US3465102A

US3465102A - Readout analyzer

Info

Publication number: US3465102A
Application number: US572864A
Authority: US
Inventors: James R Trammell
Original assignee: US Air Force
Current assignee: US Air Force
Priority date: 1966-08-16
Filing date: 1966-08-16
Publication date: 1969-09-02
Anticipated expiration: 1986-09-02

Description

J. R. TRAMMELL 3,465,102

READOUT ANALYZER 5 Sheets-Sheet l sept. 2, 1969 Filed Aug. 16, 1966 Sept. 2, 1969 .1.R. TRAMMELL READOUT ANALYZER 5 Sheets-Sheet 2 Filed Aug. 16, 1966 Sept. 2, 1969 J. R. TRAMMELL 3,465.102

READouT ANALYZER Filed Aug. 1'6, 1966 5 Sheets-Sherri'l :5

/CFLE SYSTEM 70 l l lA l l l l I FREQUEMC/ 0.9.57

@Y www ff l l l l l l l 2/0 220 e. Armen/Err Sept. 2, 1969 J. R. TRAMMELL.

READOUT ANALYZER 5 Sheets-Sheet 4 Filed Aug. 16, 1966 Wnmvws rugs mwN UANXNX Sept. 2, 1969 1. R. TRAMMELL READOUT ANALYZER 5 Sheets-Sheet 5 Filed Aug. 16, 1966 Patented Sept. 2, 1969 3,465,102 READOUT ANALYZER James R. Trammell, Cleveland, Ohio, assignor to the United States of America as represented by the Secretary of the Air Force Filed Aug. 16, 1966, Ser. No. 572,864 Int. Cl. H04m 1/ 74 U.S. Cl. 179-1 3 Claims ABSTRACT F THE DISCLOSURE A speech analyzer selects a desired voice from a multi- -plicity of voices by examining the harmonic content of the voices. The desired voice is separated from interfering voices by filtering the fundamental along with its related harmonics of the desired voice from the interfering voices.

This invention relates to a readout analyzer and more particularly to a method and system of segregating certain types of voices and other complex sounds by their harmonic content, and separating the desired sound from interfering sounds for further analysis of the information content by filtering the fundamental along with its related harmonics from the unwanted sounds.

Speech consists of a complex acoustic wave caused by the vibration of a persons vocal chords modulating the stream of air passing through the throat. This complex wave consists of the fund-amental frequency and a large number of exact multiple harmonics recurring at the fundamental frequency or pitch of the voice. When this complex wave passes through the resonant cavities of a persons mouth, certain regions of frequencies are reinforced by this resonance, which results in formants which are recognized as vowels and consonants as a person rapidly changes the resonant size of the mouth to form words.

The fundamental pitch of various voices will range from as low as 90 cycles per second for a deep male voice to as high as 300 cycles per second for a high pitched female voice. The pitch variation 0f male voices ranges from 90 to 140 cycles per second. During this discussion We will concern ourselves only with the male voice.

A number of recordings were made of a male voice and the energy level of the fundamental, and all of the harmonics of various words were measured, using a Bruel- Kjear frequency analyzer, and the line spectra charts were made. All of these words were spoken at the same pitch, and consist of energy at the fundamental and each exact harmonic, the difference between words being the -amplitude of various harmonics, depending on the resonant size of a persons oral cavities while speaking that particular syllable or word.

Tape recordings were -made of three diiferent male voices, each voicing the syllable i as in tip. A closed loop of tape was made of this sound and analyzed with .the Bruel-Kjear analyzer to determine the fundamental frequency of each of the three voices. The lowest fundamental frequency was 95 c.p.s., the highest 120 c.p.s., and one between at 108 c.p.s. It is quite `apparent that a series of sharply tuned lters, each tuned to one of the harmonies of a particular voice, this voice would predominate by whatever level difference is Iaccomplished by the Width `and shape of the filter used. The level by which one voice would predominate will depend on the difference in fundamental frequency between the desired voice and the fundamentals of the unwanted voices. In the case of the three voices referred to above, this difference amounted to 12 and 13 c.p.s. Using a lter of the type employed in the Bruel-Kjear frequency analyzer would made possible `a level difference between two tones 12 cycles apart of 3 db, so that a readout analyzer of the type described would cause a voice, selected by harmonically related filters tuned to its fundamental frequency to predominate the other voices by 3 decibels, provided they had been spoken at the same level. In previous tests it was determined by recording three voices at the same time that one voice would become predominate and intelligible if it was recorded at a level approximately 3 to 6 decibels above the other two voices.

The limitations of the system of the present invention will be determined by the characteristics of the lter used, the difference in fundamental frequency of the voices to be separated, and the amount of deviation of the fundamental pitch due to inflection of the voice at the end of sentences and during periods of emphasis.

The present invention utilizes a technique of heterodyning the frequencies with which it is concerned up to some intermediate frequency, and using crystals to give ya very selective band-pass. Thus a heterodyne system is used for the readout analyzer with a total of at least 25 crystal lat-tice filters, one for each harmonic and covering a frequency range of c.p.s. to 2500 c.p.s. The frequency chosen for the filters was 100.1 kc. to 102.5 kc., which would place them 100 c.p.sapart and establish this as the pitch of the system. Any recorded voice could be 'adjusted to this pitch by adjusting :the speed of =a tape transport until all harmonics of that voice would simul- Itaneously pass through the 25 lters, Each of the filters would have to have the narrowest possible bandwidth, which would still allow a rise and decay time sufficiently rapid to enable them to follow the syllabic rate of the human voice without causing syllables Ito run together.

The object of this invention is to provide a method and system of selecting all the related harmonics of complex sounds, such as group conversations, and to separate the voices so one of them will predominate.

Another object of the present invention is to provide a readout analyzer for segregating certain types of voices and other complex Isounds by their h-armonic content, and separating the desired sound from interfering sounds for further analysis of the information content of a system of filtering the fundamental along with its related harmonies from the unwanted sounds.

The various features of novelty which characterize this invention are pointed out with particularity in the claims annexed to `and forming -a part of this specification. For a better understanding of the invention, however, i-ts `advantages and specific objects obtained with its use, reference should be had to the laccompanying drawings and descriptive matter in which is illustrated and described `a preferred embodiment of the invention.

In the drawings:

FIGURE 1 shows the block diagram of the basic readout analyzer of the present invention;

FIGURE 2 illustrates the line spectrum of two voices saying A as in TAN;

FIGURE 3 shows the passband characteristics of filter 47;

FIGURE 4 is a block diagram of the basic system of speed control for the capstan motor of tape player 10;

FIGURE 5 shows two reproducer heads separated by a predetermined distance and their associated tapes that are included in tape player 10 to correct for delay and hunting; and

FIGURE 6 shows a complete readout analyzer in block diagram form.

Now referring to FIGURE 1, there is shown tape player 10 which includes a capstan motor. A 100 c.p.s. voice is included in the information on the tape in tape player 10. This information in electrical signal form is passed through audio amplifier 11 to be fed to balanced modulator 12. Balanced modulator 12 also simultaneously receives a 100 kc. (kilocycle) signal from crystal oscillator 13. Modulator 13 with the 100 kc. insertion signal from crystal oscillator 13 creates upper and lower sidebands for all of the frequencies on the tape.

There are also shown 25 parallel filters which are numbered 16 through 40 and are in channels Li6-60, respectively. The inputs thereto are supplied by the output of balanced modulator 13 by way of line 15 and the outputs from filters 16-40 are fed to product detector 41 by way of line 15. Product detector 41 simultaneously receives the 100 kc. signal from crystal oscillator 13.

Filters 16, 17-40 have been designed to accept the upper sideband information of the desired signal (100 c.p.s.), that is, 100.1 kc., 100.2 kc. through 102.5 kc. Only this information appears in the output of the filters and when it is recombined with the carrier signal of 100 kc. in product detector 41, the 100 c.p.s. voice appears at the output thereof and is further processed in audio amplifier 42 which then provides an output `at terminal 43. The passband characteristics of a representative filter is shown in FIGURE 3.

The complete action may be made clearer if it is assumed that the tape only has two voices, both saying A as in TAN (See FIGURE 2 which is a line spectrurn of two voices saying A as in TAN-solid line at a fundamental frequency of 100 c.p.s.-dotted line at a fundamental frequency of 115 c.p.s.).

Voice No. 1 (solid line) is at a fundamental pitch of 100 c.p.s. and has harmonic energy at each 100 cycle multiple. Voice No. 2 (dotted line) is at a fundamental pitch of 115 c.p.s. and it therefore has its harmonic energy at each 115 cycle multiple. The lters will .accept all of the upper sidebands of 100 c.p.s. voice and with a filter passband of i6 cycles only the following harmonics of 115 c.p.s. voice will pass: the seventh at 805 cycles, the thirteenth at 1495 cycles, and the twentieth at 2300 cycles. Thus almost 87% of the frequencies necessary for intelligibility of the 115 c.p.s. voice are missing in the output and because the 100 c.p.s. voice is allowed to pass with almost no attenuation, it then becomes predominant and easily understood. The recombination of the reinsertion 100 k.c. signal from oscillator 13 and product detector 41 results in only the desired voice being reproduced.

The choice of 100 c.p.s. as a fundamental frequency is arbitrary, because regardless of the actual pitch of the voice, it can be changed to the acceptable frequency of the filter system simply by changing the speed of the tape recorder. Numerous techniques may be used, but in this instance the use of a synchronous motor and a variable frequency 115 volt supply provided the best results.

-It is also necessary to correct for another factor in the voice and that is inflection, because a change of one cycle in the fundamental of the voice shifts the seventh harmonic seven cycles and so the seventh and higher harmonies are out of the passband of the filters. The stringent stability requirements of motor speed correction dictated the use of crystal type control. The variable frequency signal for aforesaid capstan motor supply is derived from the heterodyne difference between two crystal oscillators. The crystal oscillators are made variable over a range of cycles for each by the use of voltage Variable capacitors in series with each crystal. FIGURE 4 is a block diagram of the basic system of speed control for the capstan motor of tape player 10. A manual adjustment is accomplished by the selection of an isolated voltage in manual frequency set component 70 which is applied differentially to voltage

variable capacitors

71 and 72. As the voltage is increased in the positive direction, the action of the voltage variable capacitor is to increase the frequency of

oscillators

73 and 74 and conversely for reduction of voltage.

The speed change necessary to compensate for inflection is accomplished by applying the 200 c.p.s. of the desired voice from tape player 10 to 200 c.p.s. lter 77 to discriminator 78 which provides a corrective voltage to oscillator 74 by way of voltage variable capacitor 72. The voltage from the manual control (which is labeled manual frequency set component and the correction network are series added so that if the voice inflects upward, the capstan motor is made to run slower and compensates for this change.

A more detailed description now is presented for the system shown in FIGURE 4. The output of mixer amplitier is 10() volts peak to peak over the frequency range of 50 c.p.s. through 70 c.p.s. which is applied to capstan motor 76. The frequency change is accomplished by utilizing the heterodyne difference between crystal controlled

variable frequency oscillators

73 and 74. For example, the actual oscillator frequencies for a 60 c.p.s. output are 102.280 kc. for oscillator 73 and 103.340 kc. for oscillator 74. Manual frequency set component 70 is a conventional voltage source operating differentially, that is if the voltage applied to voltage variable capacitor 71 is increased; simultaneously, the voltage applied to voltage variable capacitor 72 is decreased. When the voltages are changed in this manner, the frequency of oscillator 73 increases and oscillator 74 frequency decreases and the difference frequency is now less than 60 c.p.s. The converse occurs if an increase in output frequency is desired.

Oscillator 74 only is used or the automatic control of the output frequency. The DC voltage from discriminator 78 will vary from 0.78 volt to +0.88 volt. The discriminator voltage is applied to voltage variable capacitor 72 which then serves to alter the frequency of oscillator 74.

The outputs of

oscillators

73 and 74 are combined in mixer amplifier 75 which is then applied to capstan motor 76 for the purpose of controlling the speed thereof.

The introduction of the inflection compensation circuitry produced two additional problems. All narrow band filters exhibit some delay in the transit of a signal. The delay is actually a composite of four factors: A delay in the start of the signal; a finite rise time to full amplitude; storage time; and a nite time for the signal to decay to zero.

For the purpose of this discussion it is expedient to lump the delays into rise and decay time and only consider the rise time. The lters used have -a total rise time of milliseconds, but because the corrective elements are only frequency sensitive a usable amplitude of signal appears in the output after 135 msec. Without any compensation, the corrective signal from the discriminator would change the speed of the tape 135 milliseconds after that portion of the tape had passed the reproduce head which is part of tape player 10. To correct for this deliciency, refer to FIGURE 3 where a second reproduce head was installed one inch following the signal head (one inch tape travel at 71/2 i.p.s. requires 133 msec.). The first reproduce head 80, the signal head, now supplies information to the automatic speed control, the 135 msec. delay allows this portion of the tape to tarvel to the second reproduce head 81, the information head, where lthe proper speed correction is applied and the voice appears as a monotone at the frequency which is acceptable to the filter system.

The second problem is hunting The discriminator has zero output voltage when the desired signal is in the center of the passband of the filter system. It must also have zero output voltage when there is no signal so that the speed can be accurately set with the manual control. FIGURE 5 shows a tape which has blocks of information at different frequencies recorded on Track No. 1. Without the anti-hunt provision the following occurs:

The first block of information passes reproducer head 80, the c.p.s. signal goes to the discriminator and issues a plus 5 corrective signal which speeds up the tape just as that block of information reaches reproducer head 81 so that frequency is now read as 200 c.p.s. and

is acceptable to the filters. But the second block of information is now under reproducer head 80 and because the tape speed has increased, this frequency is now read as 200 c.p.s. and when it is applied to the discriminator, the correction voltage becomes zero, the tape starts to slow down and reproducer head 81 reads this block as 195 c.p.s., which is out of the passband of the filters. For purposes of explanation the foregoing has been overly simplified because a rigorous explanation of the step by step action of the hunting would be too time-consuming. It is suicient to say that if the signal is not exactly in the center of the band-pass of the filters, the system will hunt.

The hunting occurs only when the automatic speed control has changed the speed of the tape so that the desired signal is in the center of the band-pass of the filters. At that time the correction voltage disappears and the tape returns to its original speed, which again causes a correction voltage, etc.

The fault lies with discriminator 104 shown in the block diagram of FIGURE 6. Some compensation must be provided so that a lcorrection voltage will always be supplied except when the original recording is at the proper frequency (without automatic speed correction). The carrier signal for modulator 102 is now supplied by variable frequency oscillator 115 whose frequency is controlled by a 200 c.p.s. signal recorded on Track No. 2 of the information tape (see FIGURE 5). The antihunt provision can be demonstrated by going through the same sequence as explained in the paragraph on hunting. In FIGURE 5, when the 195 c.p.s. block (Track No. 1) passes under head 80, a plus 5 corrective signal is developed at discriminator 104 (135 msec. delay), the 200 c.p.s. tone (Track No. 2) produces zero correction. When the 195 c.p.s. block reaches head 81, it is then corrected to 200 c.p.s. Head 80 reads the second block (195 c.p.s.) of information as 200 c.p.s. because of the corrected speed but at the salme time head 80 is reading the 200 c.p.s. on Track No. 2 as 205 c.p.s., which causes a five cycle change in the insertion frequency and discriminator 104 is being supplied with a frequency that is five cycles less than the frequency being read by head S0, Track No. 1. In effect although head 80 is reading the second block of 195 c.p.s. information as 200 c.p.s. the change in speed has caused a cornpensating change in insertion frequency so that the discriminator is always supplied with a signal at a frequency that will properly correct the speed of the tape when it appears under head 81.

Now referring in detail to FIGURE 6 showing a readout analyzer including correction apparatus, there are shown three

input terminals

100, 110, and 120. Terminal 100 receives the electrical signal representative of speech from audio head 80 of FIGURE 5. Terminal 110 receives 200 c.p.s. signal from Track No. 2 0f FIG- URE 5. Terminal 120 receives lan input signal from head 81 of FIGURE 5. The input signals to tenrninals 100, 110, and 120 are fed through

buffer amplifiers

101, 111 and 122 and are recived by

modulators

102, 112 and 122, respectively. Modulator 102 `also receives an input signal from variable crystal oscillator 115.

Modulators

112 and 122 also receives input signals from fixed crystal oscillator 125. Modulator 122 supplies the signal to the filter system, modulator 102 is part of the automatic speed control apparatus, and modulator 112 is part of the antihunt system. These three modulators are identical.

Modulator -122 serves to convert the audio information contained in the tape to a frequency acceptable to the lter system comprised of crystal filters 126-140 which serve as -a separation network for signal frequencies from 100.1 kc. to 102.5 kc., respectively. -Filters 12'7- 140 all operate in parallel, having a common input from modulator 122 and a common output to product detector 141. Gate 142 is in the nonconductive state but when a signal is passed by filter 126 (which is the desired signal) then the gate is open and a 100 kc. reinsertion signal from fixed crystal oscillator is fed through gate 142 to product detector 141. The output from product detector 141 is fed through audio amplifier 143` to output terminal 144. Audio amplifier 143 also serves to filter out the undesired portions of the 100 kc. components. Fixed frequency oscillator 125 was designed to have a stability of one part per million per 24 hours.

As aforementioned modulator 112 receives simultaneously two inputs, one from buffer amplifier 111 and the other from oscillator 125. The output from modulator 112 is fed to discriminator 114. It is to be noted that discriminator 114 is used with variable frequency oscillator 115. Discriminator 114 gets its signal from the continuous 200 c.p.s. signal on Track No. 2 of the tape sho/wn in FIGURE 5. The signal from discriminator 114 is the double sideband information from modulator 112. The output from discriminator 114 is the familiar DC shape ranging from 1.87 volts to +19 volts.

Variable crystal oscillator 115 is required to vary m10 cycles which is well within the range of tuning than can be applied to a crystal and still insure an oscillator that exhibits crystal stability. The frequency may be first adjusted manually to a desired frequency and then an automatic control of the frequency is provided by the DC voltage from discriminator 114 which is applied to a voltage variable capacitor included in oscillator 115. This is a conventional mode of tuning an oscillator.

As previously described, modulator 102 receives simultaneously two inputs, one from buffer amplifier 101 and the other from variable crystal oscillator 115. The output of modulator 102 is passed through filter 103 to the input of discriminator 104. Discriminators 104 and 114 are identical in design except that discriminator 114 gets its signal from the continuous 200 c.p.s. signal on Track No. 2 of the tape shown in FIGURE 5 and discriminator 104 utilizes the second harmonic of the voice signal that appears on Track No. 1 of the tape shown in FIGURE 5. It is to be noted that the discriminator 104 receives a single sideband signal and the output therefrom varies from 0.78 volt to +0.88 volt DC.

Speed control component 105 consists of components 70-75 as shown connected and operated in FIGURE 4. The output of speed control component 105 is fed through power amplier 106 to output terminal 10'7 which is then connected to capstan motor which is utilized t0 control the speed of the tape player associated with the tape and heads and 81 of FIGURE 5. It is emphasized that modulator 122 supplies the voice signal to the filter system, modulator 102 is part of the automatic speed control, and modulator 112 is part of the antihunt system.

The present readout analyzer is a new approach in the field of voice readout and/or analysis. This system is capable of performing the following functions: Take an individual voice that has been recorded on tape at a monotone c.p.s.i5 cycles), separate this voice into its harmonics, then reassemble any number or all of these harmonics in the output until the voice has an intelligibility similar to the original. The intelligibility is very apparent to a trained listener.

The system will allow the selection of either of two voices provided these voices are speaking in a monotone and separated by a m-inimum of eight cycles. Assume the voices at 96 c.p.s. and 104 c.p.s., if the speed of the tape player is reduced, the 104 c.p.s. voice will be changed to 100 c.p.s. and the 96 c.p.s. will be changed to 92 c.p.s., and 92 c.p.s. is outside of the passband of the filter system. Then the 104 c.p.s. voice only will appear in the output. Conversely, if the speed of the tape player is increased, the 96 c.p.s. voice will be changed to 100 c.p.s. and the 104 c.p.s. voice will be changed to 108 c.p.s., and then this time the higher frequency voice will be out of the passband.

If the two voices described in the preceding paragraph are made to speak alternately with Voice No. 1 at 96 c.p.s. and No. 2 at 104 c,p.s., with the tape player speed set so that these voices actually appear at their frequencies, it is possible to automatically pull each voice individually into the center Of the passband so that both voices appear in the output and both will be heard in the output speaking in the same monotone pitch. This section demonstrates the ability of the instrument to remove infiection.

The most important aspect of the foregoing demonstrations is the ability for this instrument to select any voice in a group of voices and reproduce only that voice in the output.

I claim:

1. A speech analyzer to select in intelligible form a desired voice from a multiplicity of undesired voices comprising a variable speed tape player with an associated reproducer head and including said desired and undesired voices recorded upon tape, each of said voices having a different fundamental frequency and associated harmonics, said variable speed tape player includes a capstan motor controlling the speed of said tape player, first and second variable frequency crystal oscillators having a preselected frequency range, first and second voltage variable capacitors connected to said first and second variable frequency oscillators, respectively, for varying the frequencies thereof, a variable voltage source connected to said first and second voltage variable capacitors, said voltage source operating differentially to increase the voltage to said first voltage variable capacitor as the voltage to said second voltage variable capacitor decreases and conversely, a mixer receiving the outputs of said first and second frequency variable oscillators to provide a difference frequency signal to said capstan motor for controlling the speed thereof, a first audio amplifier receiving electrical signals representative of said voices recorded on said tape, a balanced modulator receiving the output signal from said audio amplier, a fixed crystal oscillator generating a continuous wave signal at a preselected frequency, sa-id continuous Wave signal also being received by said balanced modulator, said balanced modulator creating upper and lower sidebands for all the frequencies recorded on said tape, a preselected number of filters connected in parallel and receiving at the input thereof said upper and lower sidebands, each of said filters having a preselected bandwidth and band-pass and each passing only said upper sideband, said preselected number of filters being harmonically related to each other in accordance with a preselected fundamental voice frequency, said variable speed tape player being adjusted in speed to coincide with said preselected fundamental voice frequency to permit only said fundamental frequency and associated harmonics of said desired voice to pass through all of said preselected number of filters While partially passing said frequencies of said undesired voices, means to corubine the output from said filters and said continuous wave signal in a product detector, and a second audio amplifier receiving the output signal from said product detector to provide at the output thereof only said desired voice in intelligible form.

2. A speech analyzer as defined in claim 1 further including compensation for change of infection of said desired voice comprising means to pass a preselected electrical component of said desired voice from said tape player at a preselected frequency bandwidth, said electrical component representing said change in said infiection, and a discriminator receiving said electrical component and providing a varying direct current output in accordance with said change in inflection, said varying direct current voltage being received by said first voltage variable capacitor for control thereof.

3. A speech analyzer to select in intelligible form a desired voice from a multiplicity of undesired voices, each of the voices having a different fundamental frequency and associated harmonics, comprising a variable speed tape player, first and second reproducer heads associated with said tape player, said first reproducer head positioned a preselected distance from said second reproducer head, a two track tape associated with said reproducer heads, the first of said tracks having recorded thereupon electrical signals representative of said desired and undesired voices and the second of said tracks having recorded thereupon a preselected control signal of predetermined frequency for anti-hunting utilization to control said variable speed tape player, a capstan motor controlling the speed of said tape player in accordance with an A.C. signal, a first audio amplifier receiving electrical signals of said voices from said first reproducer head, a first balanced modulator receiving the output signal from said first audio amplifier, a first oscillator generating a first continuons wave signal of a preselected frequency, said first continuous Wave signal being received by said first balanced modulator, said first balanced modulator providing upper and lower sidebands of said desired and undesired electrical voice signals, a preselected number of filters connected in parallel and receiving at the inputs thereof said upper and lower sidebands and passing only said upper sidebands, said preselected number of filters passing a preselected fundamental voice frequency and harmonically related frequencies, said variable speed tape player being adjusted in speed to coincide with said preselected fundamental voice frequency to permit only said preselected fundamental frequency and associated harmonies of said electrical signals of said desired voice to pass through all of said preselected number of filters while partially passing said frequencies of said electrical signals of said undesired voices, a gate normally nonconductive, receiving said first continuous wave signal for passage therethrough upon the receipt of an output signal from one of said preselected number of filters, a product detector receiving simultaneously first continuous wave signal by way of said gate and output signals from said preselected number of filters to provide at the output thereof only said electrical signals of said desired voice, in intelligible form, a second balanced modulator receiving said antihunting control signal from said second track of tape player, said second balanced modulator also receiving said first continuo-us wave signal, a first discriminator receiving the output signal from said second balanced modulator to provide a first D.C. control isgnal, a voltage variable oscillator receiving said second D.C. signal to control the frequency thereof, a third balanced modulator receiving electrical signals of said desired and undesired voices from said second reproducer head, said third balanced modulators also receiving the output signal from said voltage variable oscillator, a second discriminator receiving the output signal from said third balanced modulator to provide a second D C. output signal, and a voltage variable frequency generator receiving said second D.C. voltage and varying the frequency thereof in accordance with said second D.C. voltage, the output of said second voltage variable frequency generator being received by said capstan motor for controlling the speed thereof.

References Cited UNITED STATES PATENTS 3,079,464 2/1963 Baumel 179-1 RODNEY D. BENNETT, JR., Primary Examiner C. E. WANDS, Assistant Examiner