common uncomplicated modulation and transmission practices may be exploited to particular ad vantage. For these purposes, speech input is converted into unique modulation-signal form wherein speech intelligibility is characterized without involving most of the components ofspeech waves to which the listener is accustomed. Reproduction of the intelligence in terms of audible signals is readily accomplished, however, through heterodyn-ing effects such as are developed within conventional single-side-band receivers. Whereas the usual transmission practices involve a relation of transmitted power to the instantaneous peak amplitudes of speech signals, the improved techniques instead relate power to average values of the speech signals and thereby render the transmissions significantly more effective at any given power level. In addition, highly eflicient non-linear amplifiers may be employed to distinct advantage in connection with the unique signals wherein intelligence is conveyed other than by amplitude characteristics.

It is one of the objects of this invention, therefore, to provide improvements in the coding or resolution of intelligibility in speech into related electrical signals wherein the characterizing variations occupy a narrow band of frequencies.

A further object is to provide novel and improved apparatus for translating speech into related electrical signals involving only low rates of change to express the intelligibility content of the speech.

Another object is to provide improvements in electronic communications of the intelligibility in speech whereby both power and bandwidth are conserved.

A yet further object is to provide novel and improved speech-communications equipments of low-cost and uncomplicated construction which operate with low power within narrow bandwidths and which are compatible for system uses involving single-side-band receivers.

It is well known that speech communication is possible over a system that introduces infinite peak clipping, i.e., clipping that reduces speech to a succession of rectangular waves of constant amplitude in which the discontinuities correspond to the crossings of the time axis in the original speech signal (e.g., see Efiects of Differentiation, Integration, and Infinite Peak Clipping Upon Intelligibility of Speech by J. C. R. Licklider and I. Pollack, Journal of the Acoustical Society of America, January 1942). Since the only variable in the output of an infinite peak clipper is the repetition rate of the wave (i.e., the rate at which the polarity of the constant-amplitude rectangular wave reverses relative to its average value), it follows that the intelligence contained in the speech producing the wave can be conveyed from one point to another by converting the repetition rate of the wave to a modulation signal which can then be modulated on a carrier. Recovery of the modulation signal at a remote receiver permits the output of the infinite peak clipper to be substantially duplicated with the result that the information in the speech is made available.

The present invention is concerned with the modulation of a carrier with the repetition rate of the output of the clipper, and with the recovery of the information contained in the original speech. The basic concept involves the recognition that causing the instantaneous deviation in frequency, on one side of the center frequency of a variable frequency oscillator producing an amplitudelimited signal, to be the same as or a submultiple of the instantaneous repetition rate of the output of the clipper, will produce a signal which simulates a singleside-band (SSB) signal and is compatible with a conventional SSB receiver. Furthermore, since the transmitter output is essentially a constant amplitude wave, the transmitter operates at peak power independently of the intelligence in the speech and the signal will exhibit good noise rejection characteristics. All of the advantages of SSB transmission plus transmission at peak power for all modulating frequencies are achieved without the need for carrier-suppression or side-band filtering.

Briefly, the invention involves the conversion of the output of the infinite peak clipper to a unidirectional signal whose amplitude at any instant is a measure of the repetition rate of the output. This unidirectional signal is used to control the frequency deviation of an RF oscillator from its center frequency so that the oscillator output is the same as an SSB transmission of a constant amplitude but variable frequency sine-wave signal. For this reason, this aspect of the present invention is compatible with conventional SSB receivers when the instantaneous frequency deviation of the transmitter oscillator is the same as the instantaneous repetition rate of the output of the clipper.

In another aspect of the invention, the instantaneous frequency deviation of the transmitter oscillator can be made a submultiple of the repetition rate of the output of the clipper. In such case, the output of the oscillator is still an SSE-type of signal, but is no longer compatible with a conventional SSB receiver for the reason that conventional demodulation will not reproduce a signal whose instantaneous frequency is the same as the original, with the result that the output of the receiver will be unintelligible. By modifying the conventional SSB receiver so that the frequency of the received signal is multiplied by the inverse of the submultiple by which the frequency deviation of the transmitter oscillator is related to the repetition rate of the output of the clipper, the intelligence in the original speech can be recovered after heterodyning with the local oscillator of the receiver. This last-mentioned aspect of the invention has two important advantages: it permits a reduction in bandwidth by the factor of the submultiple; and it permits the transmission of secured messages since recovery of the intelligence in usable form requires knowledge of the submultiple factor.

Although the features of this invention which are considered to be novel are expressed in the appended claims, further details as to preferred practices of the invention, as well as the further objects and advantages thereof, may be most readily comprehended through reference to the following description taken in connection with the accompanying drawings, wherein:

FIGURE 1 comprises a block diagram of an improved narrow-band communications system exploiting the teachings of this invention;

FIG. 2 is a circuit diagram, partly in schematic and partly in block-diagram form, of speech-conversion and transmitter apparatus for efficient voice communication within a limited band of frequencies;

FIG. 3 illustrates graphically a set of four related wave forms which characterize the conversions of relatively low-frequency vowel sounds into a narrow-bandwidth unidirectional signal;

FIG. 4 illustrates graphically a set of four related wave forms which characterize the conversions of relatively high-frequency vowel sounds into a narrow-bandwidth unidirectional signal; and

FIG. 5 illustrates graphically a set of four related wave forms which characterize the conversions of successively occurring relatively lowand high-frequency vowel sounds into a narrow-bandwidth unidirectional signal.

The speech-communications system depicted in FIG- URE 1 includes a cooperating radio transmitter 6 and remote receiver '7 which are linked via radiated modulated carrier waves the modulation of which may embrace, in one form of the invention, a bandwidth as small as about 500 cycles, which is highly restricted in comparison with the 3 to 4 kilocycle bandwidths commonly required in voice-transmission systems. As radiated, these waves possess modulated characteristics which are uniquely related to the intelligibility of voice signals impinging upon a microphone 8, but these modulation characteristics are nevertheless unintelligible as speech per so until a conversion is performed to impart characteristics which restore speech qualities. The unique modulation signal is developed within the transmitter by amplifier and pre-emphasis circuitry 9, clipper 10, and frequency meter circuitry 11. Voice signals are translated into corresponding electrical signals by microphone 8, and, thereafter, electrical pre-emphasis in circuitry 9 accentuates the higher-frequency components while subordinating or removing the low-frequency components. In suitable amplified form, the accentuated higher-frequency components are then clipped, or limited, whereupon substantially all amplitude variations are eliminated and the resulting signal is in essentially square-wave pulse form wherein the preserved relationships to the original intelligence in the voice signals is expressed by the pulse repetition rate or frequency. Variations in the slopes of leading and trailing edges of the pulses can introduce unwanted spurious frequency components which would cause distortions, and for this reason the clipping is preferably of the so-called infinite type produced by a trigger circuit the square-wave output pulses of which may be of large amplitude and will have substantially uniform edge slopes. Frequency, or pulse periodicity, of the clipped signals is then metered by a known form of frequency-metering circuit to produce an average or direct current proportioned to the periodicity or frequency of the pulses being metered. In one preferred construction, the frequency metering function is performed by differentiating the clipped signal, rectifying to produce unidirectional pulses, and passing such pulses through a simple low-pass filter, such as one tuned to pass signals up to a selected low-maximum frequency. The unidirectional output of the frequency meter circuitry 11 is in this manner caused to exhibit a wave form the lowof speech, or'the like.

frequency variations of which characterize the variations in the frequency of higher-frequency components of the original speech waves and, hence, the intelligibility in the speech.

It is found that only a relatively low band of frequencies is required for the unidirectional output to express the speech intelligibility, particularly in that the rates of change in speech frequencies needed to express the communicated intelligence are not high, even though the actual frequencies sounded at various times may themselves be very much higher than the frequencies accornmodated by the unidirectional output signals. Theintelligibility thus resolved into its simplest form is also found to be remarkably isolated from the individualities or personal qualities of the voices of different speakers, such that substantially the same intelligibility communicated by different voices is characterized by substantially the same wave forms in the unidirectional outputs of the network thus far described. Accordingly, such networks and practices in the resolution of speech lend themselves to exploitation in the automatic recognition and processing of voice signals, for such varied purposes as voicecontrolled automatic telephone dialin automatic printing However, the specific equipment illustrated in FIGURE 1 processes the resolved intelligihility of the voice input -by translating it into a conventional modulation of radio waves for direct voice-oommunication purposes. The translation is accomplished by applying the unidirectional low-frequency output signals to a suitable frequency modulator 12, which in turn appropriately shifts the frequency of the RF. carrier generated within the transmitter circuitry 13 and radiated by the transmitting antenna 14, the deviation being the same as or a submultiple of the frequencies of the output delivered by the clipper. At a remote receiver 7, the receiving antenna 15 delivered the intercepted uniquelymodulated RF. signals to appropriate input stage cir cuitry 16 and, thence, the intelligence-conveying output signals are processed in translating circuitry 17 which yields reproductions of the accentuated speech signals in approximations of the conditions in which they -left the clipper circuitry in the transmitter. Thereafter, suitable output circuitry 18, such as a known form of audio amplifier circuitry delivers the amplified reproduced elecplitude variations and most of the low-frequency characteristics are substantially eliminated, the quality of sound in the communication is substantially uniform, and there is an unvarying volume despite varying dynamics in the speech input. The latter aspects of the speech translations are advantageous in that they entail the suppression of two non-essential variables.

The constructional detail of receiver 7 may of course exploit a variety of known techniques and forms for present purposes. However, maximum advantages are believed to be realized when the system transmissions are of the frequency-modulated type and the receiver is of a known type employed in the reception and translation of single-sideband signals. Modulator 12 in transmitter 6 is then conveniently in the form of a known reactance modulator governing the react-an'ce and, hence, frequency of a carrier developed by a suitable variable oscillator in transmitter circuitry 13. It is to be noted that the resulting transmitted signal need not itself be a single-sideaband signal, but, instead, conveniently assumes the form of a frequency-modulated wave merely having frequency variations in accordance with the variations in the unidirec-- tional signals applied to the modulator from frequencymetering circuitry 11. Because the intelligence being communicated is not represented by amplitude variations, the transmitter power stages nee-d not include critical linear amplifiers and preferably utilize vastly more efficient Class C amplifiers. The same is true of the input amplifier circuitry 16a in the receiver, and important rejection of noise is there achieved through further use of a conventional form of limiter 16b. Receiver translating circuitry 17 is conveniently in the form of a frequency multiplier 17a and a single-side-band demodulator including a mixer 17b and a local oscillator 170, these being followed by a filter 17d which passes only the desired speechfrequency band corresponding to the band of frequencies appearing in the output of the transmitter pre-emphasis circuitry 9. Frequency multiplier 17a functions to multiply the frequency of the transmitted signal so that the resulting frequency deviation equals the frequency of the output signal derived from clipper 10. Output of local oscillator 170 is heterodyned with the incoming frequencymodulated high-frequency signals in mixer 17b to develop modulation products including the audio-frequency signals which are of interest. Filter 17d separates the desired audio-frequency signals fro-m the mixer output and delivers them to the audio reproducer through a conventional type of output circuitry 18, which may include a de-emphasis network and amplifier, for example.

A schematic layout in FIG. 2 exemplifies constructional details involved in one practice of these teachings. Insofar as the transmitter components there are the same or the functional equivalents of those rep-resented for transmitter 6 in FIGURE 1, these are, for convenience in description, identified by the same reference characters, with distinguishing single-prime accents added. Speech input from microphone 8 is applied to the first stage 20, of a simple three-stage audio amplifier, between the second and

third stages

21 and 22 of which there is interposed a pre-emphasis filter 23 wherein the capacitor 24 and resistanm 25 impart an accentuation to the higherfrequency speech components and effect a suppression of the lower-frequency components of the speech, for purposes explained earlier herein. Suitable silicon diodes 26 and 27 in back-to-back interconnection function as a limiter, to prevent overdriving in the circuitry which follows, their limiting action being especially advantageous in that the intelligence being tracked through the higherfrequency components of the speech signals is not diminished while, at the same time, the unwanted high signal amplitudes or peaks are eliminated. This practice is uniquely enabled by the distinctive character of the present teachings which recognize that only the frequency of the speech signals, within a limited range of the higher audio frequencies, need be relied upon to develop efficient and effective voice communications. Output of stage 22, though involving a certain limiting of the reproduced signals, is preferably rendered more accurately limited and squared by the clipping stage 10 which is in the form of a Schmidt trigger circuit. This circuit responds to the applied signals by abruptly alternating the flows of currents in its two tube sections 28a and 28b when the applied signal changes polarity, and, importantly, its intrinsic current-conduction characteristics are such that the output pulses each have a substantially constant slope for their leading and trailing edges, as well as having substantially constant amplitudes. One advantage which accrues from use of the trigger circuit for clipping. or limiting, is that it inherently produces an output of high signal strength responsive to even weak input signals, without need for separate amplification. A companion and cardinal advantage lies in the fact that the infinite clipping through use of a trigger circuit removes the distortions of the intelligence which could otherwise result from the unpredictable frequency components represented by the varying slopes of the leading and trailing edges of the input signals. Here it should be borne in mind that the desired intelligence is intended to be characterized solely t5 by frequency or periodicity of the input, such that the extraneous and spurious frequency components introduced with the aforesaid variable slopes are sources of error. When the square-wave output of the clipper involves substantially constant rates of change of output (i.e. constant slopes of leading and trailing edges), this variable is removed, and any unwanted frequency components are fixed, do not impair the intelligence being communicated and can be filtered out readily if need be. Both signal gain and efficiency are thus enhanced by this circuitry. The two square-wave outputs of the trigger circuit, carried vai couplings 2-9 and 30, are of opposite polarity and are each individually differentiated, by difierentiators involving capacitors 29a and 30a and resistances 29b and 3%, respectively, before being applied to the double-diode rectifier 31. Output of the latter rectifier is unidirectional and is at twice the repetition rate of the input to circuit 10, and consists of sharp pulses having abrupt leading edges, rather than square waves. These sharp pulses or spikes are next applied to a single-shot multivibrator 32, which develops an output pulse of one predetermined height and one predetermined width in response to each input spike. Multivibrator parameters are selected, in accordance with known techniques and practices, to insure production of uni-directional pulses having pulse widths less than the smallest expected time separation between successive spikes of input. Low-pass filter 33 serves to integrate these unidirectional pulses of predetermined proportions, whereupon the resulting output is a frequency-metering unidirectional signal the variations in magnitude of which characterize the variations in fre quency of the original speech wave in the higher-frequency range. The frequency-metering by filter or integrator 33 is conveniently performed by a simple LC filter network, although others such as an RC network will also perform satisfactorily for-these purposes, and the tuning is to pass only low frequencies such as those below about 1,000 cycles in one example. Reactance modulator 12 responds to the frequency-metered unidirectional signals by exercising a related control over the frequency of output signals generated by the FM transmitter circuitry 13', and it will be understood that the frequency variations imparted to the radiated FM signal may be the same as or merely some fraction of the actual frequency in the higher-frequency components of the speech intelligence being communicated. In the latter case, there is effected a frequency compression in the transmitted signal and the receiver then may utilize a frequencycxpansion circuit to multiply the intelligence-carrying signals back to the approximate values of the related speech frequencies. Practice of the present invention therefore necessitates the use of but a relatively narrow frequency spectrum for highly effective communications. This invention thus permits a greater number of transmissions to take place within an available communication band than would normally be possible using conventional techniques. Although this is in itself an important advance, it is a further and distinctly major advantage that these teachings entirely circumvent the notorious complexities, costly manufacture, and critical adjustments involved in sin-gle-side-band transmitters, which are perhaps the most nearly akin a-mon-g know prior equipments. However, it should here be recognized that the present invention does not require the generation or transmission of single-sideband signals, even though conventional single-side-band receivers may be used, and that carrier-suppression difficulties are thus never encountered. Moreover, there are other major advantages in that the efiiciencies, effective use of power, and immunities from noise are inherently of the highest order. The latter benefits accrue principally from the immediate translations of the intelligence components of the original speech input into limited pulse-type signals which do not entail amplitudemodulation information, whereby the most common and troublesome noise influences are isolated and whereby average power is equal to peak power such that the rated power output is fully and continuously utilized. Simple non-linear amplifications can thus be resorted to, and the transmitting equipment operates at high eiticiency, with high levels of whatever power is available being radiated to establish significantly more effective communications than would be the case were peak power levels informative in the system. For any given available transmitter power, therefore, practice of the present invention effects more reliable and distant communications. Character of the radiated signals is such that limiting may be exploited in the receivers, to reject noise.

Operating characteristics of the equipment for translating speech intelligence into the narrow-band unidirectional signals are portrayed graphically by the four timecorrelated curves in each of FIGS. 3, 4 and 5, each of which is a representation of an actual oscillogra-m trace. A voice input corresponding to the speech sound develops a directly-related electrical wave form 34 in the output of a microphone and of initial amplifier stages. Filtering and accentuation in the pre-emphasis circuitry removes the lower-frequency components of this Wave, leaving principally the higher-frequency components, such as those of about 1 kc. and above to 3-4 kc, as characterized by wave 34a. These accentuated frequency com- .ponents trigger generation of the clipped square-wave signal 34b, wherein the intelligence is identified by frequency characteristics of the wave. Finally, frequencymetering, as described hereinabove, develops the unidirectional signal 34c in which the rates of changes in signal magnitudes are low and the frequency band is correspondingly narrow. The resolution of inputs into highly simplified form in accordance with these coding techniques is more dramatically evidenced in the cases of wave forms 35-650 and lid-35c, which represent, respectively, the processing of the speech sound cc and the combined speech sound we (or oo-ee), both of these involving higher-frequency components than in the case of the wave form 34. It should be noted that the time scale (abscissa) for the combined speech sound curves 36-36c is more compressed than in the other two cases, such that the resolved unidirectional output 360 does not possess the high frequencies which otherwise might seem to be present, and, elsewhere in the Wave terms, there are minor evidences (disappearances of traces, misalignment of zero level, etc.) of the fact that the curves originated from oscillograms rather than being idealized curves.

The speech conversion equipment (811, FIGURE 1, and 8'11', FIG. 2) lends itself particularly well to use with existing forms of FM transmitter apparatus, whereby the addition of this relatively uncomplicated equipment as a separate package makes possible a unique type of transmission having pronounced advantages over singleside-band transmissions. It will also be understood by those skilled in the art that forms of circuitry other than those discussed herein may be employed with like advantageous effects.

While specific practices have been described, and while particular embodiments have been illustrated and referred to in the descriptions, it should be appreciated that various changes, modifications and substitutions may be made without departure from these techings, and it is aimed in the appended claims to embrace all such variations as fall within the true spirit and scope of the invention.

What is claimed is:

1. Apparatus for communicating the intelligibility in speech by way of electrical waves of narrow bandwith, comprising means producing electrical signals character izing only the higher-frequency components of speech waves, means limiting the amplitudes of said signals to a substantially uniform value, frequency-metering means responsive to the limited signals producing unidirectional electrical waves having magnitudes related to the frequencies of said signals, means for producing and transmitting a carrier wave of radio frequency, modulator means'for frequency modulating said carrier wave in accordance with said unidirectional electrical waves, means for receiving the frequency-modulated carrier Wave and including oscillator means and means mixing the out put of said oscillator means with the received frequencymodulated wave to produce audio-frequency signals having frequencies related to the magnitudes of said unidirectional electrical waves, and means forconverting said audio-frequency signals into sound waves.

2. Apparatus for communicating the intelligibility'in speech by way of electrical waves of narrow bandwidth, comprising speech pre-emphasis means producing electrical signals characterizing only the higher-frequency components of speech waves, trigger circuit means responsive to said signals producing substantially rectangular waves having edges of substantially predetermined slope which coincide with alternations in said signals, multivibrator means producing a single pulse of one pre determined amplitude, width and polarity responsive to each of said rectangular waves, means averaging the pulses from said multivibrator means to produce unidirectional electrical Waves having magnitudes related to the frequencies of said signals, means for producing and transmitting a carrier wave of radio frequency, re-actance modulator means for modulating the frequency of said carrier wave in accordance with said unidirectional electrical Waves, means for receiving the frequency-modulated carrier wave and including oscillator means and means mixing the output of said oscillator means with the received frequency-modulated wave to produce audio-frequency signals having frequencies related to the magnitudes of said unidirectional electrical waves, and means for converting said audio-frequency signals into sound waves.

3. Apparatus for communicating the intelligibility in speech by way of electrical waves of narrow bandwidth, comprising means producing electrical signals characterizing only the higher-frequency components of speech Waves, means limiting the amplitudes of said signals to a substantially uniform value, frequency-metering means responsive to the limited signals producing unidirectional electrical waves having magnitudes related to the frequencies of said signals, means for producing and transmitting a carrier wave of radio frequency, modulator means for frequency modulating said carrier wave in accordance with said unidirectional electrical waves to produce frequency deviations which are in a submultiple relationship to the periodicities of said limited signals, means for receiving the frequency-modulated carrier wave, including frequency-multiplier means raising the frequency of the deviations in the received wave to produce substantially the same periodicities as those of said limited signals, oscillator means, and means mixing the output of said oscillator means with the output of said frequency-multiplier means to produce audio-frequency signals having frequencies related to the magnitudes of said electrical waves, and means for converting said audio-frequency signals into sound waves.

4. In a voice communication system:

(a) a source of speech signals;

(b) infinite peak clipper means responsive to said speech signals for producing a constant amplitude rectangular wave;

(c) means for producing a unidirectional signal whose instantaneous amplitude is a measure of the instantaneous repetition rate of said constant amplitude rectangular wave;

(d) variable frequency oscillator mean for generating a carrier frequency signal in the frequency range exceeding the audio range;

(e) means responsive to said unidirectional signal for causing the frequency of said oscillator means to have an instantaneous value which differs from the value when said repetition rate is zero by an amount equal to said instantaneous repetition rate;

(f) the frequency of said oscillator means, when said repetition rate is zero, being said carrier frequency; and

(g) means to transmit the output of said oscillator means.

5. A voice communication system in accordance with claim 4 including means for receiving the transmitted output of said oscillator means comprising:

(a) a source of CW signal whose frequency is the same as the frequency of said oscillator means when the value of said repetition rate is zero;

(b) mixer means to mix said CW signal with the output of said oscillator means; and

(c) filter mean responsive to the output of said mixer means for producing a signal Whose instantaneous frequency is the same as said instantaneous repetition rate.

6. In a voice communication system:

(a) a source of speech signals;

(d) variable frequency oscillator means for generating a carrier frequency signal;

(f) the frequency of said oscillator means, when said repetition rate is zero, being said carrier frequency;

(g) means to transmit the output of said oscillator means; and

(h) means for receiving the transmitted output of said oscillator means including:

( 1) a source of CW signal whose frequency is the same as the frequency of said oscillator means when the value of said repetition rate is zero;

(2) mixer means to mix said CW signal with the output of said oscillator means; and

(3) filter means responsive to the output of said mixer means for producing a signal whose in stantaneous frequency is the same as said instantaneous repetition rate.

7. In a voice communication system:

(a) a source of speech signals;

(d) a variable frequency oscillator means; and

(e) means responsive to said unidirectional signal for causing the frequency of said oscillator means to have an instantaneous value which differs from the value when said repetition rate is zero by .an amount which is a submultiple of said instantaneous repetition rate.

8. A voice communication system in accordance with claim 7 including a receiver comprising:

References Cited by the Examiner UNITED STATES PATENTS 4/ 1935 Steinberg 17915.55 10/ 1950 Pierson 324--7 8 10/ 1958 Schroeder 17915.S5 11/1961 KWast 324-78 2/ 1962 Edson 324-78 8/ 1964 Firestone l7866 DAVID G. REDINBAUGH, Primary Examiner.

S. J. GLASSMAN, Assistant Examiner.

Claims

1. APPARATUS FOR COMMUNICATING THE INTELLIGIBILITY IN SPEECH BY WAY OF ELECTRICAL WAVES OF NARROW BANDWITH, COMPRISING MEANS PRODUCING ELECTRICAL SIGNALS CHARACTERIZING ONLY THE HIGHER-FREQUENCY COMPONENTS OF SPEECH WAVES, MEANS LIMITING THE AMPLITUDES OF SAID SIGNALS TO A SUBSTANTIALLY UNIFORM VALUE, FREQUENCY-METERING MEANS RESPONSIVE TO THE LIMITED SIGNALS PRODUCING UNIDIRECTIONAL ELECTRICAL WAVES HAVING MAGNITUDES RELATED TO THE FREQUENCIES OF SAID SIGNALS, MEANS FOR PRODUCING AND TRANSMITTING A CARRIER WAVE OF RADIO FRQUENCY, MODULATOR MEANS FOR FREQUENCY MODULATING SAID CARRIER WAVE IN ACCORDANCE WITH SAID UNIDIRECTIONAL ELECTRICAL WAVES, MEANS FOR RECEIVING THE FREQUENCY-MODULATED CARRIED WAVE AND INCLUDING OSCILLATOR MEANS AND MEANS MIXING THE OUTPUT OF SAID OSCILLATOR MEANS WITH THE RECEIVED FREQUENCYMODULATED WAVE TO PRODUCE AUDIO-FREQUENCY SIGNALS HAVING FREQUENCIES RELATED TO THE MAGNITUDES OF SAID UNIDIRECTIONAL ELECTRICAL WAVES, AND MEANS FOR CONVERTING SAID AUDIO-FREQUENCY SIGNALS INTO SOUND WAVES.