US3418429A

US3418429A - Speech analysis system

Info

Publication number: US3418429A
Application number: US495427A
Authority: US
Inventors: Genung L Clapper
Original assignee: International Business Machines Corp
Current assignee: International Business Machines Corp
Priority date: 1965-10-13
Filing date: 1965-10-13
Publication date: 1968-12-24
Anticipated expiration: 1985-12-24
Also published as: CH452913A; DE1547030B2; FR1494395A; NL148731B; BE687178A; DE1547030A1; ES331289A1; GB1160593A; NL6614325A; SE321958B

Abstract

1,160,593. Speech recognition. INTERNATIONAL BUSINESS MACHINES CORP. 15 Sept., 1966 [13 Oct., 1965], No. 41169/66. Heading G4R. [Also in Divisions H3 and H4] In a speech analysis system, gating of coded signals (representing speech characteristics) into a store is inhibited on detection of noise. A speech waveform from a microphone 1 is amplified at 3 and split into frequency bands at 60 to feed formant location means 129 and automatic gain control 66. The latter produces a signal in response to the largest of the frequency band amplitudes to control the gain of amplifier 3, feed noise clamp generator 90 and start multivibrator 160. The multivibrator 160 then drives a ring 199 to read the outputs of the formant location means 129 into a storage matrix 150. The multivibrator 160 is kept operating after the start signal from the automatic gain control 66 disappears by a signal on line 201 from ring 199. The noise clamp generator 90 differentiates the input from the automatic gain control 66 to detect noise. While noise is present, generator 90 inhibits multivibrator 160. A potentiometer 179 allows the frequency and width of the pulses from the multivibrator 160 to be controlled to compensate for differences between different speakers. The stages of the ring 199 have indicator lamps e.g. 243, as has the automatic gain control 66 at 73.

Description

1968 cs. L. CLAPPER 3,418,429,

SPEECH ANALYSIS SYSTEM Filed Oct. 13, 1965 4 Sheets-Sheet 2 SPEECH AMPLIFIER 1a 17 SENSITIVITY CONTROL AUTOMATIC GAIN CONTROL 66 Dec. 24, 1968 e. L. CLAPPER 3,413,429

SPEECH ANALYSIS SYSTEM Filed Oct. 13. 1965 4 Sheets-Sheet 5 NOISE CLAMP GENERATOR 90\ 100 SENSITIVITY CONTROL 160 VOICE OPERATED GATED MULTIVIBRATOR Dec. 24, 1968 C ER 3,418,429

SPEECH ANALYSIS SYSTEM Filed Oct. 13. 1965 4 Sheets-Sheet 4 FIG. 6

R6 OPEN RING United States Patent 3,418,429 H SPEECH ANALYSIS SYSTEM Genung L. Clapper, Vestal, N.Y., assignor to International Business Machines Corporation, Armonk, N.Y., a corporation of New York Filed Oct. 13, 1965, Ser. No. 495,427 7 Claims. (Cl. 179-1) ABSTRACT OF THE DISCLOSURE Speech data is stored without noise by distinguishing the noise from the speech signals and operating the storage matrix by a timing signal which is generated only in the absence of noise.

The invention relates to a voice operated speech amplifying system and more particularly to a speech analyzing system which rejects noise to provide speech patterns consisting primarily of speech characteristics free of unwanted noise characteristics.

In general, speech analyzing systems are of the type in which operation is initiated by manual control, specifically by manipulation of a switch, push button, or the like, by an operator. With this type of initialization control, unwanted frequencies, mainly noise frequencies are passed into the system prior to and after voice signals because of delays between the time the system is operative and the time the voice frequencies are generated.

Yet in other systems in which control is initiated by sound frequencies the presence of a noise burst can generate undesirable characteristics which will be stored along with significant sound characteristics.

The present invention is, therefore, directed to a speech analysis system which is activated only when desirable sound characteristics are generated, and deactivated when unwanted noise frequencies arise during the generation of the voice frequencies. This is achieved by the utilization of a noise clamp generator which gates a voice operated multivibrator that coordinates the timing and sampling of the voice signals entering the system.

It is the primary object of the invention to provide a voice operated speech analyzing system which is responsive to the desirable portions of the speech spectrum and provides a spectrum of speech characteristics free of undesirable noise characteristics,

Another object of the invention resides in the provision of a speech analyzing system in which manifestations representing voice frequencies are sampled under control of a noise clam-p generator during intervals which are free of undesirable frequencies thereby providing noisefree speech signal characteristics.

Still another object resides in a speech analyzing system having an extremely fast and highly responsive gain control means which continually monitors the magnitude of the voice frequencies to provide a controlled output voltage and alerts the system to the presence of unwanted noise signals, whereby the latter are rejected, thereby enabling only true and meaningful speech sound characteristics to be recorded.

The foregoing and other objects, features, and advantages of the invention will be apparent from the following more particular description of a preferred embodiment of the invention, as illustrated in the accompanying drawings.

In the drawings:

FIG. 1 is a block diagram showing the cooperative re- 'lationship among the various components constituting the invention.

FIG. 2 shows the details of the speech amplifier.

FIG. 3 shows details for the automatic gain control circuit.

FIG. 4 shows the detailed circuitry for the noise clamp generator.

FIG. 5 shows the detailed circuitry for the voice operated gated multivibrator.

FIG. 6 shows a detail of a portion of the ring drive.

FIG. 7 shows a detailed circuit for a portion of the open ring.

As a preliminary to the explanation of the present invention, a brief description will now 'be given of the speech analyzing system described in a pending application Ser. No. 474,230, filed by the same inventor. In this system (Ser. No. 474,230) speech sounds are fed into the system by way of a microphone. These sounds are analyzed by means of a frequency analyzing system containing a plurality of frequency selectors each tuned to a particular band of frequencies lying in a range extending from 3,750 c.p.s. to 260 c.p.s. By virtue of these frequency selectors, formants, present in the voice spectrum, are detected by a formant location system. The presence of these formants are in turn transmitted to a formant transition detection system and signal characteristics representative of these transitions are developed and stored in a storage matrix. These formant characteristics or transition characteristics are either of a falling or rising type, or, in the absence of both, a steady state condition exists for which an invariant characteristic is established. The system is also provided with an automatic gain control means which communicates with a pre-amplifier to keep the voltage gain at substantially a consistent output level. Additionally, the ouput of the gain control means is passed through a slope detector which scans the output of the gain control waveform for the presence of sharp negative transients which are indicative of sudden bursts in voice intensity These burst signals are then utilized for developing meaningful burst characteristics forming a part of the speech spectrum.

In the present invention, as shown in FIG. 1, speech sounds entering the system by way of a microphone 1 are transformed from sound energy to electrical energy and passed by way of line 2 into a speech amplifier 3. The latter issues a complex speech waveform along an output line 48 to frequency analyzer 60. The speech waveform is maintained at a constant amplitude by virtue of a fast operating automatic gain control means 66 working in conjunction with a noise clamp generator 90. The frequency analyzers 60 having adjustable band width control means perform a frequency analysis upon the waveform and provide appropriate output signals on lines 60a-60g communicating with a formant location means 129. This means employs comparison hysteresis and differential logic means, as fully described in a pending application, Ser. No. 291,344, filed June 28, 1963, to provide local maxima indications to a storage matrix by Way of lines M1 through M7 under control of a gating means 129a through 129g, and a noise clamp gate signal issued over a line 124a connected to the output of the noise clamp generator 90. The automatic gain control means 66 which constantly monitors the amplitude of the frequencies present in the complex waveform provides the resultant automatic gain control voltage also to a voice controlled gated multivibrator by way of

lines

84 and 84b. Following a reset action by reset means 200, the operation of the voice controlled multivibrator 160 is initiated by speech sounds entering the system and generating positive signals on input line 84b; thereafter, the multivibrator action continues by virtue of the presence of a positive gate signal on the if line 201. The multivibrator operation is subject, however, to the presence of a negative gate signal on the line 124, which gate signal is indicative of the absence of unwanted noise signals in the complex speech waveform. The presence of an unwanted 3. noise signal in the complex waxeform is transmitted through the automatic gain control and into the noise clamp generator which develops a positive noise clamp signal on the line 124 thereby providing an inhibiting infiuence on the operations of the voice controlled gated multivibrator. This inhibiting action terminates the action of the multivibrator and, as a consequence, a ring drive means 190 which controls, by way of line 195, the advancement of an open ring means 199 is disabled for the interval that the unwanted signal appears in the complex waveform. The open ring 199 is reset to an R position (the first stage of the ring) under control of appropriate reset means 200 to provide a state of readiness for storing of the speech characteristics. As soon as speech sounds enter the system, the automatic gain control signal line 84 transmits a positive signal through the line 84b to initiate operation of the multivibrator 160. The latter, accordingly, transmits the timing signals to actuate the ring drive means 190 which, by way of line 195, advances the open ring from the first stage Rqb, to the next stage, R1. From this stage R1 and up until the final ring position is encountered, the E signal is positive to maintain the multivibrator under continuous operation. The presence of an unwanted noise signal, however, causes the voltage on line 124 to drop to a negative level to terminate multivibrator operations. The periodicity of the unwanted signals is considerably finer than the periodicity of the voice which is approximately 7 ms. in duration. Since no integration is used in the automatic gain control means 66, the response is limited only to the response of the frequency selector means 60. A sudden rise in the output of any one of the frequency analyzing lines 61a-61g, results in a substantially immediate response from the gain control means 66 to cause the latter to issue, by way of line 84, a regulating voltage to the speech amplifier means 3. This action is so prompt that the automatic gain control means 66 provides an immediate compensatory voltage for the normal drop off to the damped wave trains within the periodicity of the speech waveform. By virtue of this fast action, the :presence of a noise spike immediately, within a matter of a 2 volt rise above the 4 volt level, initiates the generation of a noise clamp signal which is well ahead of the time that a voltage level, approximately 0.7 volt above ground is reached whereat multivibrator action commences. Thus, clamping action precedes multivibrator action to prevent manifestations, representing noise characteristics, from entering the storage matrix, and permit ting entry of only the meaningful noise free sound characteristics. Waveforms generated by explosive speech sounds containing harsh frequencies which approach the slope of noise spikes are rejected by the action of the capacitor 91 in conjunction with the diode network 92.

Among the various components constituting the invention, the principal ones will now be described in detail.

Speech amplifier The speech amplifier comprises essentially four

transistors

7, 15, 38 and 45 interconnected in the circuit network shown in FIG. 2. Transistor 7 has its base connected to line 2 by way of a capacitor 4 and a 1K resistor 6. The emitter thereof is connected to the line 2 and the collector is connected to a 12 volt supply by way of a line including resistors 9 and 11 each of 4.7K ohms. The base and collector are interconnected by way of a line 5 which includes an 18K ohm resistor. Transistor 15 is connected to a circuit network which includes a diode 16, a resistor 17 and a capacitor 18. This network is connected to a sensitivity control means 21 by way of a 1K resistor 20. The base of transistor 15 is coupled to the collector of transistor 7 by means of capacitor 12. The collector of transistor 15 is coupled to the collector of transistor 7 by way of a capacitor 12 in series with a 47K ohm resistor 14. The collector is further connected to a l2 voltage supply by way of a line 22 which includes resistors 23 and 24 the midpoint of which is connected to ground by way of capacitor 25. Transistor 38 is connected to the collector of transistor 15 by way of a resistive capacitative network which includes line 26, resistor 27,

capacitors

29 and 31, capacitor 33, and resistor 34 and line 32 which is connected to the base of transistor 38. The base also has an output level control means which includes a line 35, resistor 36, and a potentiometer 37 connected between ground and 12 volts. The emitter of transistor 38 is connected to a line 84a by way of a resistor 40 and also to a diode 39 connected to ground. By virtue of the diode connection, a constant current is fed through the emitter even though variations appear in the current passing through the line 84a. This constant emitter current is achieved by virtue of the diode 39 which provides a compensating current while the current in the line 84a varies. Transistor 45 has its collector connected to ground by way of a line 46 and its emitter is connected to a -12 volt source by way of a resistor 47. The base of transistor 45 is connected to the transistor 38 by way of line 42. Transistor 45 supplies to the frequency selectors 60 an output whose amplitude is maintained constant by the automatic gain control by way of line 48.

The function of transistor 7 serves as the input preamplifier for the incoming speech signals entered by way of the mircrophone. Transistor 15 serves the function of gain control. The setting of a potentiometer 21 determines current fiow through resistor 17 connected to the emitter of transistor 15. With large current flowing in this resistor, equal in amount to the emitter current, substantially no current flows through diode 16 this being due to the high resistance, in the order of several hundred K-ohms, of the diode 16. Conversely, if the potentiometer is adjusted so that less current flows through resistor 17, a proportionately greater current flows through diode 16 because the impedance of the diode under this condition is correspondingly smaller. Thus, the greater the current flow through diode 16, the smaller is its resistance. Since the gain of transistor 15 is a function of emitter impedance, it is then possible to control the gain of this stage by means of the potentiometer setting 21.

The coupling network, between

transistors

15 and 38, serves to compensate for the normal fall off of the high frequencies in the voice waveform. The potentiometer 37 serves as a level setter for the final output from transistor 45 by adjusting the base current to transistor 38. This final output is maintained at approximately 4.5 volts, average level. The transistor 38 has the function of automatic gain control for the speech amplifier. The variations in the automatic gain control voltage applied to resistor 40 causes current variations through the latter in much the same Way that different potentiometer voltages cause corresponding current variations through the resistor 17. For example, at an automatic gain control voltage of 5 volts, current flow through diode 39 represents the current flow through the emitter and the current flow through resistor 40. Hence, at this time, the diode impedance is at a minimum and the gain of the stage is maximum. The condition at ground potential, however, is such that the current flow through diode 39 is just the emitter current with substantially no current flow through the resistor 40 thereby producing an intermediate gain for the stage. Between ground potential and positive voltage the current flowing through resistor 40 is in opposition to the constant emitter current and thus provides a proportionately smaller current flow through the diode 39, thus decreasing the gain of the stage. Transistor 45 serves as an output means that supplies to the frequency selectors 60, by way of line 48, an amplitude controlled speech signal.

Automatic gain control The automatic gain control means 66, shown in FIG. 3, comprises

transistors

67, 71 and connected in the manner shown. The transistor 67 has its collector grounded and the emitter thereof connected to line 68 which terminates at a 12 volt supply. Included in this line is a potentiometer 69 by which the operating voltage level is established at 6 volts and communicated to the base of transistor 71, the latter having its collector connected to an automatic gain control indicator 73 by way of line 72, and to a grounded resistor 70. The purpose of the latter is to provide a pre-energization path for the automatic gain control means 73 in order to aflord a greater degree of sensitivity when the transistor 71 becomes conductive. The emitter of transistor 71 is connected to a --12 volt supply by way of line 74, resistor 75, line 77, and resistor 78. The transistor 89 has its emitter also connected to the l2 volt supply by way of resistor 76, line 77, and resistor 78. The collector thereof is connected to a +12 volt supply by way of line 81 and resistor 82. The base of transistor 80 is connected to a 6 volt supply by way of line 83. The controlled output voltage appears on the collector output line 84 and is connected to the noise clamp generator 90 by way of line 84a, and to the voiceoperated gated multivibrator 160 by way of line 8417. In addition, the controlled output is communicated to the Speech amplifier 3 by way of line 84.

Inputs to the automatic gain control are derived from output lines 61a-61g connected to lines a-60g extending from the frequency analyzers 60. These lines are mixed by appropriate diodes 62 and terminate at a comrnon line 63 connecting the base of transistor 67. Any one of these mixed lines which transmits the highest voltage at any particular instant controls the speech amplifier gain through the automatic gain control voltage. The operating voltage level is compared with a fixed reference, 6 volts, and the positive difference provides an automatic gain control indication as well as a controlled voltage output. On the other hand should the inputs to the transistor 67 fall below the fixed reference level, the automatic gain control indicator cuts off as the transistor 67 drops the potential at the base of transistor 71 below the reference level. This causes an increase in the current to transistor which results in a voltage drop in the automatic gain control output. in turn this causes an increase in the gain to the speech amplifier 3 and causes a larger output to be transmitted from the rectifiers, not shown, constituting a portion of the frequency selector means, 60. Since the automatic gain control means utilizes on integration means, the response is thus limited to the rectifier response.

Noise clamp generator The noise clamp generator shown in FIG. 4 comprises essentially four

transistors

96, 98, 102, and 113 interconnected in the network shown. The output from the automatic gain control means 66 is applied to the base of transistor 96 by way of

lines

84a and 95. Interconnected between these lines are a capacitor 91 and low level noise filter 92 consisting of a pair of

diodes

92a and 92b. This network has a ground connection by way of line 93 and resistor 94. Transistor 102 is connected to transistor 96 by way of line 97 which includes a pair of resistors 97a and 971) by virture of which the emitters are interconnected. The collectors of these transistors are both connected to a l2 volt supply by means of lines 103 and including

resistors

104 and 106 respectively. Transistor 98 has its collector connected to the line 97, its base connected to a +6 volt supply by way of a line 99 and its emitter connected to a sensitivity control 101 by way of a line including a resistor 100. Transistor 113 has its base interconnected to the

transistors

96 and 102 by way of

circuit paths

107 and 109, respectively, containing

diodes

108 and 110 respectively. These two circuit paths connect at a common point 111 which is in turn connected to the collector of transistor 113 by way of a circuit path 116 which includes resistors and 117 The base of transistor 102 is connected to ground by a circuit path 120 which includes a resistive capacitative network containing resistor 119a and capacitor 11%. The terminal 111 is connected to capacitor 118 one side of which is connected to ground. The output of the noise clamp generator is fed from the collector of transistor 113 by way of line 114 into the base of transistor 122 whose collector is connected to a +6 volt supply and the emitter, to a 6 volt supply by way of resistor 123. The output of the transistor 122 is fed to a line 124 connected to a drive 124a in turn connected to the transistor 164 of the voice operated gated multivibrator.

In operation the automatic gain control waveform passed into the input network consisting of capacitor 91,

diodes

92a and 92b, and resistor 94 is differentiated and then amplified by means of

transistors

96 and 102. A rising waveform produces a negative going signal from transistor 96. A falling waveform causes a similar output from the transistor 102. During this action transistor 98 controls the current in the system so that the outputs are normally above 6 volts. During the transmission of the negative going signals from

transistors

96 and 102 to the transistor 113, by way of

lines

107 and 109, the transistor 113 is held nonconductive by virtue of the negative going signals. This results in an output of +6 volts from the collector which is connected through line 124 to transistor 122 which controls line 124, the input to the voice operated gated multivibrator 160. In addition, a branch path 124a also conveys this output voltage to the gating means 129a129g located in the formant location means 129.

Voice operated gated multivibrator The voice operated gated multivibrator 160, shown in FIG. 5, comprises

transistors

162, 164, 174, 175 and 183 connected in the manner shown.

Transistors

162 and 164 have their emitters connected in common to a grounded line 163. The collectors thereof are connected to a 12 volt supply by way of lines 166 and respectively. The collector of transistor 162 is coupled to the base of transistor 164 by way of a line 167 containing a resistor 168. The input to the base of transistor 162 is applied, by way of an OR circuit 159, to which are connected lines 84b by way of a diode 15912 and the R line 201 by way of a diode 159a. The cathodes of these diodes are connected in common through a resistor 158 to a 12 volt supply. This input network to the transistor 162 constitutes a positive OR circuit. The input to transistor 164 is applied by way of line 124 through a diode 161. The presence of a noise signal in the waveform applied to the input of the system causes a noise clamp signal of +6 volts to be transmitted through this line 124 into the base of transistor 164. The collector output from transistor 164 is coupled to the base of transistor 174 by way of a line 170 and resistor 171. A capacitative coupling exists from the line 17 0 to the collector of transistor by way of a line 173 which includes a capacitor 172. The collectors of

transistors

174 and 175 are connected to a +6 volt supply respectively by way of lines and 181. The emitters of these transistors are connected to a 6 volt supply. The collector of transistor 174 is coupled to the base of transistor 183 by way of line 176 including a resistor 182. Also, the base of transistor 175 is capacitatively coupled to the base of transistor 183 by Way of a capacitor 177. Both the base of the transistor 175 and the capacitor 177 are commonly connected to a line 178 in turn connected to a speed control means 179. The transistor 183 has a grounded emitter and the collector thereof connected to a l2 volt supply by way of line 185. The output of the transistor 183 extends from the line 185 to an output line 186.

In operation, the transistor 162 is cut ofi when either the R line 201 or the automatic gain control input line 84b are above ground potential. The absence of a noise clamp signal on line 124 and the drop at the collector of transistor 162 enables the transistor 164 to conduct thereby raising the voltage level at the gating resistor 168 from ---12 volts to ground. As a consequence, transistor 174 is driven to conduction following termination of a 20 millisecond delay interval occasioned by the charging of capacitor 172. In turn, transistor 175 is cut off to cause transistor 183 to conduct to produce an output signal on the line 186. Multivibrator action is controlled between the

transistors

174 and 175 as each alternately conducts. This action continues as long as transistor 164 holds the gate on. Upon the application of the noise clamp pulse to the base of transistor 164 the latter cuts off and turns the gate off during the noise clamp pulse. Upon termination of the latter, multivibrator action is resumed.

Ring drive The ring drive 190, shown in FIG. 6, comprises

transistors

213, 216 and 224 connected in the network shown. Input timing signals enter the ring drive by way of a capacitor 210, line 212 to the base of transistor 213 whose emitter is connected to a 6 volt supply, and the collector to +6 volts by way of a resistor 217. Transistor 216 has its base connected to the collector of transistor 213 by way of a coupling capacitor 219 and also to ground by way of a 10K resistor 220. The collector output of transistor 213 is fed into the base of transistor 224 by way of a line 221 and a resistor 222. A 12 volt supply is connected to the base of transistor 224 by way of resistor 223. The emitter of transistor 224 is connected to a +6 volt supply and the collector, a 6 volt supply. The collector output is passed on to line 225 which is connected to the open ring drive line 195.

In operation, timing signals issued by the voice controlled gated multivibrator 160 enter the ring drive means 190 by way of the capacitor 210. Transistors 213 and 216 in conjunction with the capacitor 219 and resistor 220 function as a pulse generator to produce pulses of definite length, the length being dependent upon the time constant of the resistor 220 and the capacitor 219. The purpose of the transistor 224 is to provide drive as well as a shift in the pulse level. The output pulse from the instant circuit varies between 6 volts and +6 volts and has a pulse period of approximately 130 ,uS.

Open ring The open ring 199 comprises a plurality of stages R through R8 each constituted of a pair of transistor triggers. As seen in FIG. 7, only three stages are shown, namely: the first, second and last stages. The first stage R comprises

transistors

233 and 238. The ring drive line 195 is connected to the collector of transistor 233 by way of a network including a diode 230 and a resistance capacitance network 231 consisting of resistor 231k and capacitor 231a. The base of transistor 238 is connected to the resistance capacitance network by way of a line 239. The collector of transistor 233 is connected to a +6 volt supply through a resistor 233a and also to an inter-stage coupling line 244 which is connected to the second stage R1. The emitter of transistor 233 is connected to a -6 volt supply, which voltage is also applied to the base of the transistor by way of a resistor path 234. The base of the transistor 233 is also connected to the collector of transistor 238 by way of a path 235 containing

resistors

235a and 235b at the midpoint of which is connected a reset line 237 having a capacitor 236 interposed therein. The output from the second transistor 238 is passed on to an output line 241 connected to the storage matrix. The output line 241 is also connected to a line 242 which includes an indicator 243 connected to a 12 volt supply.

When the ring is off, both transistors in each stage are off, i.e., non-conducting. To initiate operation of the ring a positive reset signal is transmitted from the reset means 200 by way of reset line 237, through the capacitor 236 to cause transistor 233 to conduct. Conduction in the latter causes a voltage drop that is passed through the network 231 to the base of transistor 238 to cause the latter to conduct. Thus, with both

transistors

233 and 238 conducting, the first stage R is now in its on state. In

this state output line 242 is slightly below ground potential and causes the indicator 243 to turn on. When the ring drive signal is transmitted through the line 195, which is connected in common to all stages of the ring, only one stage at a time becomes responsive. In this instance with stage R being on and the next'su-cceeding stage R1 off, the ring drive signal causes transistor 238 to turn off which action causes transistor 233 to turn off. The turning off of transistor 233 causes a positive signal (complement) to be transmitted through interstage coupling line 244 to the base of transistor 250 of the second ring stage R1. Transistor 255 is maintained in a nonconducting state even though transistor 250 is on. This condition prevails until the right drive pulse terminates whereupon transistor 255 is driven into conduction. The second ring stage is now in its on state while stage 1, R 5 and the remaining stages are off. The next ring drive signal transmitted through the line 195 causes the second stage, R1, to go off and stage 3, not shown, to turn on. In this manner, each succeeding stage will be turned on while all the remaining stages are in their ofi state.

Having described the details of the important com ponents constituting the invention, it may be appreciated that the system, following a restart action by means of he reset means 200, is initiated into action by the entry of speech sounds into the system by way of the microphone 1. The speech sounds are converted into appropriate waveforms and, among other things, develop appropriate voltage on the

lines

84 and 84b to maintain the action of the multivibrator which provides the timing signals to the ring drive in turn controlling the advancement of the open ring 199. This activity is permitted only in the absence of noise spikes arising by mechanical means other than the normal speech sound originating means. Moreover, the action of the diode network 92 in conjunction with the differentiating action of the capacitor 91 prevents noise transients, arising from harshness in the speech sounds, from entering the storage matrix 150.

In addition, the novel emitter control for the

transistors

15 and 38 in the speech amplifier 3 provide for a highly sensitive gain control which is relatively linear and free of distortion over a wide range. This control also provides a high speed action by virtue of the absence of the time constants associated with the base capacitor since the base current remains substantially fixed.

The system is further provided with speed control means 179 which controls the speed and duration of timing signals issued by the multivibrator 160. This feature, together with the action of the RC network in the ring drive 190 which controls the pulse width of the ring drive signals, provides a highly flexible means for adjusting the system to accommodate variations, however wide, in speech output by different individuals.

While the invention has been particularly shown and described with reference to a preferred embodiment thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.

What is claimed is:

1. A voice controlled speech analysis system comprising:

a plurality of frequency analyzers responsive to voice frequencies of the speech spectrum to provide appropriate output signals;

formant energy detecting means connected to said frequency analyzers and providing formant signal manifestations indicative of speech frequencies present;

timing signals generating means responsive to said appropriate output signals from said frequency analyzers for generating timing signals;

a storage means responsive jointly to said timing signals and said formant signal manifestations for storing the latter; and,

noise detecting means for disabling the generation of said timing signals upon the detection of unwanted noise frequencies within the speech spectrum to prevent storage of unwanted signal manifestations.

2. A Voice controlled speech analysis system comprising:

a plurality of frequency analyzers responsize to voice frequencies of the speech spectrum to provide appropriate output signals;

formant gating means jointly responsive to said timing signals and said signal manifestations to provide timed formant signal manifestations;

a storage means responsive jointly to said timing signals and said timed formant signal manifestations for storing the latter; and,

3. A voice controlled speech analysis system comprising:

formant energy detecting means connected to said frequency analyzers and providing formant signal manifestations indicative of the speech frequencies present;

gain control means for monitoring said output signals and providing a controlled input of fixed amplitude to said frequency analyzers;

timing signals generating means responsive to the output of said gain control means for generating timing signals;

a storage means jointly responsive to said timing signals and said formant signal manifestations for storing the latter; and,

noise detecting means for disabling the generation of said timing signals upon the detection of unwanted noise frequencies to prevent storage of unwanted signal manifestations.

4. A voice controlled speech analysis system comprising:

formant energy detecting means connected to said frequency analyzers and responsive to said output signals to provide formant signal manifestations indicative of the speech frequencies present;

a ring responsive to said timing signals to provide ring controlled gating signals;

a storage matrix responsive to said ring controlled gating signals and said timed formant signal manifestru tions for storing the latter; and,

noise detecting means for disabling the generation of said timing signals upon the detection of unwanted noise frequencies Within the speech spectrum to prevent storage of said unwanted signal manifestations 5. A system as in claim 4 further including an amplifier interposed between said frequency analyzers and said gain control means;

said amplifier including means for monitoring the output levels of the Waveforms representing the voice frequencies present in the speech spectrum.

6. A voice operated speech analyzing system for issuing noise free speech sound characteristics comprising:

amplitude control means in conjunction with speech amplification means for providing an amplitude controlled speech waveform in response to generated speech signals;

a noise clamp generator comprised of a differentiating network feeding a differential amplifier which issues a negative output level in response to speech waveforms free of noise, and a positive output level in response to noise frequencies detected in said speech signals;

means for connecting the output from said amplitude control means to said differentiating network;

timing signals generating means connected to the output of said noise clamp generator and operable in response to a negative output to generate timing signals, and non-operable in response to a positive out- P means responsive to said generated speech signals to provide speech sound characteristics; and,

gating means jointly responsive to said generated timing signals and said speech sound characteristics to provide said noise free speech sound characteristics.

7. A voice operated speech analyzing system for generating speech signal characteristics in response to speech signals derived from the frequencies in the speech spectrum comprising:

a noise clamp generator responsive to said speech signals to provide a positive or negative output depending respectively on the presense or absence of noise frequencies in said speech signals;

a timing signal generator initiated into operation in response to initial speech signals to generate timing signals and thereafter maintaining generation in response to said negative output, indicative of noise free speech signals, from said noise clamp generator, and interrupting generation of said timing signals in response to a positive output, indicative of the presence of noise frequencies; and,

means jointly responsive to said timing signals and said speech signals for generating noise free speech signal characteristics.

References Cited UNITED STATES PATENTS 3,109,066 10/1963 David 1791.8 3,126,449 3/1964 Shirman 179l.8 3,261,916 7/1966 Bakis.

KATHLEEN H. CLAFFY, Primary Examiner.

R. P. TAYLOR, Assistant Examiner.