US2957134A - Fundamental frequency extractor from speech waves - Google Patents

Description

Oct. 18, 1960 vHKALFAlAN 2,957,134

FUNDAMENTAL FREQUENCY EXTRACTOR FROM SPEECH WAVES Filed Sept. 16, 1957 2 Sheets-Sheet 1' z I l l FLIP-HOP \1 B aurpur WAVES J J I l AS MARKER l POINTS or mans rams ---TII1E BASE Fig.5

VOICE INPUT :E OUTPUT b C28 1 R4G: "F.

INVENTOR.

. Oct. 18, 1960 M. v. KALFAIAN FUNDAMENTAL FREQUENCY EXTRACTOR FROM SPEECH WAVES 2 Sheets-Sheet 2 Filed Sept. 16, 195';

R m m m United Fat-eat FUNDAMENTAL FREQUENCY EXTRACTOR FROM SPEECH WAVES Meguer V. Kalfaian, 962 Hyperion Ave., Los Angeles 29, Calif.

Filed Sept. 16, 1957, Ser. No. 684,205

1 Claim. (Cl. 324-77) This invention relates to the analysis of speech sound waves and, more particularly, to systems and means for instantaneous determination of the fundamental frequency of speech waves.

The fundamental frequency of speech waves is usually referred to the repetition of wave patterns which are formed by certain combinations of frequency components having definite amplitude levels and frequency ratios, one with respect to another. In ordinary speech, howover, the frequency positions of these components in any combination for a particular sound, shift widely in the entire voice spectrum, and hence, the fundamental frequency also changes widely for each phonetic sound. In various modes of speech wave analysis, each wave pattern is analyzed separately for phoneme recognition, and accordingly, instantaneous determination of these swinging fundamental frequencies (marking points at the beginning and ending of each wave pattern) is extremely desirable.

The system described in the present invention is based upon the original disclosure in US. Patent No. 2,613,273, issued to M. Kalfaian on October 7, 1952, and filed January 23, 1947, wherein it had been shown that the fundamental components can be determined by selection of the major peaks of the complex wave. As a brief reference to a portion of the disclosure in which major peak selection is described, there is shown in the schematic of Figure (in said patent) an arrangement comprising RC elements 89, 88 (having long time constant) connected in series with rectifier 90, the combination of which receives complex wave voltages from across inductance 91. The purpose of such an arrangement is to charge the capacitor 88 unidirectionally to the level of a major peak, and eliminate the minor peaks by way of slow discharge through the large resistor 89. Thus, each time a major peak arrives the capacitor charges suddenly to the peak and produces a marker pulse indicating the arrival of a wave pattern. The description of this part of the reference disclosure may be referred to column 8, lines 15 to 39.

While the simple RC network arrangement disclosed in the above mentioned patent issue will operate theoretically, the extreme complexity of speech sound waves requires further control systems to standardize, at least partly, the amplitude of major peaks prior to selection of same. These undesirable complexities of speech wave patterns may be classified as follows: (1) enormous variation of amplitude level between vowel, semi-vowel and consonant sounds; (2) insufiicient amplitude difference between major and minor peaks; (3) random changing of the polarity of the asymmetric wave in which the major peaks are most prominent; and (4) random variation in amplitude of the major peaks, the condition of which renders the major and minor peaks often indistinguishable one from another. The first undesired condition may be remedied by the provision of a possible automatic volume control which is instantaneous in its action. The second condition may be remedied partly ice.

by unidirectional square-law amplification of the arriving sound wave; but this condition will further exaggerate the first undesired condition. The third condition may be remedied either by full-wave rectification of the phoneme sound waves or by utilizing a phase inverting instantaneous switch that is sensitive to the wave portion having larger magnitude, and select the major peaks from this phase-inverted wave portion. The fourth undesired condition, which is the most serious of all, may be remedied, at least partly, by a novel arrangement of gain suppressing saturation feedback, the function of which will be described later in the present specification. It is. wished to be noted here, however, that this novel arrange-- ment, including the entire system, is an improvement over the basic circuitry as described in the above mentioned. patent issue, and further improvements made by the same; inventor as disclosed in his later issued patents consisting; of Patent No. 2,673,893, March 30, 1954, Figures 5, 9;. and Patent No. 2,708,688, May 17, 1955, Figures 3 and 5, the specification of which may be referred to said patent issues.

As mentioned in the foregoing, a highly accurate system of major peak selection is most desirable. This is made possible in the present invention, at least substantially so,, by a system of eliminating, or at least minimizing, the undesired conditions of speech waves that have been mentioned in the foregoing. In its broader aspects, this system provides the following: Means for amplifying the speech waves unidirectionally and non-linearly, whereby partly exaggerating the amplitude of major peaks from that of minor peaks; gain controllable amplifying means for amplifying the non-linearly amplified waves in the same peak exaggerated direction; gain suppressing feedback means operatively associated with the amplifying means for feeding back these amplified waves for suppressing the gain of said amplifying means to a saturation level; and gradually decaying unidirectional storage means operatively associated with said gain suppressing means for storing a quantity of said feedback suppression waves, whereby causing suppression-storage of the gain of said amplifying means in larger magnified proportion at the beginning of each of said major peaks than at the minor peaks during gradual decaying time of said suppression-storage as an identification signal of the major peak.

From the various stages of the system as described above, it is noted that the original speech waves are first distorted non-linearly so that the differences in amplitude between major and minor peaks are partly exaggerated to a practical degree. The function of the second stage is to equalize the peak amplitude of the major peaks by way of saturation feedback. The final corrective function is provided by the storage element which causes non-linear suppression of the amplifier. For example, whenever a major peak arrives, the amplifier is at its state of maximum gain. As suppression feedback starts, the gain of the amplifier decreasesnon-linearly until the gain almost vanishes. At this point, a substantially saturated suppression signal-quantity is stored in the storage element which decays gradually until the following major peak arrives. During decaying time of the storage element, the amplitudes of the minor peaks at the output of the amplifier-means will be much less than the original, since the amplifier gain has been reduced from its previous state. Thus, a sharp rise of storage will be formed in the storage element at the beginning of each major peak, and greatly reduced storage of minor peaks in between major peaks. This particular function equalizes the amplitude of the. major peaks, and by doing so, most of the other undesirable conditions are minimized to negligible values. Actual tests have shown that this latter function also overcomes the undesirable condition of random changing of the polarity of the asymmetric wave in which the major peaks are most prominent. By the utilization of several stages of non-linear suppression feedback, and

in combination with a novel bi-stable switching circuit, which the present invention contemplates to provide, a very accurate selection of major peaks from speech waves may be secured.

With the brief description given above, my invention will be better understood from the following detailed specification and by reference to the accompanying drawings, wherein, Figure 1 is a schematic arrangement describing the principles (by way of the graphical waveforms given in Figure 2) of operation of the system in accordance with the present invention; Figure 3 is the complete schematic arrangement of the system in accordance with the invention; Figure 4 is a modification of Fig. 3; and Figure 5 is a waveform showing the actual speech wave and marker Waves coincident with the major peaks, as obtained at the output of the arrangement in Figure 3.

The character of each phonetic sound is determined by a certain combination of frequency components, as well as by their relations with regard to frequency positions and amplitude levels, one with another. These frequency components with their varying amplitudes are so related one with another that the combination produces a series of recurring major peaks spaced apart at intervals corresponding to the fundamental frequency of that particular combination of frequencies. Due to various complexities in ordinary speech, however, the amplitude of these major peaks varies tremendously, and often it becomes diificult to differentiate them from the amplitudes of minor peaks. For simplicity of illustration, a partly irregular waveform of the sound N is shown in Figure 5. In this drawing (read the time base from right to left hand of the sheet), the major peaks are designated as p1, p2, p5, p8, and marked by pulses of the square waves drawn immediately below the sound waves. The major peaks between p1 and p5 are quite distinguishable, but between p5 and p8, they start declining, and the minor peaks could be mistaken for major peaks when passed through a series connected rectifier and an RC network of short recovery time constant. Accordingly, the principal object of the present invention is to provide a circuit arrangement which is functionally capable of standardizing these amplitude changes as fast as they may occur.

In describing the principles of operation of such an arrangement, as given in Figure I, assume first that the electron tube V1 is of the commercially available type 5915, which contains two separate electron-control grids, so that its plate conductance as well as the transconduct ance may be controlled by any one of these two control grid elements. One of the control grids, for example G2, is zero biased with respect to the cathode element, and the other grid, for example G1, is biased at three volts negative with respect to the cathode element. The normal three volt negative bias renders this particular tube operative at a non-linear curve of the grid-versusplate swing with minimum transconductance. In this state of the arrangement, assume that a sine wave of three volts is applied to the grid G1, the voltage of which is amplified in the plate circuit resistance R1. This amplified voltage is rectified through rectifier device D1 and applied upon the control grid G2 in degenerative direction. During the positive excursion of the input sine wave at G1, the transconductance of the tube increases, while at the same time the rectified voltage applied upon G2 decreases the transconductance. Here it will be noted that the greatest degenerative feedback will occur at the peak of the input sine wave at G1; but at this point the bias at G2 must also be zero to achieve it, and accordingly, the voltage at G2 and R2 drops to maximum negative until the tube becomes almost non-conductive. When a storage capacitor C1 is added across R2, this negative voltage charges in C1 suddenly and decays gradually,

keeping the tube transconductance low, so that the positive swing upon grid G1 does not cause appreciable feedback Voltage from across R1, until the negative voltage at G2 has resumed close to cathode potential. The net result is that, first, during the initial degenerative feedback, a maximum negative potential is developed across C1, with greatly reduced excitation immediately after, and second, the magnitude of initial charge across C1 becomes almost identical with different amplitudes of applied sine wave voltage upon grid G1, since as stated, the amplified potential across R1 causes maximum swing of the tube transconductance even with small voltage swing upon the input grid G1.

To illustrate graphically the waveform diiferences between control grid terminals G1 and G2, Figure 2 shows oscillograms of two different operating conditions. The graph at A shows a sine wave of three volts applied upon the control grid G1, and the modified waveform at control grid G2, or capacitor C1. The graph B shows how the amplitude of the voltage waveform obtained at the control grid G2 is increased with respect to the sine wave voltage of small amplitude applied upon the control grid G1. Both the amplitudes and waveforms of the applied and derived potentials at A and B in- Figure 2 are drawn approximately proportionally, so as to show the peak exaggeration and instantaneous gain control action of the arrangement in Figure 1. The resulting waveforms as shown at A and B in Figure 2 are obtained by 60 cycle sine wave applied upon the control grid G1 of the arrangement of Figure 1. Because of this very low frequency sine wave used for the test pattern, the peaks of the waveform at G2 is shown lagging in phase with respect to the sine wave at G1. At higher frequencies, however, which will be the actual operating conditions, this phase delay will be negligible.

The graphical illustration of Figure 2 indicates that complete amplitude control is not obtainable from a single stage of the arrangement of Figure 1. Also, for complete isolation of the major peaks from minor peaks of the speech waves, it is desirable that more than a single stage of the arrangement of Figure l, with further modifications thereof, be used. Accordingly, it is contemplated herein that the modified arrangement of Figure 3 will perform satisfactorily for the selection of major peaks from complex speech waves during propagation of the waves.

Referring to Figure 3, the spoken speech wave is received by a microphone 1, the voltage variations of which are amplified in the plate circuit resistance R3 through cathode excitation of a double triode vacuum tube V2. The signal voltage developed across R3 is coupled through capacitor C3, to the control grid of vacuum tube V3, which is chosen to be of the screen grid .type. The control grid of this tube is biased to a point where the curvature of grid voltage swing versus plate current characteristics is non-linear, and the transconductance of the tube is at a minimum. In such a state, the input speech sound waves of positive excursions are non-linearly exaggerated during amplification in negative polarity in the plate circuit resistance R4. This amplified voltage is applied upon the control grid of V2 through coupling capacitor C4 and resistors R5, R6 and R7, so that its phase is inverted in the plate circuit resistance R8 of V2. At this point it is seen that the received speech waves in microphone 1 are amplified through V2, and the positive excursions of these waves are magnified non-linearly through V3, so that the major peaks of one polarity of the speech waves are partly exaggerated by square-law amplification.

These amplitude exaggerated speech waves are applied upon the control grid of vacuum tube V4 through coupling C5, so polarized that the major peak excursions tend to increase the transconductance of V4. The function of this latter tube is similar to V1, as described by way of the circuit arrangement in Figure 1, and the associated like parts consist of plate circuit resistance R9;

enthuse .5 degenerative coupling capacitor C6; rectifier diode D2; and RC parallel connected network comprising resistance R and capacitance C7, which are coupled to one of the independent electron intensity control grids of V4. Accordingly, the function of V4 is to produce and store a substantially saturated negative potential across G7 at the peak arrival of each succeeding major peak. As mentioned in the foregoing, it is desirable that more than a single stage of this degenerative circuit be utilized for complete elimination of the minor peaks. Since, however, this circuit is operated by an input signal having positive polarity, it is necessary that the stored negative potential in capacitor C7 be first inverted in polarity prior to application upon the control grid of the following stage. This is accomplished by the first section of double triode tube V5, which receives the stored signal of negative potential from C7 at its control grid, by direct cou pling, and inverts the polarity of the same signal in its plate circuit resistor R11. This phase inverted signal across R11 is applied, through coupling capacitor C8, upon one of the control grids of vacuum tube V6, which constitutes a second stage of degenerative circuit with its associated parts, comprising plate circuit resistor R12; decoupling capacitor C9; rectifier diode D3; and parallel RC network comprising resistor R13 and capacitor C10. From the foregoing description it is evident that each time the original speech waves pass through one of these degenerative stages, the major peaks are exaggerated and the minor peaks are minimized. I have found by practice that three stages of these degenerative circuits will suflice for satisfactory selection of the major peaks, although it is by no means limited to three stages, as a single stage might be found satisfactory for particular purposes, or

even more than three stages might be desirable to be utilized. The purpose herein is accordingly to provide a unique arrangement which is believed to be satisfactory for general purposes.

The third stage of above described degenerative circuit comprises vacuum tube V7; plate circuit resistor R14; decoupling capacitor C11; rectifier diode D4; and RC parallel network comprising resistor R15 and capacitor C12. As described in the paragraph supra, the negative signal potential across capacitor C10 is first phase inverted prior to application upon one of the control grids of V7. This is accomplished by direct coupling of the signal voltage across C10 to the control grid of V5, the potential of which is phase inverted in the plate circuit resistance R15 of this tube. The phase inverted signal voltage across R15 is applied, through coupling capacitor C13, upon the first of two independent control grids of V7, and by way of the degenerative coupling from anode to the second control grid of this tube. The major peaks of the original speech waves are developed across capacitor C12 at substantially constant amplitude with some residual minor peaks still being present. To further eliminate these remaining minor peaks, a trigger circuit of predetermined sensitivity may be utilized. Since the voltage of the succeeding major peaks across C12 is of approximately constant amplitude, by way of saturation .decoupling as described in the foregoing, the sensitivity of the trigger circuit may be adjusted to respond to only a part of the major peaks, so as to cancel out the remaining minor peaks.

The particular trigger circuit utilized in connection with the present invention consists of two separate trigger circuits so coupled one with respect to the other that when one trigger circuit has accomplished in changing its state of conductance, it causes the other to change its state of conductance. This particular type of crosscoupled trigger circuit is considered herein to be novel in its kind, and it comprises a first trigger circuit consisting of trigger tube V8 with two exciter tubes V9, V10, and a second trigger circuit consisting of trigger tube V11 with two exciter tubes V12, V13. The first trigger circuit consists of a double triode tube V8, the

anode elements of which are connected to resistors R17 and R18. The anode and control elements of V8 are cross coupled by coupling capacitors C14 and C15; the paths between grid to cathode elements being left floating. For the initial operation, the capacitors C14 and C15 will charge to the value of supply plate potential, for example, 250 volts as shown in the drawing, through grid current flow in each section of the double triode; the time period of this charge being short enough not to damage the tube. As these capacitors become fully charged, the potential between each of the control grid elements with respect to the cathode elements becomes zero. The anode currents flowing through resistors R17 and R18 cause voltage drop upon each one of the control grids; but since unbalance occurs in ordinary component parts, greater amount of current will flow through one of the resistors with accelerated magnification and result in conduction of only one section of the double triode. In order to reverse the state of conduction of this stabilized trigger circuit, either exciter tube V9 or exciter tube V10 is made conductive by an external signal, so that the voltage reduced in either resistor R17 or R18, by current flow therethrough, the control grid of the operating section will receive cut-oft potential and render that particular section of the triode to become non-conductive, while at the same time allowing the other section to become conductive. The second trigger circuit is a replica of the first, and it comprises a double triode trigger tube V11; anode circuit resistors R19, R20, and coupling capacitors C16, C17.

The exciter tubes V9, V10 and V12, V13 of the first and second trigger circuits, respectively, are each chosen to have two independent electron intensity control grids, for example, the commercially available type 5915, so that the anode current of each tube may be controlled by any one of these control grids. The direct cross-coupling arrangement is as follows: The second control grids of V9 and V10 are electrically connected in parallel and zero biased with respect to their cathode elements through resistor R21. The second control grids of V12 and V13 are connected in parallel and biased (through resistor R22) with respect to their cathode elements to anode current cut-off, for example, minus 20 volts, so that these tubes are normally rendered inoperative. The plate of V9 is directly connected to the plate of the left handed triode section of V8, and the plate of V10 is directly connected to the right handed triode section of V8. Similarly, the plate of V12 is directly connected to the left handed triode section of V11, and the plate of V13 is directly connected to the right handed triode section of the V11. In this manner, the state of conduction of trigger tube V8 may be controlled by exciter tube V9 or V10, and the state of conduction of trigger tube V11 may be controlled by exciter tube V12 or V13. The control grid of the left handed triode section of trigger tube V8, for example, the grid G1, is directly connected to the first grid of V12; the grid G2 of V8 is directly connected to the first grid of V13; the grid G3 of V11 is directly connected to the first grid of V10; and the grid G4 of V11 is directly connected to the first grid of V9. In this manner, it is seen that the on or off anode currents of the exciter tubes V9 and V10 of the first trigger circuit are controlled by the conductive state of V11 of the second trigger circuit, and the on or off anode currents of the exciter tubes V12 and V13 of the second trigger circuit are controlled by the conductive state of V8 of the first trigger circuit.

In operation, assume first that the right handed triode section of trigger tube V11 is normally conducting. This means that the bias voltage from grid G4 to the cathode element is zero, and the bias voltage from G3 to the cathode element is highly negative. Since G4 is directly connected to the first control grid of V9, and G3 is directly connected to the first grid of V10, it is evident that V9 has become conductive and V10 non-conductive. Similarly, since V9 draws current through R17, the high negative voltage developed across thi resistor is applied upon the control grid G2, and consequently, the left handed triode section of trigger tube V8 becomes conductive while the right handed triode section of V8 becomes nonconductive. Now assume that the input triggering signal is applied upon the second control grids of exciter tubes V9, V10 in negative polarity, and the same signal is applied upon the second control grids of exciter tubes V12, V13 in positive polarity. As this input signal is applied, the exciter tubes V9 and V10 are secured to non-conductive states, while the positive signal applied upon exciter tubes V12 and V13 drives them to conductive states. At this point, it is seen that the first control grid of exciter tube V13 had already been driven highly negative by direct connection to the control grid G2 of trigger tube V8, so that the incoming positive signal upon its second control grid cannot render it operative, while the exciter tube V12 becomes conductive and draws current through R19. This action renders the left handed triode section of trigger tube V11 non-conductive and the right handed triode section of this tube conductive, with no further action upon the first trigger circuit, since at this point the exciter tubes V9 and V10 are both non-conductive. When the input triggering signal subsides to approximately zero value, the bias voltage upon the second control grids of exciter tubes V9 and V10 become zero with respect to their cathode elements, and since at this time the first control grid of exciter tube V10 receives zero bias from the control grid of G3, it becomes conductive and draws current in series with R18pchanging the state of conductance of trigger tube V8. Thus, it is seen that each time the input triggering signal arrives in positive polarity upon the second control grids of exciter tubes V12, V13, and in negative polarity upon the exciter tubes V9, V10, the trigger tube V11 changes its state of conductance, and when the input signal subsides, the trigger tube V8 changes its state of conductance.

The input triggering signal just mentioned is produced by phase inverting action in the right handed triode section of V14, for example, by producing thi signal in positive polarity across cathode circuit resistor R23, and in negative polarity across plate circuit resistor R24. Finally, this signal voltage is applied upon the second control grids of V9, V10 through coupling capacitor C18, and to the second control grids of V12, V13 through the coupling capacitor C19. The left handed triode section of V14 is utilized to amplify and phase invert the negative major peak signal across capacitor C12, by way of its plate circuit resistor R25 and coupling capacitor C20 to the control grid of the right handed triode section of V14.

In the prototype model of the arrangement given in Fig. 3, I have found that certain modifications will improve the over-all performance of major peak selection. For example, it had been stated in the foregoing that in ordinary speech the amplitude of different phonetic sounds varies tremendously. This variation may be partly controlled by a system of mechanical relays for switching in and out different values of impedances. For example, the amplitude of speech sound waves applied to the control grid of the right handed triode section of V2 may be controlled in three steps by short 'circuiting action of the resistors R5, R6 and R7. In its operation, assume first that the resistors R and R6 are normally short circuited by the closed mechanical relay contacts 2, 3, and 4, 5 of idle relays 6 and 7, respectively. In thi state, the signal voltage coupled by capacitance C4 to the control grid of the right handed triode section of V2 is maximum. Whereas, when the contact points 4 and 5 of relay 7 are opened, the amplitude of the original signal upon said grid of V2 is reduced by one step. Further, when the contact points upon said grid of V2 is reduced by another step. Thus, the wide amplitude variation of the original speech sound waves is reduced into a narrow band of stepwise variation.

The operation of relays 6 and 7 are effected by the double triode tubes V15 and V16. The sound wave signal across resistor R4 is applied simultaneously upon the control grids of the left handed triode sections of V15 and V16 through coupling capacitor C4. These last mentioned triode sections are utilized mainly to amplify the incoming signal at their plate circuit resistors R26 and R27, respectively. The signal voltage across R26 is coupled to the RC network comprising resistor R28 and capacitor C21, through coupling capacitor C22 and rectifier diode D5. The unidirectionally rectified signal voltage across C21 is applied upon the right handed triode section of V15 in positive polarity, so that whenever a large negative voltage is developed across C21 and the control grid of the right handed triode section of V15, this triode section becomes conductive from a normal non-conductive state and energizes relay 6; closing the relay contact points 2 and 3. Similarly, the signal voltage across R27 is coupled to the RC network comprising resistor R29 and capacitor C23, through coupling capacitor C24 and rectifier diode D6. The unidirectional rectified signal voltage across C23 is applied upon the right handed triode section of V16 in positive polarity, so that whenever a large positive voltage is developed across C23 and the control grid of the right handed triode section of V16, this triode section becomes conductive from its normal nonconductive state and energizes relay 7; closing the relay contact points 4 and 5. The operating sensitivities of relays 6 and 7 may be adjusted 'by either the values of resistors R26 and R27, or cathode circuit resistors R30, R31 and R32, R33 of vacuum tube V15 and V16, so that one relay will operate with larger incoming signal than the other for stepwise control of short circuiting the resistors R5 and R6. The large negative bias normally applied upon the control grids of the right handed triode section of V15 and V16 render them non-conductive, so that the relays 6 and 7 normally remain in idle states. If desired, the associated component parts of relays 6 and 7 may remain identical, and the last mentioned negative bias may be adjusted in stepwise values for said stepwise operation of the relays.

In order that the skilled in the art may be able to build the device, as given in Fig. 3, the important values of certain component parts are listed as below:

The component part values as given above are not critical, and they may have wide tolerances. The plateto-ground bypass capacitor C25 is used to avoid noise being transmitted to the following stages, as the system is very sensitive and will respond to noise in the same manner as voice signals. Due to the extreme sensitivity of the arrangement of Fig. 3, it may be necessary to .2 and 3 of relay 6 .are also opened, the signal ypltage ,7? utilize uvell regulated supply voltages, or, the critical stages may be individually filtered out, for example, by resistors R37--R45 in conjunction with filter capacitors as shown in the drawing. The screen grid potentials of V4, V6 and V7 are adjusted to approximately 75 volts by a gaseous regulator tube V17, which is not absolutely essential and it may be dispensed with.

With regard to the particular type of trigger circuit utilized in conjunction with the arrangement of Fig. 3, it will be noted that the control grid elements of trigger tubes V8 and V11 are floating without termination to their cathode elements. The object for this type of arrangement is to avoid discharge of the coupling capacitors C14 to C17. The very small inherent capacity elements between grid to cathode elements, however, will cause gradual discharge of capacitors C14C17 and set the trigger circuit into oscillation. The frequency rate of this oscillation depends upon the ratio of capacitance values between C14C17 and the internal capacitances of the vacuum tubes V8, V11. Since this ratio is very large, the oscillation (abrupt change in state of conduction) will take place once in several minutes when in idle state; but this will be avoided when the device is functionally active by the voice signals. Of course, other types of trigger circuits may be employed if so desired.

As stated in the foregoing, the circuit arrangement of Fig. 3, and the particular values of the component parts, are given herein for the purpose that the skilled in the art of electronics may be able to build it without encountering diificulty in its operation. It is also wished here to be stated that the arrangement of Fig. 3 has been built in different forms with equally good performance. For example, the minor peaks of the voice signal wave may be first partly eliminated prior to application of the Wave uponthe control grid of the first stage degenerative circuit, for example, the stage comprising vacuum tube V4. Fig. 4 shows a simple schematic arrangement for minor peak reduction, wherein the capacitor C28 charges quickly to an incoming major peak in positive direction, through diode D7, and discharges slowly through resistor R46 during the idle period until the following major peak arrives. Thus, the minor peaks having low amplitude during said idle period will be completely eliminated.

In another modification, the incoming voice signals may be first full wave rectified prior to application upon the control grid of V4, so that the random changing of the polarity of the asymmetric wave in which the major peaks are most prominent may be stabilized prior to major peak selection. Such full wave rectification of signal waves is generally employed in conventional radio; television; and various other electronic devices, and accordingly, further schematic drawings and descriptive matter thereof are not deemed necessary to be given herein.

As described in the foregoing, the circuit arrangement of Fig. l is capable of equalizing the varying amplitudes of the major peaks. Accordingly, it is not absolutely essential to utilize the mechanical relays 6 or 7, or both; and if so desired, more than two relays may be employed in conjunction with the device so as to obtain greater number of amplitude controlling steps.

In order to provide graphical illustration of the operating conditions of the arrangement of Fig. 3, the draw ing of Fig. shows the speech sound waves at A and the output marker waves at B coincident with the major peaks, as obtained from the anode circuit of any one of the trigger circuits consisting of either V8 or V11.

While in the foregoing I have described the principles of my invention in connection with specific apparatus, I wish it to be understood that this description and the accompanying drawings thereof are made by way of limited examples only, as it will be obvious to the skilled in the art of electronics that various substitutions of parts, adaptations and modifications are possible without departing from the spirit and scope thereof. I also wish it to be understood that the schematic arrangement of Fig. 3 is not particularly limited to the use of vacuum tubes as shown in the drawing, as transistors will also provide the functions pertinent to the particular system disclosed herein. Of course, the physical characteristics of vacuum tubes and transistors are recognized to be different one from the other, for example, their inherent differences in impedance values, and accordingly, the particular component values provided herein, for reference purposes, will not be applicable to the circuitry for transistor operation. And further, due to said differences in their operating properties, the circuitry of the arrangement in Fig. 3 will require revision. As stated, however, the basic functions will be the same whether vacuum tubes or transistors are utilized in conjunction with the present invention. Accordingly, the limitations of this invention will be defined only by the appended claim.

Iclaim:

A complex wave analyzing apparatus for producing marker pulses at major peaks of an applied input complex wave, comprising a gain-controllable amplifying means of the electron discharge type having an electron emitter, electron collector, and at least first and second intensity control elements each capable of controlling the intensity of electron current passing from emitter to collector; a first resistor element in series with the collector element; means for producing said complex wave and applying same to said first intensity control element for amplifying same across said resistor element; a parallel connected capacitor and second resistor network across said second intensity control element and said emitter element, said network having a decaying time constant equal to or longer than the average time period occurring between said major peaks; means for coupling unidirectionally the amplified complex wave across said first resistor to said capacitor for storing in the capacitor gain-suppression electrical quantities, thereby causing a sudden storage of gain-suppression electrical quantity in large magnified proportion at the beginning of each of said major peaks of the amplified complex wave; and means for deriving marker pulses from said sudden storage of electrical quantities.

References Cited in the file of this patent UNITED' STATES PATENTS 2,361,648 Petty Oct. 31, 1944 2,383,571 Shook Aug. 28, 1945 2,401,214 Worcester May 28, 1946 2,468,687 Schmitt Apr. 26, 1949 2,489,126 Fay Nov. 22, 1949 2,593,694 Peterson Apr. 22, 1952 2,613,273 Kalfaian Oct. 7, 1952 A 2,662,125 Stafford Dec. 8, 1953 2,673,893 Kalfaian Mar. 30, 1954 2,756,328 Braak July 24, 1956 2,832,931 Crain Apr. 29, 1958 2,857,481 Philips Oct. 21, 1958

Images