US3816660A

US3816660A - Speech synthesizer with glide control

Info

Publication number: US3816660A
Application number: US00213079A
Authority: US
Inventors: A Nishiyama; E Tsuboka; H Ohwaki; H Yoshino; T Yamaguchi
Original assignee: Matsushita Electric Industrial Co Ltd
Current assignee: Panasonic Holdings Corp
Priority date: 1970-12-29
Filing date: 1971-12-28
Publication date: 1974-06-11
Anticipated expiration: 1991-06-11

Abstract

A film plate with a series of signal patterns representative of successions of individual speech-sound wave units stored thereon is disposed in front of a cathode ray tube having an optical fiber face plate. By appropriately controlling the scanning system of the cathode ray tube, the scanning beam is directed to the pattern of a signal representing the speech-sound wave unit on the film plate which is required to be read out. The intensity of the electron beam itself is controlled in such a manner as to smooth the glide portion between adjacent speech sounds. The corresponding speech sound is reproduced by detecting the signal for the desired speech sound by a photosensor disposed near the film plate.

Description

0 United States Patent 1 1111 3,816,660

Nishiyama et al. 1 June 11, 1974 [54] SPEECH SYNTHESIZER WITH GLIDE 3,115,605 12/1963 Coulter 179/1 VS O O 3.158.685- 11/1964 Gerstman.... 179/1 SA 3,166,640 l/1965 Dersch 179/1 SA Inventors: Aklra Nlshlyamfi, y g 3,344,239 9/1967 Ragland 179/1 SA Hirokazu Yoshino, Katano; Tetsuo Yamaguchi, Hirakata; Eiichi FOREIGN PATENTS OR APPLICATIONS Tsuboka, Nara; Haruki o 498.659 12/1953 Canada 179/1 SA Yokohama 0f Japan OTHER PUBLlCATlONS [73] Assignee: Matsushita Electric Industrial Co., Dersch IBM Technical Disdosure Bulletin, voice Osaka, Japan Readout System, V01. 3, N0. 2, p. 37, 7/1960.

[22] Filed: Dec. 28, 1971 Primary Examiner-Thomas W. Brown [21] Appl 2l3079 Assistant Examiner.lon Bradford Leaheey Attorney, Agent, or Firm-Stevens, Davis, Miller & [30] Foreign Application Priority Data Mosher Dec. 29, 1970 Japan 45-124824 Dec. 29, 1970 Japan 45-124839 [57] ABSTRACT g japan A film plate with a series of signal patterns representa- 1970 45':24843 tive of successions of individual speech-sound wave units stored thereon is disposed in front of a cathode ray tube having an optical fiber face plate. By appropriately controlling m6 scanning System Ofthe cathode I? 6 ra tube the Scannin beam is to attern 1581 Field of Search 179/1 SA 15 55 1 vsy P 35/35 35 of a signal representmg the speech-sound wave umt on the film plate which is required to be read out. The in- 56] R f Ct d tensity of the electron beam itself is controlled in such e erences e a manner as to smooth the glide portion between ad ja- UNITED STATES PATENTS cent speech sounds. The corresponding speech sound 2.137.888 11/1938 Fuller 35/35 A is reproduced by detecting the signal for the desired 1793-249 5/1957 Vilbig l79/1 SA speech sound by a photosensor disposed near the film 3.059.064 10/1962 Lebe" 35/35 A platg 3,102,165 8/1963 Clapper 179/1 SA 3.114.980 12 1963 Davis 35/35 9 Claims, 14 Drawing Figures GATE MEMO/F) D-A C/HCU/T C/RCU/T com mm? 7 5 MEMORY :73: MEMORY 0-21 u/r CIRCUIT com 7' Z) i 2 5 C/RCU/T 5/ a) 0 X g MJL T/PL/E/P DU/PAT/O/ i/VTE/VS/TY J sou/v0 TYPEDETECTOR f fl f ca/vmor 1 C/RCU/T C/RCU/T C/l-PCU/T MUL T/ AV PL/cArlo/v /2, AMPL/HER SWEEP 6E HFATO/F MFG PA FEWEMM 1 1914 SHEET 3 0F 6 RSUQB Akbm Alla v 1 SPEECH SYNTHESIZER WITH GLIDE CONTROL BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech-sound synthesizer and more particularly to a device which utilizes a flying-spot scanner system and which can produce synthesized speech sound wave unit by detecting information signals corresponding to speech sounds stored on a recording medium kept in contact with the scanning face of the cathode ray tube in the system by means of a photosensor disposed near the medium.

2. Description of the Prior Art There are two'types of conventional speech-sound synthesizers. One of them has a resort to the compilation of recorded speech sounds. According to this type, successions of desired speech sounds are obtained by selecting successively and compiling some of many speech sounds or corresponding digital signals stored in a magnetic rotating drum. This type of the synthesizer is widely in current use in the field where only a limited number of words are concerned.

However, this system needs the help of an electronic computer and a rotating drum for the storage, readingout and synthesis of the speech sounds and therefore the proper control of the overall system is very difficult.

The other type of the conventional speech-sound synthesizers is based on the synthesis by rule, which is further divided into two subtypes. The first type of synthesizer synthesizes speech sounds by modulating signals from a source of sound signals through a multistage variable LC circuit which simulates the motion of the muscles involved in the speech organ of a human being. On the other hand, the second type composes series of speech sounds by modulating such signals from a source through a multi-stage resonance circuit which simulates the acoustic spectrum of the human speech sounds. These synthesizers generate a specific speech sound by the use of the phenomenon that a base signal from a source of sound signals is necessarily subjected to modulation when it passes through a plurality of resonance circuits. Therefore, it is possible to control not only the fundamental frequency of the speech sounds treated by these devices but also the equivalents to the glides," that is, the transitory sounds connecting successions of speech sounds proper, produced by the organ of speech when it shifts from a position for one speech sound to that for another. In this case, however, the overall control system necessarily becomes intricate and a large computor having a high memory capacity is needed for processing the data of speech sounds and the control signal. For this reason, no such devices have been put into practice.

SUMMARY OF THE INVENTION The object of the present invention is to provide a speech-sound synthesizer which comprises a memory plate having wave form signals corresponding to a plurality of speech sound wave unit stored in the respective addresses thereon, which memory plate is disposed in front of a cathode ray tube; a deflecting system including a means for sweeping the electron beam of the cathode ray tube in one direction and a means for switching over the positions of the beam in response to an information signal consisting of an address information for selecting desired speech sounds and a synchronizing signal; a detector for detecting the electron beam or the energy converted corresponding thereto; and a demodulator for reproducing speech sounds from the output of the detector.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 4 shows another type of film plate on which PCM signals corresponding to speech sound wave unit are printed and which is also considered to be an embodiment of the invention.

FIG. 5 shows a series of speech-sound signals including a synchronizing signal, a horizontal (or X-axis) sweep signal and a blanking signal, with respect to the same time base.

FIG. 6 shows a block diagram of a speech-sound synthesizer as another embodiment of the invention.

FIG. 7 is a schematic block diagram of a circuit which generates duration signals to maintain each of individual speech sounds continuously for a certain period of time.

FIGS. 8a-8e show wave forms necessary for illustrating the operation of the circuit shown in FIG. 7.

FIG. 9 is a schematic block diagram of a control circuit which generates glides or transitory sounds produced when the organ of speech shifts from one position to produce a particular speech sound to another position to produce a different speech sound.

FIG. 10 shows wave forms necessary for illustrating the operation of the control circuit shown in FIG. 9.

FIG. 11 is a schematic block diagram of a circuit which controls the fundamental speech-sound frequency, accent and amplitude of speech sounds to be produced, depending upon whether the sounds to be produced are of a male adult, a female adult or a infant.

FIGS. l2, l3 and 14 show wave forms necessary for illustrating the operation of the circuit shown in FIG. 11.

DESCRIPTION OF THE PREFERRED EMBODIMENTS In FIG. I, which shows a speech-sound synthesizer embodying the present invention, a speech-sound signal is a coded signal consisting of a Y-axis address information which is needed when the film plate is scanned by the beam of the fiber optical tube and a synchronizing signal, and the speech-sound signal is applied to both a D-A converter 51 and a sync separator circuit 52. The D-A converter 51 delivers an analog control voltage to determine the amplitude of deflection along the Y-axis to a Y-axis deflecting circuit 54 each time the converter 51 receives a synchronizing signal from the sync separator circuit 52. An X-axis deflecting circuit 53 generates a saw-tooth wave having a constant amplitude and frequency in response to a signal from the sync separator circuit 52 and accordingly the beam of a fiber optical tube 58 scans at film plate 59 along the X-axis thereof. Also, a blanking circuit 55 delivers a blanking signal to perform the function of blanking the scanning lines during the horizontal retrace. FIG. should be referred to for speech-sound signal, X-axis sweep signal and blanking signal which are shown in a comparative manner with the same time base. An output from the sync separator circuit 52 actuates a triangular wave generator circuit 56. The output of the generator circuit 56 and the output of the Y- axis deflecting circuit 54 are mixed together through a mixer 57, the output of which in turn sweeps the beam of the fiber optical tube 58 along the Y-axis.

By deflecting the electron beam of the fiber optical tube 58 with such X-axis or horizontal deflecting voltage and Y-axis or vertical deflecting voltage as described above, the locus of the swept beam assumes a triangular wave form as shown in FIG. 3. If the film plate 59 shown in FIG. 2 which has a wave form corresponding to a single speech sound wave unit printed thereon is scanned with the swept beam, the speech sound wave unit can be put out of a photomultiplier 60 as a PWM (pulse width modulation) signal as shown in FIG. 3. This PWM signal is converted to an acoustic wave form having an audio frequency through a lowpass filter 61, and the audio signal corresponding to the acoustic wave form is then amplified by an amplifier 62 and-thereafter drives a speaker 63 to produce a synthesized speech sound.

FIG. 4 shows another type of film plate according to the invention, on which speech sound wave unit are stored in the form of a PCM signal, the PCM signals being represented as a series of small transparent rectangular portions formed in the opaque film plate. If a row of the PCM signals corresponding to a speech sound wave unit are swept by a horizontal sweep signal (X-axis sweep signal), the photomultiplier will receive a light signal modulated in accordance with the PCM signals. The photomultiplier converts the optical signal to an electric one and the electric signal can easily be transformed into a corresponding synthesized speech sound. The speech sound wave unit to be stored on the film plate may be in the form of a series of codes represented by different degrees of transparency. Although there are different ways of printing the sound information on the film plate, synthesized speech sounds can be produced in almost the same manner with a slight change in the deflecting system to be used.

FIG. 6 is a block diagram of a speech-sound synthesizer, in which the synthesizer shown in FIG. 1 is further provided with a circuit which generates the glides between two successive speech sounds proper and controls the fundamental speech-sound frequency. In FIG. 6, coded speech-sound signals are applied to an input terminal S and temporarily stored in a memory 1, which sequentially sends out them at the same speed as the ordinary spoken speech has. Gate circuits 2 and 3 are alternately opened as the coded signals arrive sequentially. Memory circuits 6 and 7 are adapted respectively to temporarily store corresponding coded speech-sound signals from the gate circuits 2 and 3, respectively, until the next speech-sound signals arrive at the memory circuits 6 and 7. If new coded speechsound signals reach the memory circuits 6 and 7 after the gate 2 and 3 have been opened. the old coded signals having been stored till then are cancelled and the new coded signals are stored instead in the memories 6 and 7. D-A (digital-analog) converters 9 and 10 generate analog voltages corresponding to the coded speech-sound signals which are received at the outputs of the converters 9 and 10. The outputs of the D-A converter 9 and 10 are applied to the vertical deflecting inputs Y and Y of a fiber optical tube 13 and sweep the electron beam of the tube to read the speech sound wave unit stored on a film plate 14. On the other hand,

the coded speech-sound signals are applied also to a speech-sound type detector circuit 5 which extracts the distinction between the vowel and the consonants and the distinction of the antecedent consonants from other consonants.

The output of the speech-sound type detector circuit 5 is applied to the input terminals of a duration'measurement circuit 8 which determines the durations of individual speech sounds. The output of the duration measurement circuits 8 is applied, on the one hand, to the gate circuit 2 to open the same and, on the other hand, to the gate circuit 3 through a polarity inverting circuit 4 to open the gate 3. Accordingly, the gates 2 and 3 open alternately in accordance with the duration of each speech sound. In addition, the output of the duration measurement circuit 8 is fed to an intensity control circuit 11 which generates a glide for each speech sound on the basis of the duration signal from the circuit 8 corresponding to the speech sound. The output from the intensity control circuit 11 is .then applied to the intensity control terminal Z of the fiber optical tube 13 to intensity-modulate the beam of the tube 13. A sweep generator and controller circuit 12 is so controlled as to generate a variety of the pitch frequencies of individual speech-sound, accents and amplitudes of sounds in order to represent a variety of sounds produced by men and women. The output of the sweep generator and controller circuit 12 is applied to the sweep control terminal of the fiber optical tube 13 to sweep the electron beam of the tube 13. The swept beam scans a film plate 14 on which individual speech sound wave units are stored, and the light which is passed through the film plate 14 is received by a photosensor such as a photomultiplier 15. The change in the light detected by the photomultiplier 15 is converted to the corresponding electric signal, which is in turn fed to and amplified by a multiplication amplifier 16 to allow a speaker 17 to produce continuous speech sounds. The amplification factor of the multiplication amplifier 16 may be externally controlled.

FIG. 7 shows in detail the duration measurement circuit 8 used in the speech-sound synthesizer shown in FIG. 6, which circuit derives a duration characteristic of an individual speech sound from a speech-sound code corresponding to the individual speech sound. FIG. 8 shows wave forms for illustrating the operation of the circuit shown in FIG. 7. In FIG. 7, a pulse voltage is applied to a terminal CV or V according as the speech-sound duration extractor circuit 5 shown in FIG. 6 delivers a discrimination signal indicative of the extraction of a consonant or a vowel. Namely, if a con sonant is extracted, a pulse voltage appears at the terminal CV and the duration of the consonant is determined by a monostable multivibrator 21. One can see this state in the wave form in FIG. 8a, in which the horizontal axis represents the time base. Also in this case, the duration of the antecedent (e.g., K in the Japanese word "ka) consonant which together with an attributive vowel (e.g., a in the Japanese word ka) constitutes the above mentioned consenant. is determined by a monostable multivibrator 22 (see wave form in FIG. 8b). This is done as follows. The duration characteristic of the antecedent consonant is extracted by the antecedent consonant extractor circuit contained in the speech-sound duration extractor circuit 5 shown in FIG. 6 and the signal representative of the duration is applied to one of terminals t,, t t and t associated with switching elements connected in series respectively with capacitors C,, C C and C, which determine the duration of the nonstable state of the monostable multivibrator 22. Accordingly, upon the switching on of a particular switching element, the duration of the previously mentioned antecedent consonant is determined in accordance with the capacitance of the capacitor associated with the particular switching element. Moreover, the duration of the attributive vowel of the consonant in question is determined by detecting with a flip-flop 24 a signal indicating the termination of the duration of the antecedent consonant, delivered from the multivibrator 21 and a signal indicating the termination of the duration of the consonant as a whole, delivered from the vibrator 22 (see wave form in FIG. 80). Now, if a vowel proper is extracted by the speech-sound duration extractor circuit 5 shown in FIG. 6, a pulse voltage appears at the terminal V and the duration of the vowel is determined by a monostable multivibrator 23 (see wave form in FIG. 8d). The outputs of the

multivibrators

22 and 23 and of the flipflop 24 are all applied to an adder (OR circuit) 25, the output of which drives a flip-flop 26, which delivers an output representing a series of continuous speechsounds, each having its proper duration (see wave form in FIG. 8e).

FIG. 9 shows in detail the intensity control circuit 11 used in the synthesizer shown in FIG. 6, which circuit controls the intensity of the beam of the fiber optical plate and therefore generates glides, i.e., transitory sounds between successive speech-sounds. FIG. shows wave forms for illustrating the operation of the circuit shown in FIG. 9. In FIG. 9, a duration signal for speech sounds in the form as shown in the wave form E in FIG. 8 is received from the duration measurement circuit 8 shown in FIG. 6. The received signal is applied to a NOT circuit 31 which delivers an output whose polarity is opposite to that of the input signal. The polarity-inverted signal from the NOT circuit 31 is then applied to an integrating circuit 34. The received signal is also applied directly to an integrating circuit 33. The output wave forms from the integrating circuits 33 and 34 are indicated respectively at Y, and Y, in FIG. 10, the wave forms Y, and Y being voltages obtained by integrating the wave form E in FIG. 8 but opposite in polarity to each other. A high frequency square wave generator circuit 32 delivers a chopping signal to alternately switch over the outputs Y, and Y, of the D-A converter circuits 9 and 10 shown in FIG. 6 and to alternately open gate circuits 35 and 36. The outputs of the gate circuits 35 and 36, which are intensity modulating signals, are applied together to the terminal Z of the fiber optical tube 13 to intensity-modulate the beam of the tube.

Namely, as seen from FIG. 10 where the horizontal axis is the time base, the intensity due to the signal Y, start from zero level, gradually rises and reaches a maximum during the time from t,, to r,. The intensity due to the signal is at a maximum at t but no output is delivered since the signal Y, has not a stored speech sound at t The vertical axis or the ordinate in FIG. 10 represents the voltage but for convenience sake it is assumed to represent the intensity. Therefore, the maximum intensity produces the corresponding maximum acoustic output while the zero intensity corresponds to the zero acoustic output. The intensity due to the signal Y, is the maximum during the time from t, to I; while that due to the signal Y is zero during the same time.

Accordingly, the photomultiplier 15 in FIG. 6 receives.

only the signal Y,. During the time from l to r, the intensity due to the signal Y, gradually decreases from maximum to zero while that due to the signal Y increases gradually from zero to maximum. As to the signal to the photosensor 15, the signal Y, gradually decreases with the gradually increasing signal Y, so that one speech sound C, is continuously superseded by another speech sound V,. During the time from t, to t.,, only the Y signal is received by the photosensor l5, and from t, to t the signal Y, gradually increases with the gradually decrease in the signal Y From t to the photosensor 15 receives only the signal Y, having the maximum intensity corresponding to speech sound V During the time from 1,, to t, the signal Y, gradually falls and the signal Y gradually rises. If the signal Y, sweeps no speech sound wave unit on the film during the time from t,, to t-,, the photosensor l5 delivers no output or a zero output during that time. Through the intensity modulation method described above, a wave form as shown in FIG. 14 can be obtained which is the envelope of the output wave forms of the multiplication amplifier 16 shown in FIG. 6 obtained by passing the output wave forms through a low-pass filter, and which gradually rises from zero, fluctuates in amplitude according to the quality of speech sound and return to zero at the end of the sound. The coordinate of FIG. 14 has time scale on the horizontal axis and voltage scale of the amplitude for the envelope on the vertical axis.

FIG. 11 shows in detail the sweep generator and controller circuit 12 used in the synthesizer shown in FIG. 6, which circuit can generate a variety of the pitch frequencies of individual speech-sound and accents produced by men and women or infants and adults if it is externally controlled and can internally and automatically control the change in the pitch frequency due to the amplitude. In FIG. 11, a saw-tooth wave generator circuit 41 sweeps the beam of the fiber optic tube 13 in FIG. 6. The output of the circuit 41 is applied to the sweep input terminal X of the fiber optical tube 13 in FIG. 6. The output signal from the multiplication amplifier 16 shown in FIG. 6 is applied to a low-pass filter 45 in FIG. 11, which filter delivers such an envelope wave form as shown in FIG. 14. The signal wave form representing the envelope of amplitudes is then sent to a phase inverter circuit 46 to have its phase changed oppositely or shifted by The output of the phase inverter circuit 46 is fed to an adder 44, which may receive also at the terminal ACC an external control signal to control the accents by controlling the pitch frequency. The output signal from the adder 44 and the saw-tooth voltage output from the saw-tooth wave generator circuit 41 are compared through a comparator 43, which delivers a control signal only when both the outputs received becomes equal to control the peak of the saw-tooth voltage from the generator circuit 41. This means that the pitch frequency can be controlled, that is, the change in the pitch of speech sounds can be provided due to the change in the accents or ampli tudes, by shifting the position of the end of the wave form corresponding to a single speech sound wave unit as shown in FIG. 12 stored on the film plate as shown in FIG. 2.

In FIG. 12, the wave forms characterized by the intervals a, b and c correspond respectively to a high, standard and low pitch. The intervals a, b and c can be determined respectively by setting the peak of the sawtooth voltage generated by the circuit 41 at the levels A, B and C as seen in FIG. 13. Therefore, according to the invention, the pitch can be arbitrarily controlled by controlling the voltage output from the adder 44 in FIG. 11. A sweep time setting circuit 42 can vary the sweep time by applying a control voltage to a terminal M.F.C. and therefore can synthesize approximated sounds produced by male adults, female adults and infants.

Furthermore, according to the present invention, the fiber optical tube can be replace by a monoscope. In such a case, a plurality of patterns corresponding to desired speech sound wave units are stored on the aluminum foil as a target provided in the monoscope and the speech sound wave units are electrically read out through the use of the secondary electrons emitted by scanning the foil with the primary beam to synthesize the desired sounds.

We claim:

1. A speech-sound synthesizer comprising a memory plate for storing a plurality of speech-sound information signals corresponding to a plurality of speechsound wave units in respective addresses on said memory plate, said memory plate being disposed in front of a cathode-ray tube; means for moving an electron beam to a desired address on said memory plate in response to a coded address signal; means for sweeping said electron beam along a stored speech-sound information signal in said desired address; means for controlling the intensity of the electron beam of said cathode-ray tube, and thereby the image on said tube so that said intensity gradually changes at the beginning and end portions of each of said speech-sounds; and means for detecting light from said cathode-ray tube through said memory plate and for converting said light modulated by said memory plate into corresponding speech-sounds.

2. A speech-sound synthesizer according to claim 1, wherein waveforms of speech-sound wave units are recorded on said memory plate as speech-sound information, and there are provided a first deflection circuit for generating a signal scanning said electron beam in the amplitude. I

3. A speech-sound synthesizer according to claim 1, wherein PCM modulated speech-sound wave units are recorded on said memory plate as speech-sound information signals.

4. A speech-sound synthesizer according to claim 1, wherein said speech-sound information signals are represented by densities corresponding to the amplitudes of speech-sound wave units.

5. A speech-sound synthesizer according to claim 1, further comprising a means for determining the duration of said individual speech-sounds, said sweeping means repeatedly sweeping said electron beam on each of selected speech-sound wave units.

6. A speech-sound synthesizer according to claim 5, wherein said means for determining the duration of said individual speech-sounds comprises a first monostable circuit to determine the duration of a consonant consisting of an antecedent consonant and an attributive vowel, a second monostable circuit to determine the duration of said antecedent consonant by selecting appropriate capacitance from among a group of different capacitances, and a third monostable circuit to determine the duration of said attributive vowel with the aid of said first and second monostable circuits.

7. A speech-sound synthesizer according to claim 5, wherein said means for controlling the intensity of said cathode-ray tube comprises a first integrating circuit to integrate the outputpf said means for determining the duration of individual speech-sounds, a second integrating circuit to integrate the signal opposite in phase to said output signal from said duration determining means, and a gate means to alternately pass the outputs from said first and second integrating circuit, the gated integranting circuit outputs being applied as a deflecting signal to said cathode ray tube as an intensity modulating signal to generate glides" in said speech-sounds.

8. A speech-sound synthesizer according to claim 1, wherein a means for variably controlling the sweep width of said individual speech-sound information signals is provided in said sweeping means, thereby to variably control the pitch frequencies of speechsounds.

9. A speech-sound synthesizer according to claim 1, wherein said sweeping means comprises a phase inverting circuit to phase-invert the envelope of acoustic wave forms, an adder to add an external signal for controlling the frequencies of speech-sound pitches to said inverted envelope, a comparator circuit to compare the output signal from said adder with the output signal from a saw-tooth wave generator circuit which is controlled by the output of said comparator, whereby said pitch frequencies are controlled by using the signal from said saw-tooth wave generator circuit as a deflecting signal.

Claims

1. A speech-sound synthesizer comprising a memory plate for storing a plurality of speech-sound information signals corresponding to a plurality of speech-sound wave units in respective addresses on said memory plate, said memory plate being disposed in front of a cathode-ray tube; means for moving an electron beam to a desired address on said memory plate in response to a coded address signal; means for sweeping said electron beam along a stored speech-sound information signal in said desired address; means for controlling the intensity of the electron beam of said cathode-ray tube, and thereby the image on said tube so that said intensity gradually changes at the beginning and end portions of each of said speech-sounds; and means for detecting light from said cathode-ray tube through said memory plate and for converting said light modulated by said memory plate into corresponding speech-sounds.

2. A speech-sound synthesizer according to claim 1, wherein waveforms of speech-sound wave units are recorded on said memory plate as speech-sound information, and there are provided a first deflection circuit for generating a signal scanning said electron beam in the direction of the time axis of the waveform of a speech-sound wave unit and a second deflection circuit for deflecting said electron beam in the direction of the amplitude of the waveform of a speech-sound wave unit to oscillate in a constant frequency with a predetermined amplitude.

7. A speech-sound synthesizer according to claim 5, wherein said means for controlling the intensity of said cathode-ray tube comprises a first integrating circuit to integrate the output of said means for determining the duration of individual speech-sounds, a second integrating circuit to integrate the signal opposite in phase to said output signal from said duration determining means, and a gate means to alternately pass the outputs from said first and second integrating circuit, the gated integranting circuit outputs being applied as a deflecting signal to said cathode ray tube as an intensity modulating signal to generate ''''glides'''' in said speech-sounds.

8. A speech-sound synthesizer according to claim 1, wherein a means for variably controlling the sweep wIdth of said individual speech-sound information signals is provided in said sweeping means, thereby to variably control the pitch frequencies of speech-sounds.