EP0128065B1 - Method of generating an excitation signal for a channel or linear prediction vocoder, and excitation signal generator therefor - Google Patents

Method of generating an excitation signal for a channel or linear prediction vocoder, and excitation signal generator therefor Download PDF

Info

Publication number
EP0128065B1
EP0128065B1 EP19840400915 EP84400915A EP0128065B1 EP 0128065 B1 EP0128065 B1 EP 0128065B1 EP 19840400915 EP19840400915 EP 19840400915 EP 84400915 A EP84400915 A EP 84400915A EP 0128065 B1 EP0128065 B1 EP 0128065B1
Authority
EP
European Patent Office
Prior art keywords
signal
excitation signal
input
excitation
output
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired
Application number
EP19840400915
Other languages
German (de)
French (fr)
Other versions
EP0128065A1 (en
Inventor
Jean-Frédéric Zurcher
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Publication of EP0128065A1 publication Critical patent/EP0128065A1/en
Application granted granted Critical
Publication of EP0128065B1 publication Critical patent/EP0128065B1/en
Expired legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/08—Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters
    • G10L19/10—Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters the excitation function being a multipulse excitation
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/02—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders

Definitions

  • the invention relates to a method for generating an excitation signal for a channel or linear prediction speech synthesizer, and to an excitation signal generator for such a synthesizer.
  • the invention relates to the synthesis chain of channel vocoders and linear prediction vocoders.
  • the vocoders can be of three types, either of the conventional type, that is to say with detection of the fundamental frequency of the speech signal, or of the baseband type, or of the voice excitation type. The general principle of such vocoders is illustrated in FIG. 1.
  • a vocoder can be broken down into three parts: an analysis sub-assembly 2, a transmission member 4 and a synthesis sub-assembly 6.
  • the analysis sub-assembly 2 will transform a speech signal analog into a digital signal.
  • time coding of the speech signal is not carried out, as in the case for example of a telephone transmission by Pulse Modulation and Coding (MIC ), but we will perform a frequency coding of the speech signal.
  • the bit rate of the speech signal thus coded is of the order of 2 to 12 kbit / s, which is much lower than the MIC coding which requires 64 kbit / s.
  • the speech signal is represented by two types of parameters.
  • a first type of parameters which describes the instantaneous spectral envelope of the speech signal and a second type of parameters which describes the fine structure of the spectrum of the speech signal. This fine structure is often characterized by a single parameter which is the value of the fundamental frequency.
  • a first means 8 of the analysis sub-assembly 2 analyzes and codes the instantaneous spectral envelope of the speech signal and a second means 10 of the analysis sub-assembly codes the fine structure of the spectrum of the speech signal.
  • the two digital signals obtained at the output of the first means 8 and of the second means 10 of the analysis sub-assembly 2 are then transmitted in the transmission member 4, for example in the form of a frame.
  • the two digitized signals are received by the synthesis sub-assembly 6.
  • the digital signal encoding the spectral envelope of the speech signal P is then applied to the input of a means 12 consisting of a filter or a set of filters to restore this spectral envelope.
  • the second digital signal coded by the second means 10 of the analysis sub-assembly 2, is applied to the input of a means 14 for generating an excitation signal which has a fine structure spectrum similar to that of the signal analyzed but of neutral spectral envelope and which delivers a signal to the means 12 which restores a spectral envelope similar to that of the speech signal analyzed.
  • the first means 8 encodes the entire spectral band analyzed, for example between 200 Hz and 6,000 Hz and the means 10 codes a parameter allowing the development of the excitation signal at the reception.
  • the spectral band is cut into a low band, called the base band, for example between 200 Hz and 1000 Hz, and a high band.
  • the low band is coded according to a conventional temporal method by the means 10; on reception, the excitation signal is taken from the decoded base band and the high band is coded by the first means 8 as the total spectral band of the vocoders of the first type.
  • the processing performed by the third type of vocoder differs from that performed by the second in that the excitation signal is not deduced from the decoded baseband, but is obtained from parameters provided by a temporal analysis of the signal. of speech.
  • the baseband In known baseband vocoders, the baseband is generally around 800 Hz in width. Indeed, it is necessary that the baseband contains the fundamental frequency of the speech signal and several of its harmonics in order to generate an excitation signal comprising all the harmonic frequencies of the fundamental frequency which are contained in the spectral band processed. . In addition, in known baseband vocoders, the development of the excitation signal requires sophisticated spectrum equalization treatments to deliver an approximately flat excitation signal, i.e. a signal excitation whose instantaneous spectral envelope does not have too strong ripple.
  • Known baseband vocoders therefore have the drawbacks, on the one hand, of a high data rate between the analysis sub-assembly and the synthesis sub-assembly because of the width of the base band to be transmitted and on the other hand a high cost price because of the sophisticated processing carried out to generate the excitation signal.
  • the invention overcomes these drawbacks. It resides in a method for developing the excitation signal obtained without sophisticated processing, which is a significant improvement in particular in the case of baseband vocoders, or with vocal excitation.
  • the excitation signal has a substantially constant energy which does not depend on the value of the fundamental frequency.
  • the excitation signal has an energy significantly higher than the energy of an excitation signal consisting of single pulses spaced from the fundamental period. This has the advantage of improving the signal / noise ratio of the synthesized signal.
  • the synthetic signal obtained provides a more natural auditory impression (less metallic), if the comparison is made with what is obtained with an excitation signal consisting of pulses spaced from the fundamental period.
  • the subject of the invention is a method of developing an excitation signal for a channel synthesizer or a linear prediction, in which a signal consisting of a sequence of pseudo- pulses is delivered. predetermined random frequency, and in which said sequence is reset when a determined event occurs whose repetition frequency is linked to the fundamental frequency of the speech signal, the amplitude of the excitation signal having a predetermined fixed value and the polarity of said signal d excitation being determined by the logical level of the pulses of the pseudo-random sequence.
  • a signal representative of the value of the fundamental frequency is received at the synthesis and is generated, in relation with the occurrence of the determined event, a signal to reset the pseudo-random pulse sequence, the frequency of this signal being linked to the value received from said basic frequency.
  • the determined event is constituted by the passage of the signal of the baseband received by the value zero in a determined direction, or by the passage of the baseband signal through a determined extremum.
  • an excitation signal generator for a speech synthesizer comprising generator means for delivering a predetermined pseudo-random pulse sequence, means for resetting the generator means in response to the reception of a control signal, means for producing the excitation signal sensitive to the logic level of the pulses of said sequence to generate an excitation signal of predetermined fixed amplitude and polarity determined by the logic level of said sequence and detection means of a determined event whose frequency is linked to the fundamental frequency of the speech signal, said detection means controlling said reset means.
  • the generator means comprise a shift register, at least one NON-OR-EXCLUSIVE gate whose inputs are connected to two output stages of the register, the output of said door being connected to the data input of the register, the latter further comprising a reset input connected to the output of the reset means.
  • the generator means comprise, in the case of a channel synthesizer, a logic inverter whose input is connected to the output of the NOR-OR gate. EXCLUSIVE.
  • FIG. 2 represents the central core of the excitation signal generator according to the present invention.
  • This generator 40 comprises a shift register 46 whose outputs of two of the stages Q7 and Q9 are looped back to its data input D via a NOR gate 48.
  • the shift register is clocked by a HOR clock signal applied to its input H, and also includes a reset input R, to ensure the reset of the register in response to the application of a reset signal.
  • the generator 40 also includes an AND gate 56, one of which input receives the aforementioned clock signal, and two flip-flops 58 and 60 of type D which are mounted in the following manner: the clock input of flip-flop 58 is connected to the output of gate 56 and its output Q is looped back to its input terminal D and connected to the clock input of the flip-flop 60. The output Q of the latter is also looped back to its input terminal D and also to one of the inputs of the AND gate 56.
  • the reset inputs to "1 RAU (Set in English terminology) of flip-flops 58 and 60 are connected permanently to ground, and the reset inputs to “0” RESET (Reset in English terminology) are connected to a line intended to receive a control signal which has a repetition period corresponding to the frequency of the fundamental of the signal of speech.
  • the output Q of the flip-flop 58 is connected to the input R of the register 46.
  • An inverter 54 has also been provided, the input of which receives the output signal from the gate 48.
  • the input signals S1 and output S2 of the inverter 54 are in phase opposition, and are intended to produce respectively the excitation signals for the odd and even channels of the synthesis subset in the case a channel synthesizer.
  • the clock signal HOR received by the shift register 46 has, in the case of using a digital synthesizer, the same frequency as the sampling frequency of the digital synthesis filters, for example a frequency of 16 kHz.
  • the two flip-flops 58 and 60 have the role of synchronizing the reset pulses applied to the shift register 46, with the clock pulses.
  • the generator 40 operates as follows: when the signal applied to the reset input of flip-flops 58 and 60 is at "1 •, the outputs Q of these scales are kept at "1 •.
  • the RESET input of flip-flops 58 and 60 receives a “0 •, the reset to zero of these flip-flops is no longer imposed, and on the first rising edge of the clock signal HOR which follows this event, the output Q of flip-flop 58 goes to state “0” given the looping back from to input D.
  • the zeroing of output Q of flip-flop 58 acts at the level of reset input R, which puts the outputs Q of register 46 to zero, that is to say, which reinitializes the pseudo-random pulse sequence delivered by the gate shift register system NON-OR-EXCLUSIVE.
  • the output Q of flip-flop 58 changes to "1 since its input D is at" 0 •, previous value on output Q. This transition acts on the clock input of flip-flop 60 and the output Q of the latter goes to zero.
  • a pulse on the entry R of the register 46 which is in phase with the clock signal HOR and whose width is equal to the duration separating two clock pulses.
  • the gate 48 delivers a "1 which will be transferred in the different stages of the register at the rate of the clock. This level "1 is maintained until a" 1 "propagates to a stage n whose output Qn is connected to the input of gate 48 (stage Q7 if we consider FIG. 2) . At this moment, the logic state changes at the output of gate 48, and in the remainder of the sequence, this gate delivers a series of pulses of pseudo-random width.
  • FIG. 2 shows connections between the outputs of stages Q7 and Q9 of register 46 and the inputs of gate 48, other alternatives are possible for the choice of output stages.
  • the output of gate 48 delivers a signal whose logic level determines the polarity of the excitation signal of the synthesis chain (this signal will have a constant amplitude in absolute value).
  • This excitation signal is directed towards the input of the synthesis predictor filter (for a linear prediction synthesizer), or to the inputs of the bandpass filters of the synthesis channels (in the case of a channel vocoder).
  • the inverter 54 intervenes as follows: the signal S1 on the input of the latter is used to generate the excitation signal of the synthesis channels of odd rank, while the signal S2 on its output is used to generate the excitation signal of even rank channels.
  • This measurement makes it possible to compensate for phase shifts between channels in the usual case where the filters of the synthesis channels are of the fourth order and overlap at -6 dB and practically in phase opposition. An essentially flat frequency response is thus obtained over the entire frequency band thus treated.
  • the reset is done in response to the reception on the RESET inputs of a signal which has a repetition period corresponding to the fundamental frequency of the speech signal, and that the pseudo-random pulse sequence always proceeds from the same way after its reset, it appears that this sequence also has a periodicity corresponding to the fundamental frequency. Since this sequence is made up of a series of pulses, it therefore has a broad frequency spectrum, and on average flat, comprising all the harmonics of the fundamental frequency necessary for the proper functioning of a speech synthesizer. In the absence of a fundamental frequency, the reset is done randomly and the sequence has a white noise spectrum.
  • FIG. 3 a variant of the embodiment of the generator of Figure 2 to obtain a more regular and less noisy excitation signal for the first channels (in the case of a channel synthesizer) vocoder (channels covering the spectral band up to approximately 1000 Hz).
  • This excitation signal allows in particular a better reproduction of the voiced sounds.
  • This generator comprises the same elements as that of FIG. 2 which are designated by the same reference numbers.
  • a flip-flop 62 has been added to it, the reset input at “0” (RESET) being connected to the output Q of flip-flop 58, and the reset input at “1” (RAU) receiving the output signal S2 of the logic inverter 54.
  • the data inputs D and clock H are reset to zero by connection to ground.
  • the output Q of this flip-flop delivers a signal S 01 which is used to produce the signal of excitation of the first channels of odd rank, for example of ranks 1 and 3 and the output delivers a signal S 02 which is used to produce the excitation signal of the first channels of even rank, for example of rank 2 and 4.
  • the output Q of the flip-flop 62 delivers a sequence of p " 1 "successive, the length of this sequence being equal to the total length of the pulse of level" 1 ", at the output of gate 48, immediately following the reset. Thereafter, the Q output of flip-flop 62 remains at zero until the next reset. It can be seen that the flip-flop 62 behaves in the same way as a conventional flip-flop.
  • the pulses transmitted to the reset inputs of flip-flops 58 and 60, and whose repetition period corresponds to the frequency of the fundamental of the speech signal, are generated in the following manner:
  • the reset to zero is controlled by pulses spaced apart by random lengths or else the reset can be suppressed.
  • the means 10 comprises a low-pass filter 16 which allows only the base band of the spectral band of the speech signal P to pass and a coding means 18 which performs digital coding of the analog signal delivered by the low-pass filter 16
  • the width of the baseband is for example of the order of 500 Hz.
  • the coding means 18 performs time coding in a known manner.
  • the digital signal delivered by the means 10 is framed with the digital signal delivered by the means 8 corresponding to the data of the analysis channels by a framing means 20.
  • the frame is then transmitted by a transmission device 22 on a transmission line L.
  • the frame is applied to the input of a reception member 24.
  • This is connected in series to a separation means 26 which delivers the channel data to the means 12 on the one hand, and the digital signal encoding the baseband by means 14 of developing the excitation signal on the other hand.
  • This means 14 comprises in series a delay means 28, a decoding means 30, a low-pass filter 32, an adder 34, an input of which is connected to a noise generator 36, a detection means 38 and a signal generator excitation 40 of the type according to the invention and previously described in FIGS. 2 and 3.
  • the delay means 28 has the function of compensating for the delay of the channel data on the baseband signal caused by the low pass filters of the analysis channels and the transmission of the channel data. This delay is adjusted empirically. It is around 20 to 30 ms.
  • the signal delivered by the means 28 is then decoded by the means 30 which restores an analog signal.
  • This means 30 performs the reverse transformation from that carried out by the means 18 of FIG. 2a. It is followed by a low-pass filter 32, similar to the low-pass filter 16, at the output of which the baseband signal is found.
  • This signal can be noisy thanks to an adder 34 receiving the baseband signal on one input and a noise generated by a white noise generator 36 on another input. This addition of noise is optional.
  • the baseband does not transport energy, to apply to the input of the detection means 38 a signal whose zero crossings are almost random, zero crossings from which a random excitation signal can be generated.
  • the added noise is low, its level can be for example - 60 dB compared to the maximum level of the baseband.
  • the detection means 38 has the function of detecting the occurrences of a determined event in the baseband and of emitting a pulse at each of these occurrences.
  • This determined event the repetition frequency of which corresponds to the fundamental period of the speech signal, may be the zero crossing in a determined direction of the baseband signal, the detection then being carried out simply by a comparator. It is also possible to choose as an easily detectable event, for example, the passage of the baseband signal through a determined extremum.
  • the pulses delivered by the detection means 38 to each of said occurrences are applied to the reset inputs of flip-flops 58 and 60 of the generator 40.
  • the output of the pseudo-random pulse sequence is paced by the clock signal HOR and reset when a pulse is delivered at the output of the means 38.
  • Voice-activated vocoders differ from baseband vocoders in that the baseband is not transmitted. In vocoders with vocal excitation, the detection of the occurrences of the determined event mentioned above is therefore done in the subset of vocoder analysis.
  • FIG. 5a represents part of an analysis sub-assembly of a vocoder with vocal excitation.
  • the means 10 comprises in series, a low-pass filter 32 receiving the speech signal P as an input, an adder 34 receiving on a input a noise signal coming from a white noise generator 36, a detection means 38 delivering a pulse at each occurrence of the determined event and a coding means 42.
  • This coding means 42 codes the dates of appearance of the determined event.
  • This coding means which is simple to implement has the advantage of requiring only a low bit rate equal to nF bit / second.
  • the signal delivered by the coding means 42 is framed with the analysis data coming from the first means 8 in a framing means 20.
  • a transmission member 22 then delivers the frame on the transmission line L.
  • the transmitted signal is received by a reception member 24.
  • This is connected to a separation means 26 which delivers the data analysis by means 12 on the one hand and the signal coding the dates of occurrences of the event determined by means 14 of development of the excitation signal on the other hand.
  • This means 14 comprises a decoding means 44 receiving as input every T seconds, a number determining the date of occurrence of the determined event, or its absence, in said time interval T and delivering an impulse to each of said occurrences, this pulse being transmitted to the reset inputs of flip-flops 58 and 60 of the excitation signal generator 40.

Abstract

1. Process for the generation of an excitation signal for a speech synthesizer provided with channels or linear prediction, from a signal formed by a predetermined pseudo random sequence of pulses, characterized in that the said sequence is reintialized when a specified event occurs, the frequency of repetition of which event is linked to the fundamental frequency of the speech signal, the amplitude of the excitation signal having a predetermined fixed value and the polarity of the said excitation signal being determined by the logic level of the pulses of the pseudo-random sequence.

Description

L'invention concerne un procédé d'élaboration d'un signal d'excitation pour un synthétiseur de parole à canaux ou à prédiction linéaire, et un générateur de signal d'excitation pour un tel synthétiseurThe invention relates to a method for generating an excitation signal for a channel or linear prediction speech synthesizer, and to an excitation signal generator for such a synthesizer.

On connaît plusieurs types de dispositifs d'analyse et de synthèse d'un signal de parole dont les plus connus sont les vocodeurs à canaux, les vocodeurs à formants et les vocodeurs à prédiction linéaire. L'invention concerne la chaîne de synthèse des vocodeurs à canaux et des vocodeurs à prédiction linéaire. Les vocodeurs peuvent être de trois types, soit de type classique, c'est-à-dire avec détection de la fréquence fondamentale du signal de parole, ou encore du type à bande de base, ou encore du type à excitation vocale. Le principe général de tels vocodeurs est illustré sur la figure 1.Several types of devices for analyzing and synthesizing a speech signal are known, the best known of which are channel vocoders, formant vocoders and linear prediction vocoders. The invention relates to the synthesis chain of channel vocoders and linear prediction vocoders. The vocoders can be of three types, either of the conventional type, that is to say with detection of the fundamental frequency of the speech signal, or of the baseband type, or of the voice excitation type. The general principle of such vocoders is illustrated in FIG. 1.

Un vocodeur peut être décomposé en trois parties : un sous-ensemble d'analyse 2, un organe de transmission 4 et un sous-ensemble de synthèse 6. On a vu que le sous-ensemble d'analyse 2 va transformer un signal de parole analogique en un signal numérique. Pour minimiser le débit du signal numérique émis par le sous-ensemble d'analyse 2, on ne réalise pas un codage temporel du signal de parole, comme dans le cas par exemple d'une transmission téléphonique par Modulation d'Impulsion et Codage (MIC), mais on va effectuer un codage fréquentiel du signal de parole. Le débit du signal de parole ainsi codé est de l'ordre de 2 à 12 kbit/s, ce qui est très inférieur au codage MIC qui nécessite 64 kbit/s.A vocoder can be broken down into three parts: an analysis sub-assembly 2, a transmission member 4 and a synthesis sub-assembly 6. We have seen that the analysis sub-assembly 2 will transform a speech signal analog into a digital signal. To minimize the bit rate of the digital signal emitted by the analysis sub-assembly 2, time coding of the speech signal is not carried out, as in the case for example of a telephone transmission by Pulse Modulation and Coding (MIC ), but we will perform a frequency coding of the speech signal. The bit rate of the speech signal thus coded is of the order of 2 to 12 kbit / s, which is much lower than the MIC coding which requires 64 kbit / s.

Le signal de parole est représenté par deux types de paramètres. Un premier type de paramètres qui décrit l'enveloppe spectrale instantanée du signal de parole et un deuxième type de paramètres qui décrit la structure fine du spectre du signal de parole. Cette structure fine est souvent caractérisée par un seul paramètre qui est la valeur de la fréquence fondamentale. Un premier moyen 8 du sous-ensemble d'analyse 2, analyse et code l'enveloppe spectrale instantanée du signal de parole et un deuxième moyen 10 du sous-ensemble d'analyse code la structure fine du spectre du signal de parole.The speech signal is represented by two types of parameters. A first type of parameters which describes the instantaneous spectral envelope of the speech signal and a second type of parameters which describes the fine structure of the spectrum of the speech signal. This fine structure is often characterized by a single parameter which is the value of the fundamental frequency. A first means 8 of the analysis sub-assembly 2 analyzes and codes the instantaneous spectral envelope of the speech signal and a second means 10 of the analysis sub-assembly codes the fine structure of the spectrum of the speech signal.

Les deux signaux numériques obtenus à la sortie du premier moyen 8 et du deuxième moyen 10 du sous-ensemble d'analyse 2 sont ensuite émis dans l'organe de transmission 4, par exemple sous forme de trame. A l'autre extrémité de l'organe de transmission 4, les deux signaux numérisés sont reçus par le sous-ensemble de synthèse 6.The two digital signals obtained at the output of the first means 8 and of the second means 10 of the analysis sub-assembly 2 are then transmitted in the transmission member 4, for example in the form of a frame. At the other end of the transmission member 4, the two digitized signals are received by the synthesis sub-assembly 6.

Le signal numérique codant l'enveloppe spectrale du signal de parole P, est alors appliqué sur l'entrée d'un moyen 12 constitué d'un filtre ou d'un ensemble de filtres pour restituer cette enveloppe spectrale. Le deuxième signal numérique, codé par le deuxième moyen 10 du sous-ensemble d'analyse 2, est appliqué sur l'entrée d'un moyen 14 d'élaboration d'un signal d'excitation qui présente un spectre de structure fine semblable à celle du signal analysé mais d'enveloppe spectrale neutre et qui délivre un signal vers le moyen 12 qui restitue une enveloppe spectrale semblable à celle du signal de parole analysé.The digital signal encoding the spectral envelope of the speech signal P, is then applied to the input of a means 12 consisting of a filter or a set of filters to restore this spectral envelope. The second digital signal, coded by the second means 10 of the analysis sub-assembly 2, is applied to the input of a means 14 for generating an excitation signal which has a fine structure spectrum similar to that of the signal analyzed but of neutral spectral envelope and which delivers a signal to the means 12 which restores a spectral envelope similar to that of the speech signal analyzed.

Ce schéma général d'un vocodeur s'applique en particulier aux trois types de vocodeur précédemment évoqués. Dans le premier et le troisième types de vocodeurs, le premier moyen 8 code l'ensemble de la bande spectrale analysée, comprise par exemple entre 200 Hz et 6 000 Hz et le moyen 10 code un paramètre permettant l'élaboration du signal d'excitation à la réception. Dans le deuxième type de vocodeur, la bande spectrale est découpée en une bande basse, dite bande de base, comprise par exemple entre 200 Hz et 1 000 Hz, et une bande haute. La bande basse est codée selon un procédé temporel classique par le moyen 10 ; à la réception, le signal d'excitation est tiré de la bande de base décodée et la bande haute est codée par le premier moyen 8 comme la bande spectrale totale des vocodeurs du premier type. Le traitement effectué par le troisième type de vocodeur diffère de celui effectué par le second en ce que le signal d'excitation n'est pas déduit de la bande de base décodée, mais est obtenu à partir de paramètres fournis par une analyse temporelle du signal de parole.This general diagram of a vocoder applies in particular to the three types of vocoder previously mentioned. In the first and third types of vocoders, the first means 8 encodes the entire spectral band analyzed, for example between 200 Hz and 6,000 Hz and the means 10 codes a parameter allowing the development of the excitation signal at the reception. In the second type of vocoder, the spectral band is cut into a low band, called the base band, for example between 200 Hz and 1000 Hz, and a high band. The low band is coded according to a conventional temporal method by the means 10; on reception, the excitation signal is taken from the decoded base band and the high band is coded by the first means 8 as the total spectral band of the vocoders of the first type. The processing performed by the third type of vocoder differs from that performed by the second in that the excitation signal is not deduced from the decoded baseband, but is obtained from parameters provided by a temporal analysis of the signal. of speech.

Dans les vocodeurs à bande de base connus, la bande de base a une largeur de l'ordre de 800 Hz en général. En effet, il est nécessaire que la bande de base contienne la fréquence fondamentale du signal de parole et plusieurs de ses harmoniques afin d'engendrer un signal d'excitation comportant toutes les fréquences harmoniques de la fréquence fondamentale qui sont contenues dans la bande spectrale traitée. De plus, dans les vocodeurs à bande de base connus, l'élaboration du signal d'excitation nécessite des traitements sophistiqués d'égalisation de spectre pour délivrer un signal d'excitation à peu près plat, c'est-à-dire un signal d'excitation dont l'enveloppe spectrale instantanée ne présente pas d'ondulation trop forte. Les vocodeurs à bande de base connus présentent donc les inconvénients d'une part d'un débit de données élevé entre le sous-ensemble d'analyse et le sous-ensemble de synthèse à cause de la largeur de la bande de base à transmettre et d'autre part d'un prix de revient élevé à cause des traitements sophistiqués effectués pour engendrer le signal d'excitation.In known baseband vocoders, the baseband is generally around 800 Hz in width. Indeed, it is necessary that the baseband contains the fundamental frequency of the speech signal and several of its harmonics in order to generate an excitation signal comprising all the harmonic frequencies of the fundamental frequency which are contained in the spectral band processed. . In addition, in known baseband vocoders, the development of the excitation signal requires sophisticated spectrum equalization treatments to deliver an approximately flat excitation signal, i.e. a signal excitation whose instantaneous spectral envelope does not have too strong ripple. Known baseband vocoders therefore have the drawbacks, on the one hand, of a high data rate between the analysis sub-assembly and the synthesis sub-assembly because of the width of the base band to be transmitted and on the other hand a high cost price because of the sophisticated processing carried out to generate the excitation signal.

On sait par l'article « ASPIC : Analyseur synthétiseur de parole à informations codées (système CNET) de G. Ferrieu et al. paru dans la revue l'Onde Electrique, vol. 49, n° 504, mars 1969, p. 376-377, Paris que le signal d'excitation peut être produit à partir d'une séquence pseudo-aléatoire. L'auteur ne décrit pas explicitement de moyens pour délivrer cette séquence pseudo-aléatoire, mais souligne la difficulté et la complexité de la méthode mettant en oeuvre un détecteur demélo- die (cf 4.1.5). On connaît, par ailleurs, par le document DE-A-2 435 057 une structure de générateur de séquence pseudo-aléatoire qui peut être réinitialisé en fonction du signal de sortie dudit générateur.We know from the article “ASPIC: Speech synthesizer with coded information (CNET system) by G. Ferrieu et al. published in the review l'Onde Electrique, vol. 49, n ° 504, March 1969, p. 376-377, Paris that the excitation signal can be produced from a pseudo-random sequence. The author does not explicitly describe the means to deliver this pseudo-random sequence, but emphasizes the difficulty and complexity of the method using a demediodic detector (see 4.1.5). Furthermore, document DE-A-2 435 057 discloses a pseudo-random sequence generator structure which can be reset as a function of the output signal of said generator.

L'invention remédie à ces inconvénients. Elle réside dans un procédé d'élaboration du signal d'excitation obtenu sans traitements sophistiqués, ce qui est une amélioration importante en particulier dans le cas de vocodeurs à bande de base, ou à excitation vocale.The invention overcomes these drawbacks. It resides in a method for developing the excitation signal obtained without sophisticated processing, which is a significant improvement in particular in the case of baseband vocoders, or with vocal excitation.

Parmi les autres avantages apportés par la présente invention, il convient de noter que le signal d'excitation présente une énergie sensiblement constante qui ne dépend pas de la valeur de la fréquence fondamentale. En outre, le signal d'excitation présente une énergie nettement supérieure à l'énergie d'un signal d'excitation constitué d'impulsions uniques espacées de la période fondamentale. Ceci a pour avantage d'améliorer le rapport signal/bruit du signal synthétisé.Among the other advantages provided by the present invention, it should be noted that the excitation signal has a substantially constant energy which does not depend on the value of the fundamental frequency. In addition, the excitation signal has an energy significantly higher than the energy of an excitation signal consisting of single pulses spaced from the fundamental period. This has the advantage of improving the signal / noise ratio of the synthesized signal.

D'autre part, le signal synthétique obtenu fournit une impression auditive plus naturelle (moins métallique), si on effectue la comparaison par rapport à ce qu'on obtient avec un signal d'excitation constitué d'impulsions espacées de la période fondamentale.On the other hand, the synthetic signal obtained provides a more natural auditory impression (less metallic), if the comparison is made with what is obtained with an excitation signal consisting of pulses spaced from the fundamental period.

De manière plus précise, l'invention a pour objet un procédé d'élaboration d'un signal d'excitation pour synthétiseur de parole à canaux ou à prédiction linéaire, dans lequel on délivre un signal constitué d'une séquence d'impulsions pseudo-aléatoire prédéterminée, et dans lequel on réinitialise ladite séquence lorsque survient un événement déterminé dont la fréquence de répétition est liée à la fréquence fondamentale du signal de parole, l'amplitude du signal d'excitation ayant une valeur fixe prédéterminée et la polarité dudit signal d'excitation étant déterminée par le niveau logique des impulsions de la séquence pseudo-aléatoire.More specifically, the subject of the invention is a method of developing an excitation signal for a channel synthesizer or a linear prediction, in which a signal consisting of a sequence of pseudo- pulses is delivered. predetermined random frequency, and in which said sequence is reset when a determined event occurs whose repetition frequency is linked to the fundamental frequency of the speech signal, the amplitude of the excitation signal having a predetermined fixed value and the polarity of said signal d excitation being determined by the logical level of the pulses of the pseudo-random sequence.

Selon une autre caractéristique secondaire du procédé selon l'invention, dans le cas d'un vocodeur à détection et à mesure de la fréquence fondamentale, on reçoit à la synthèse un signal représentatif de la valeur de la fréquence fondamentale et on engendre, en rapport avec l'occurrence de l'événement déterminé, un signal de réinitialisation de la séquence d'impulsions pseudo-aléatoire, la fréquence de ce signal étant liée à la valeur reçue de ladite fréquence fôndamentale.According to another secondary characteristic of the method according to the invention, in the case of a vocoder with detection and measurement of the fundamental frequency, a signal representative of the value of the fundamental frequency is received at the synthesis and is generated, in relation with the occurrence of the determined event, a signal to reset the pseudo-random pulse sequence, the frequency of this signal being linked to the value received from said basic frequency.

Selon une autre caractéristique secondaire du procédé selon l'invention, dans le cas d'un vocodeur à bande de base, l'événement déterminé est constitué par le passage du signal de la bande de base reçue par la valeur zéro dans un sens déterminé, ou par le passage du signal de la bande de base par un extrémum déterminé.According to another secondary characteristic of the method according to the invention, in the case of a baseband vocoder, the determined event is constituted by the passage of the signal of the baseband received by the value zero in a determined direction, or by the passage of the baseband signal through a determined extremum.

L'invention a aussi pour objet un générateur de signal d'excitation pour synthétiseur de parole, comportant des moyens générateurs pour délivrer une séquence d'impulsions pseudo-aléatoire prédéterminée, des moyens pour réinitialiser les moyens générateurs en réponse à la réception d'un signal de commande, des moyens de production du signal d'excitation sensibles au niveau logique des impulsions de ladite séquence pour engendrer un signal d'excitation d'amplitude fixe prédéterminée et de polarité déterminée par le niveau logique de ladite séquence et des moyens de détection d'un événement déterminé dont la fréquence est liée à la fréquence fondamentale du signal de parole, lesdits moyens de détection commandant lesdits moyens de réinitialisation.Another subject of the invention is an excitation signal generator for a speech synthesizer, comprising generator means for delivering a predetermined pseudo-random pulse sequence, means for resetting the generator means in response to the reception of a control signal, means for producing the excitation signal sensitive to the logic level of the pulses of said sequence to generate an excitation signal of predetermined fixed amplitude and polarity determined by the logic level of said sequence and detection means of a determined event whose frequency is linked to the fundamental frequency of the speech signal, said detection means controlling said reset means.

Selon un mode de réalisation préféré du générateur de signal d'excitation selon l'invention, les moyens générateurs comportent un registre à décalage, au moins une porte NON-OU-EXCLUSIF dont les entrées sont reliées à deux étages de sortie du registre, la sortie de ladite porte étant reliée à l'entrée de données du registre, ce dernier comportant en outre une entrée de réinitialisation reliée à la sortie des moyens de réinitialisation.According to a preferred embodiment of the excitation signal generator according to the invention, the generator means comprise a shift register, at least one NON-OR-EXCLUSIVE gate whose inputs are connected to two output stages of the register, the output of said door being connected to the data input of the register, the latter further comprising a reset input connected to the output of the reset means.

Selon une caractéristique secondaire du générateur de signal d'excitation selon l'invention, les moyens générateurs comportent, dans le cas d'un synthétiseur à canaux, un inverseur logique dont l'entrée est reliée à la sortie de la porte NON-OU-EXCLUSIF.According to a secondary characteristic of the excitation signal generator according to the invention, the generator means comprise, in the case of a channel synthesizer, a logic inverter whose input is connected to the output of the NOR-OR gate. EXCLUSIVE.

D'autres caractéristiques et avantages de l'invention ressortiront mieux de la description qui va suivre, donnée à titre illustratif mais non limitatif, en référence aux figures annexées dans lesquelles :

  • la figure 1, déjà décrite, représente de façon schématique la structure générale d'un vocodeur,
  • la figure 2 représente un mode de réalisation du générateur de signal d'excitation selon la présente invention,
  • la figure 3 représente un mode de réalisation d'une variante du générateur selon la figure 2,
  • la figure 4a représente schématiquement les moyens d'un sous-ensemble d'analyse d'un vocodeur à bande de base, et la figure 4b représente schématiquement les moyens d'un sous-ensemble de synthèse correspondant, mettant en oeuvre le générateur de signal d'excitation selon les figures 2 ou 3,
  • la figure 5a représente schématiquement les moyens d'un sous-ensemble d'analyse d'un vocodeur à excitation vocale, et la figure 5b représente schématiquement les moyens d'un sous-ensemble de synthèse correspondant, mettant en oeuvre le générateur de signal d'excitation selon les figures 2 ou 3.
Other characteristics and advantages of the invention will emerge more clearly from the description which follows, given by way of illustration but not limitation, with reference to the appended figures in which:
  • FIG. 1, already described, schematically represents the general structure of a vocoder,
  • FIG. 2 represents an embodiment of the excitation signal generator according to the present invention,
  • FIG. 3 represents an embodiment of a variant of the generator according to FIG. 2,
  • FIG. 4a diagrammatically represents the means of a subset for analyzing a baseband vocoder, and FIG. 4b diagrammatically represents the means of a corresponding synthesis subset, implementing the signal generator excitation according to FIGS. 2 or 3,
  • FIG. 5a schematically represents the means of a subset of analysis of a vocoder with vocal excitation, and FIG. 5b schematically represents the means of a corresponding synthesis subset, implementing the signal generator d excitation according to Figures 2 or 3.

La figure 2 représente le noyau central du générateur de signal d'excitation selon la présente invention. Ce générateur 40 comprend un registre à décalage 46 dont les sorties de deux des étages Q7 et Q9 sont rebouclées sur son entrée de données D par l'intermédiaire d'une porte NON-OU-EXCLUSIF 48. Le registre à décalage est cadencé par un signal d'horloge HOR appliqué sur son entrée H, et comporte également une entrée de réinitialisation R, pour assurer la remise à zéro du registre en réponse à l'application d'un signal de réinitialisation. Le générateur 40 comporte également une porte ET 56 dont une entrée reçoit le signal d'horloge précédemment mentionné, et deux bascules 58 et 60 de type D qui sont montées de la manière suivante : l'entrée d'horloge de la bascule 58 est reliée à la sortie de la porte 56 et sa sortie Q est rebouclée sur sa borne d'entrée D et reliée à l'entrée d'horloge de la bascule 60. La sortie Q de cette dernière est également rebouclée sur sa borne d'entrée D et aussi sur l'une des entrées de la porte ET 56. Les entrées de remise à « 1 RAU (Set en terminologie anglo-saxonne) des bascules 58 et 60 sont reliées en permanence à la masse, et les entrées de remise à « 0 » RAZ (Reset en terminologie anglo-saxonne) sont reliées à une ligne destinée à recevoir un signal de commande qui a une période de répétition correspondant à la fréquence du fondamental du signal de parole.FIG. 2 represents the central core of the excitation signal generator according to the present invention. This generator 40 comprises a shift register 46 whose outputs of two of the stages Q7 and Q9 are looped back to its data input D via a NOR gate 48. The shift register is clocked by a HOR clock signal applied to its input H, and also includes a reset input R, to ensure the reset of the register in response to the application of a reset signal. The generator 40 also includes an AND gate 56, one of which input receives the aforementioned clock signal, and two flip-flops 58 and 60 of type D which are mounted in the following manner: the clock input of flip-flop 58 is connected to the output of gate 56 and its output Q is looped back to its input terminal D and connected to the clock input of the flip-flop 60. The output Q of the latter is also looped back to its input terminal D and also to one of the inputs of the AND gate 56. The reset inputs to "1 RAU (Set in English terminology) of flip-flops 58 and 60 are connected permanently to ground, and the reset inputs to “0” RESET (Reset in English terminology) are connected to a line intended to receive a control signal which has a repetition period corresponding to the frequency of the fundamental of the signal of speech.

La sortie Q de la bascule 58 est reliée à l'entrée R du registre 46. On a également prévu un inverseur 54 dont l'entrée reçoit le signal de sortie de la porte 48. Comme il sera expliqué plus en détail dans la suite de la description, les signaux d'entrée S1 et de sortie S2 de l'inverseur 54 sont en opposition de phase, et sont destinés à produire respectivement les signaux d'excitation pour les canaux impairs et pairs du sous-ensemble de synthèse dans le cas d'un synthétiseur à canaux. Le signal d'horloge HOR reçu par le registre à décalage 46 a, dans le cas d'utilisation d'un synthétiseur numérique, la même fréquence que la fréquence d'échantillonnage des filtres numériques de synthèse, par exemple une fréquence de 16 kHz. Les deux bascules 58 et 60 ont pour rôle de synchroniser les impulsions de réinitialisation appliquées au registre à décalage 46, avec les impulsions d'horloge.The output Q of the flip-flop 58 is connected to the input R of the register 46. An inverter 54 has also been provided, the input of which receives the output signal from the gate 48. As will be explained in more detail below. the description, the input signals S1 and output S2 of the inverter 54 are in phase opposition, and are intended to produce respectively the excitation signals for the odd and even channels of the synthesis subset in the case a channel synthesizer. The clock signal HOR received by the shift register 46 has, in the case of using a digital synthesizer, the same frequency as the sampling frequency of the digital synthesis filters, for example a frequency of 16 kHz. The two flip-flops 58 and 60 have the role of synchronizing the reset pulses applied to the shift register 46, with the clock pulses.

Le générateur 40 fonctionne de la façon suivante : lorsque le signal appliqué sur l'entrée RAZ des bascules 58 et 60 est à « 1 •, les sorties Q de ces bascules sont maintenues à « 1 •. Quand l'entrée RAZ des bascules 58 et 60 reçoit un « 0 •, la remise à zéro de ces bascules n'est plus imposée, et sur le premier front montant du signal d'horloge HOR qui suit cet événement, la sortie Q de la bascule 58 passe à l'état « 0 » étant donné le rebouclage de à sur l'entrée D. La mise à zéro de la sortie Q de la bascule 58 agit au niveau de l'entrée de réinitialisation R, ce qui met les sorties Q du registre 46 à zéro, c'est-à-dire, ce qui réinitialise la séquence d'impulsions pseudo-aléatoire délivrée par le système registre à décalage porte NON-OU-EXCLUSIF. Sur le front montant suivant du signal d'horloge HOR, la sortie Q de la bascule 58 passe à « 1 puisque son entrée D est à « 0 •, valeur précédente sur la sortie Q. Cette transition agit sur l'entrée d'horloge de la bascule 60 et la sortie Q de cette dernière passe à zéro. On obtient donc avec ces deux bascules une impulsion sur l'entrée R du registre 46, qui est en phase avec le signal d'horloge HOR et dont la largeur est égale à la durée séparant deux impulsions d'horloge. Entre deux présences d'une impulsion de réinitialisation, on a le fonctionnement suivant : la porte 48 délivre un « 1 qui va être transféré dans les différents étages du registre au rythme de l'horloge. Ce niveau « 1 est maintenu jusqu'à ce qu'un « 1 » se propage jusqu'à un étage n dont la sortie Qn est reliée à l'entrée de la porte 48 (l'étage Q7 si on considère la figure 2). A ce moment, l'état logique change en sortie de la porte 48, et dans la suite de la séquence, cette porte délivre une suite d'impulsions de largeur pseudo-aléatoire. On notera que, bien que l'on ait représenté sur la figure 2 des connexions entre les sorties des étages Q7 et Q9 du registre 46 et les entrées de la porte 48, d'autres alternatives sont possibles pour le choix des étages de sortie. Néanmoins, on connectera de préférence aux entrées de la porte 48, des sorties d'étages non consécutifs du registre 46. La sortie de la porte 48 délivre un signal dont le niveau logique détermine la polarité du signal d'excitation de la chaîne de synthèse (ce signal aura une amplitude constante en valeur absolue). Etant donné que les moyens de production du signal d'excitation ainsi définis peuvent être réalisés par mise en oeuvre de techniques classiques, on n'a pas représenté ces moyens sur les figures 2 et 3. Ce signal d'excitation est dirigé vers l'entrée du filtre prédicteur de synthèse (pour un synthétiseur à prédiction linéaire), ou vers les entrées des filtres passe-bande des canaux de synthèse (dans le cas d'un vocodeur à canaux). Dans ce dernier cas, l'inverseur 54 intervient de la manière suivante : le signal S1 sur l'entrée de ce dernier est utilisé pour engendrer le signal d'excitation des canaux de synthèse de rang impair, tandis que le signal S2 sur sa sortie est utilisé pour engendrer le signal d'excitation des canaux de rang pair. Cette mesure permet de compenser les décalages de phase entre canaux dans le cas usuel où les filtres des canaux de synthèse sont du quatrième ordre et se recoupent à -6 dB et pratiquement en opposition de phase. On obtient ainsi une réponse en fréquence pratiquement plate sur toute la bande de fréquence ainsi traitée.The generator 40 operates as follows: when the signal applied to the reset input of flip-flops 58 and 60 is at "1 •, the outputs Q of these scales are kept at "1 •. When the RESET input of flip-flops 58 and 60 receives a “0 •, the reset to zero of these flip-flops is no longer imposed, and on the first rising edge of the clock signal HOR which follows this event, the output Q of flip-flop 58 goes to state “0” given the looping back from to input D. The zeroing of output Q of flip-flop 58 acts at the level of reset input R, which puts the outputs Q of register 46 to zero, that is to say, which reinitializes the pseudo-random pulse sequence delivered by the gate shift register system NON-OR-EXCLUSIVE. On the next rising edge of the HOR clock signal, the output Q of flip-flop 58 changes to "1 since its input D is at" 0 •, previous value on output Q. This transition acts on the clock input of flip-flop 60 and the output Q of the latter goes to zero. One thus obtains with these two flip-flops a pulse on the entry R of the register 46, which is in phase with the clock signal HOR and whose width is equal to the duration separating two clock pulses. Between two presences of a reset pulse, there is the following operation: the gate 48 delivers a "1 which will be transferred in the different stages of the register at the rate of the clock. This level "1 is maintained until a" 1 "propagates to a stage n whose output Qn is connected to the input of gate 48 (stage Q7 if we consider FIG. 2) . At this moment, the logic state changes at the output of gate 48, and in the remainder of the sequence, this gate delivers a series of pulses of pseudo-random width. It will be noted that, although FIG. 2 shows connections between the outputs of stages Q7 and Q9 of register 46 and the inputs of gate 48, other alternatives are possible for the choice of output stages. Nevertheless, it is preferable to connect to the inputs of gate 48, outputs of non-consecutive stages of register 46. The output of gate 48 delivers a signal whose logic level determines the polarity of the excitation signal of the synthesis chain (this signal will have a constant amplitude in absolute value). Given that the means of producing the excitation signal thus defined can be produced by implementing conventional techniques, these means have not been shown in FIGS. 2 and 3. This excitation signal is directed towards the input of the synthesis predictor filter (for a linear prediction synthesizer), or to the inputs of the bandpass filters of the synthesis channels (in the case of a channel vocoder). In the latter case, the inverter 54 intervenes as follows: the signal S1 on the input of the latter is used to generate the excitation signal of the synthesis channels of odd rank, while the signal S2 on its output is used to generate the excitation signal of even rank channels. This measurement makes it possible to compensate for phase shifts between channels in the usual case where the filters of the synthesis channels are of the fourth order and overlap at -6 dB and practically in phase opposition. An essentially flat frequency response is thus obtained over the entire frequency band thus treated.

Etant donné que la réinitialisation se fait en réponse à la réception sur les entrées RAZ d'un signal qui a une période de répétition correspondant à la fréquence fondamentale du signal de parole, et que la séquence d'impulsions pseudo-aléatoire se déroule toujours de la même façon après sa réinitialisation, il en ressort que cette séquence présente aussi une périodicité correspondant à la fréquence fondamentale. Etant donné que cette séquence est constituée d'une suite d'impulsions, elle présente donc un spectre de fréquences large, et en moyenne plat comportant tous les harmoniques de la fréquence fondamentale nécessaires au bon fonctionnement d'un synthétiseur de parole. En l'absence de fréquence fondamentale, la réinitialisation se fait de manière aléatoire et la séquence a un spectre de bruit blanc.Since the reset is done in response to the reception on the RESET inputs of a signal which has a repetition period corresponding to the fundamental frequency of the speech signal, and that the pseudo-random pulse sequence always proceeds from the same way after its reset, it appears that this sequence also has a periodicity corresponding to the fundamental frequency. Since this sequence is made up of a series of pulses, it therefore has a broad frequency spectrum, and on average flat, comprising all the harmonics of the fundamental frequency necessary for the proper functioning of a speech synthesizer. In the absence of a fundamental frequency, the reset is done randomly and the sequence has a white noise spectrum.

On a représenté sur la figure 3 une variante du mode de réalisation du générateur de la figure 2 permettant d'obtenir un signal d'excitation plus régulier et moins bruité pour les premiers canaux (dans le cas d'un synthétiseur à canaux) du vocodeur (canaux recouvrant la bande spectrale jusqu'à 1 000 Hz environ). Ce signal d'excitation permet notamment une meilleure restitution des sons voisés.There is shown in Figure 3 a variant of the embodiment of the generator of Figure 2 to obtain a more regular and less noisy excitation signal for the first channels (in the case of a channel synthesizer) vocoder (channels covering the spectral band up to approximately 1000 Hz). This excitation signal allows in particular a better reproduction of the voiced sounds.

Ce générateur comprend les mêmes éléments que celui de la figure 2 qui sont désignés par les mêmes références numériques. On y a ajouté une bascule 62 dont l'entrée de remise à « 0 » (RAZ) est reliée à la sortie Q de la bascule 58, et l'entrée de remise à « 1 » (RAU) reçoit le signal de sortie S2 de l'inverseur logique 54. Les entrées de données D et d'horloge H sont mises à zéro par liaison à la masse. De la même manière que décrit précédemment en relation avec les signaux S1 et S2, la sortie Q de cette bascule délivre un signal S01 qui sert à produire le signal d'excitation des premiers canaux de rang impair, par exemple de rangs 1 et 3 et la sortie à délivre un signal S02 qui sert à produire le signal d'excitation des premiers canaux de rang pair, par exemple de rang 2 et 4. Avec ce montage, la sortie Q de la bascule 62 délivre une séquence de p « 1 » successifs, la longueur de cette séquence étant égale à la longueur totale de l'impulsion de niveau « 1 », en sortie de la porte 48, suivant immédiatement la réinitialisation. Par la suite, la sortie Q de la bascule 62 reste à zéro jusqu'à la réinitialisation suivante. On constate que la bascule 62 se comporte de la même manière qu'une bascule bistable classique.This generator comprises the same elements as that of FIG. 2 which are designated by the same reference numbers. A flip-flop 62 has been added to it, the reset input at “0” (RESET) being connected to the output Q of flip-flop 58, and the reset input at “1” (RAU) receiving the output signal S2 of the logic inverter 54. The data inputs D and clock H are reset to zero by connection to ground. In the same manner as described previously in relation to the signals S1 and S2, the output Q of this flip-flop delivers a signal S 01 which is used to produce the signal of excitation of the first channels of odd rank, for example of ranks 1 and 3 and the output delivers a signal S 02 which is used to produce the excitation signal of the first channels of even rank, for example of rank 2 and 4. With this arrangement, the output Q of the flip-flop 62 delivers a sequence of p " 1 "successive, the length of this sequence being equal to the total length of the pulse of level" 1 ", at the output of gate 48, immediately following the reset. Thereafter, the Q output of flip-flop 62 remains at zero until the next reset. It can be seen that the flip-flop 62 behaves in the same way as a conventional flip-flop.

Selon le type de vocodeur mis en oeuvre, les impulsions transmises aux entrées RAZ des bascules 58 et 60, et dont la période de répétition correspond à la fréquence du fondamental du signal de parole, sont engendrées de la manière suivante :Depending on the type of vocoder used, the pulses transmitted to the reset inputs of flip-flops 58 and 60, and whose repetition period corresponds to the frequency of the fundamental of the speech signal, are generated in the following manner:

1) Vocodeur avec détection de la fréquence fondamentale1) Vocoder with fundamental frequency detection

On détermine, à la synthèse, des impulsions qui sont espacées d'une période de temps T correspondant à la fréquence fondamentale fo transmise (pour des sons voisés), soit T = 1/fo. En l'absence de détection de fondamental, la remise à zéro est commandée par des impulsions espacées de longueurs aléatoires ou encore on peut supprimer la réinitialisation.On synthesis, pulses are determined which are spaced by a period of time T corresponding to the fundamental frequency f o transmitted (for voiced sounds), ie T = 1 / fo. In the absence of fundamental detection, the reset to zero is controlled by pulses spaced apart by random lengths or else the reset can be suppressed.

2) Vocodeur à bande de base2) Baseband vocoder

Sur la figure 4a, on a représenté une partie du sous-ensemble d'analyse d'un vocodeur à bande de base. Le moyen 10 comprend un filtre passe-bas 16 qui ne laisse passer que la bande de base de la bande spectrale du signal de parole P et un moyen de codage 18 qui effectue un codage numérique du signal analogique délivré par le filtre passe-bas 16. La largeur de la bande de base est par exemple de l'ordre de 500 Hz. Le moyen de codage 18 réalise un codage temporel de manière connue.In FIG. 4a, a part of the analysis sub-assembly of a baseband vocoder has been shown. The means 10 comprises a low-pass filter 16 which allows only the base band of the spectral band of the speech signal P to pass and a coding means 18 which performs digital coding of the analog signal delivered by the low-pass filter 16 The width of the baseband is for example of the order of 500 Hz. The coding means 18 performs time coding in a known manner.

Le signal numérique délivré par le moyen 10 est mis en trame avec le signal numérique délivré par le moyen 8 correspondant aux données des canaux d'analyse par un moyen de mise en trame 20. La trame est ensuite émise par un organe de transmission 22 sur une ligne de transmission L.The digital signal delivered by the means 10 is framed with the digital signal delivered by the means 8 corresponding to the data of the analysis channels by a framing means 20. The frame is then transmitted by a transmission device 22 on a transmission line L.

A l'extrémité de cette ligne de transmission L (figure 4b), la trame est appliquée sur l'entrée d'un organe de réception 24. Celui-ci est relié en série à un moyen de séparation 26 qui délivre les données canaux au moyen 12 d'une part, et le signal numérique codant la bande de base au moyen 14 d'élaboration du signal d'excitation d'autre part.At the end of this transmission line L (FIG. 4b), the frame is applied to the input of a reception member 24. This is connected in series to a separation means 26 which delivers the channel data to the means 12 on the one hand, and the digital signal encoding the baseband by means 14 of developing the excitation signal on the other hand.

Ce moyen 14 comprend en série un moyen de retard 28, un moyen de décodage 30, un filtre passe-bas 32, un additionneur 34 dont une entrée est reliée à un générateur de bruit 36, un moyen de détection 38 et un générateur de signal d'excitation 40 du type selon l'invention et précédemment décrit dans les figures 2 et 3.This means 14 comprises in series a delay means 28, a decoding means 30, a low-pass filter 32, an adder 34, an input of which is connected to a noise generator 36, a detection means 38 and a signal generator excitation 40 of the type according to the invention and previously described in FIGS. 2 and 3.

Le moyen de retard 28 a pour fonction de compenser le retard des données canaux sur le signal de la bande de base causé par les filtres passe-bas des canaux d'analyse et la transmission des données canaux. Ce retard est ajusté empiriquement. Il est de l'ordre de 20 à 30 ms. Le signal délivré par le moyen 28 est ensuite décodé par le moyen 30 qui restitue un signal analogique. Ce moyen 30 effectue la transformation inverse de celle réalisée par le moyen 18 de la figure 2a. Il est suivi d'un filtre passe-bas 32, semblable au filtre passe-bas 16, à la sortie duquel on retrouve le signal de la bande de base. Ce signal peut être bruité grâce à un additionneur 34 recevant le signal de la bande de base sur une entrée et un bruit engendré par un générateur de bruit blanc 36 sur une autre entrée. Cet apport de bruit est facultatif. Il est intéressant en ce qu'il permet, lorsque la bande de base ne transporte pas d'énergie, d'appliquer sur l'entrée du moyen de détection 38 un signal dont les passages par zéro sont à peu près aléatoires, passages par zéro à partir desquels on peut engendrer un signal d'excitation aléatoire. Le bruit ajouté est faible, son niveau peut être par exemple de - 60 dB par rapport au niveau maximum de la bande de base.The delay means 28 has the function of compensating for the delay of the channel data on the baseband signal caused by the low pass filters of the analysis channels and the transmission of the channel data. This delay is adjusted empirically. It is around 20 to 30 ms. The signal delivered by the means 28 is then decoded by the means 30 which restores an analog signal. This means 30 performs the reverse transformation from that carried out by the means 18 of FIG. 2a. It is followed by a low-pass filter 32, similar to the low-pass filter 16, at the output of which the baseband signal is found. This signal can be noisy thanks to an adder 34 receiving the baseband signal on one input and a noise generated by a white noise generator 36 on another input. This addition of noise is optional. It is advantageous in that it allows, when the baseband does not transport energy, to apply to the input of the detection means 38 a signal whose zero crossings are almost random, zero crossings from which a random excitation signal can be generated. The added noise is low, its level can be for example - 60 dB compared to the maximum level of the baseband.

Le moyen de détection 38 a pour fonction de détecter les occurrences d'un événement déterminé dans la bande de base et d'émettre une impulsion à chacune de ces occurrences. Cet événement déterminé, dont la fréquence de répétition correspond à la période fondamentale du signal de parole, peut être le passage par zéro dans un sens déterminé du signal de la bande de base, la détection étant alors réalisée simplement par un comparateur. On peut également choisir comme événement facilement détectable, par exemple, le passage du signal de la bande de base par un extrémum déterminé. Les impulsions délivrées par le moyen de détection 38 à chacune desdites occurrences sont appliquées sur les entrées RAZ des bascules 58 et 60 du générateur 40. La sortie de la séquence d'impulsions pseudo-aléatoire est rythmée par le signal d'horloge HOR et réinitialisée lorsqu'une impulsion est délivrée en sortie du moyen 38.The detection means 38 has the function of detecting the occurrences of a determined event in the baseband and of emitting a pulse at each of these occurrences. This determined event, the repetition frequency of which corresponds to the fundamental period of the speech signal, may be the zero crossing in a determined direction of the baseband signal, the detection then being carried out simply by a comparator. It is also possible to choose as an easily detectable event, for example, the passage of the baseband signal through a determined extremum. The pulses delivered by the detection means 38 to each of said occurrences are applied to the reset inputs of flip-flops 58 and 60 of the generator 40. The output of the pseudo-random pulse sequence is paced by the clock signal HOR and reset when a pulse is delivered at the output of the means 38.

3) Vocodeurs à excitation vocale3) Vocoders with vocal excitation

Les vocodeurs à excitation vocale diffèrent des vocodeurs à bande de base en ce que la bande de base n'est pas transmise. Dans les vocodeurs à excitation vocale, la détection des occurrences de l'événement déterminé évoqué précédemment se fait donc dans le sous-ensemble d'analyse du vocodeur.Voice-activated vocoders differ from baseband vocoders in that the baseband is not transmitted. In vocoders with vocal excitation, the detection of the occurrences of the determined event mentioned above is therefore done in the subset of vocoder analysis.

La figure 5a représente une partie d'un sous-ensemble d'analyse d'un vocodeur à excitation vocale. Le moyen 10 comprend en série, un filtre passe-bas 32 recevant en entrée le signal de parole P, un additionneur 34 recevant sur une entrée un signal de bruit issu d'un générateur de bruit blanc 36, un moyen de détection 38 délivrant une impulsion à chaque occurrence de l'événement déterminé et un moyen de codage 42.FIG. 5a represents part of an analysis sub-assembly of a vocoder with vocal excitation. The means 10 comprises in series, a low-pass filter 32 receiving the speech signal P as an input, an adder 34 receiving on a input a noise signal coming from a white noise generator 36, a detection means 38 delivering a pulse at each occurrence of the determined event and a coding means 42.

Ce moyen de codage 42 code les dates d'apparition de l'événement déterminé. Ce codage peut être simple et. très peu coûteux en débit. Par exemple, si F est la fréquence supérieure de la bande de base, on découpe le temps en intervalles de T secondes où T = 1/F. Dans chaque intervalle de temps, l'événement déterminé ne peut donc apparaître qu'au plus une fois. Si chaque intervalle de temps de longueur T est découpé à son tour en 2n - 1 sous-intervalles numérotés de 0 à 2n - 2 où n est un nombre entier, on peut repérer la date d'apparition de l'événement déterminé en transmettant le numéro du sous-intervalle de temps correspondant. On connaît ainsi la date d'apparition de l'événement déterminé à T/2" - 1 près, cette précision étant ajustable avec la valeur de n. Si l'événement n'apparaît pas dans l'intervalle de temps de longueur T considéré, il suffit de transmettre le nombre 2n - 1. Ce moyen de codage simple à mettre en oeuvre a l'avantage de ne nécessiter qu'un faible débit égal à n.F bit/seconde.This coding means 42 codes the dates of appearance of the determined event. This coding can be simple and. very inexpensive in throughput. For example, if F is the upper frequency of the baseband, the time is divided into intervals of T seconds where T = 1 / F. In each time interval, the determined event can therefore only appear at most once. If each time interval of length T is divided in turn into 2 n - 1 sub-intervals numbered from 0 to 2 n - 2 where n is an integer, one can locate the date of appearance of the event determined in transmitting the number of the corresponding time sub-interval. We thus know the date of appearance of the event determined to T / 2 "- 1 close, this precision being adjustable with the value of n. If the event does not appear in the time interval of length T considered , it suffices to transmit the number 2n - 1. This coding means which is simple to implement has the advantage of requiring only a low bit rate equal to nF bit / second.

Par exemple avec F = 500Hz et n = 5, le débit est de 2500 bits/s, ce qui est très inférieur au débit nécessaire pour transmettre la bande de base des vocodeurs à bande de base. Par ailleurs, la précision de localisation de l'événement déterminé est de 1/31 1/500 = 64 N.s, ce qui est suffisant.For example with F = 500Hz and n = 5, the bit rate is 2500 bits / s, which is much lower than the bit rate necessary to transmit the baseband of the baseband vocoders. Furthermore, the location accuracy of the determined event is 1/31 1/500 = 64 N .s, which is sufficient.

Le signal délivré par le moyen de codage 42 est mis en trame avec les données d'analyse issues du premier moyen 8 dans un moyen 20 de mise en trame. Un organe de transmission 22 délivre ensuite la trame sur la ligne de transmission L.The signal delivered by the coding means 42 is framed with the analysis data coming from the first means 8 in a framing means 20. A transmission member 22 then delivers the frame on the transmission line L.

Dans le sous-ensemble de synthèse du vocodeur à excitation vocale dont une partie est représentée schématiquement sur la figure 5b, le signal transmis est reçu par un organe de réception 24. Celui-ci est relié à un moyen de séparation 26 qui délivre les données d'analyse au moyen 12 d'une part et le signal codant les dates d'occurrences de l'événement déterminé au moyen 14 d'élaboration du signal d'excitation d'autre part.In the voice excitation vocoder synthesis subset, part of which is shown diagrammatically in FIG. 5b, the transmitted signal is received by a reception member 24. This is connected to a separation means 26 which delivers the data analysis by means 12 on the one hand and the signal coding the dates of occurrences of the event determined by means 14 of development of the excitation signal on the other hand.

Ce moyen 14 comprend un moyen de décodage 44 recevant en entrée toutes les T secondes, un nombre déterminant la date d'occurrence de l'événement déterminé, ou son absence, dans ledit intervalle de temps T et délivrant une impulsion à chacune desdites occurrences, cette impulsion étant transmise aux entrées RAZ des bascules 58 et 60 du générateur de signal d'excitation 40.This means 14 comprises a decoding means 44 receiving as input every T seconds, a number determining the date of occurrence of the determined event, or its absence, in said time interval T and delivering an impulse to each of said occurrences, this pulse being transmitted to the reset inputs of flip-flops 58 and 60 of the excitation signal generator 40.

Claims (12)

1. Process for the generation of an excitation signal for a speech synthesizer provided with channels or linear prediction, from a signal formed by a predetermined pseudo random sequence of pulses, characterized in that the said sequence is reinitialized when a specified event occurs, the frequency of repetition of which event is linked to the fundamental frequency of the speech signal, the amplitude of the excitation signal having a predetermined fixed value and the polarity of the said excitation signal being determined by the logic level of the pulses of the pseudo-random sequence.
2. Process for the generation of an excitation signal according to claim 1, characterized in that, in the case of a synthesizer provided with channels, the pseudo-random sequence of pulses is presented in the form of a first pulse train and a second pulse train (S1, S2) formed simultaneously and in phase opposition, one of the pulse trains (S1) being allocated to the odd channel, the other pulse train (S2) being allocated to the even channels.
3. Process for the generation of an excitation signal according to either one of claims 1 and 2, characterized in that, in the case of a vocoder provided with detection and measurement of the fundamental frequency, a signal, representing the value of the fundamental frequency is received on synthesis and, in relationship with the occurrence of the specified event, a signal of reinitialization of the pseudo-random sequence of pulses is formed, the frequency of this signal being linked to the received value of the said fundamental frequency.
4. Process for the generation of an excitation signal according to either one of claims 1 and 2, characterized in that, in the case of a baseband vocoder, the specified event is constituted by the passage of the signal of the baseband received through the value zero in a specified direction, or by the passage of the signal of the baseband through a specified extreme value.
5. Process for the generation of an excitation signal according to either one of claims 1 and 2, characterized in that, in the case of a voice- excitation vocoder, the specified event is detected on analysis in the baseband signal by detection of the zero passages in the same direction or by detection of the extreme values, only the value of the differences between the successive events being transmitted, and in that, on synthesis, these events are reconstituted from the received values of the differences.
6. Generator of an excitation signal for a speech synthesizer, comprising generating means (46, 48) to supply a predetermined, pseudo-random sequence of pulses, the said excitation- signal generator being characterized in that it comprises means (58, 60) for reinitializing the generating means, means for producing the excitation signal, which producing means are sensitive to the logic level of the pulses of the said sequence in order to form an excitation signal of predetermined fixed amplitude and of polarity determined by the logic level of the said sequence and means (38) for detecting a specified event, the frequency of which is linked to the fundamental frequency of the speech signal, the said detecting means commanding the said reinitializing means.
7. Generator of an excitation signal for a speech synthesizer, according to claim 6, characterized in that, in the case of a synthesizer provided with channels, the generating means simultaneously supply two pulse trains in phase opposition, means for producing the excitation signal being respectively associated with each one of the pulse trains.
8. Generator of an excitation signal according to claim 6, characterized in that the generating means comprise a shift register (46), at least one EXCLUSIVE NOR gate (48), the inputs of which are connected to two output stages (Q7, Q9) of the register (46), the output of the said gate being connected to the data input of the register, the latter further comprising a reinitialization input (R) connected to the output of the reinitializing means (58, 60).
9. Generator of an excitation signal according to claim 8, characterized in that, in the case of a synthesizer provided with channels, the generating means comprise a logic inverter (54), the input of which is connected to the output of the EXCLUSIVE NOR gate.
10. Generator of an excitation signal according to either one of claims 8 and 9, characterized in that the reinitializing means (58, 60) are sensitive to the said command signal and to the clock signal of the shift register (46) so as to provide at the reinitialization input (R) of the said register a command signal resynchronized with the clock signal.
11. Generator of an excitation signal according to any one of claims 8 to 10, characterized in that the reinitializing means comprise an AND gate (56), a first type D flip-flop (58) and a second typeD flip-flop (60), the reset-to- 1 '(RTO) inputs of which are connected to earth and the reset-to- « 0 inputs of which receive the said command signal, a first input of the AND gate receiving the clock signal of the shift register (46), the other input of this gate being connected to the inverse output (Q) of the second flip-flop (60), the output of the AND gate (56) being connected to the clock input of the first flip-flop (58), the inverse output (Q) of which is looped to its data input (D) and connected to the reinitialization input (R) of the shift register (46) and to the clock input of the second flip-flop (60), the inverse output (Q) of which is looped to its data input (D).
12. Generator of an excitation signal according to claims 9 and 11, characterized in that it comprises a bistable flip-flop (62), the reset-to- « 0 input of which is connected to the non-inverse output (Q) of the first flip-flop and the reset-to- « 1 ' input of which is connected to the output of the logic inverter (54), each one of the outputs of the bistable flip-flop (62) being connected to a means for generating an excitation signal, the non-inverse output being allocated to a predetermined number of consecutive odd channels taken with effect from the first odd channel, and the inverse output being allocated to a predetermined number of consecutive even channels taken with effect from the first even channel.
EP19840400915 1983-05-09 1984-05-04 Method of generating an excitation signal for a channel or linear prediction vocoder, and excitation signal generator therefor Expired EP0128065B1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR8307712 1983-05-09
FR8307712A FR2545966B1 (en) 1983-05-09 1983-05-09 METHOD FOR DEVELOPING AN EXCITATION SIGNAL FOR A CHANNEL OR LINEAR PREDICTION SPEECH SYNTHESIZER AND EXCITATION SIGNAL GENERATOR FOR SUCH A SYNTHESIZER

Publications (2)

Publication Number Publication Date
EP0128065A1 EP0128065A1 (en) 1984-12-12
EP0128065B1 true EP0128065B1 (en) 1987-12-02

Family

ID=9288713

Family Applications (1)

Application Number Title Priority Date Filing Date
EP19840400915 Expired EP0128065B1 (en) 1983-05-09 1984-05-04 Method of generating an excitation signal for a channel or linear prediction vocoder, and excitation signal generator therefor

Country Status (3)

Country Link
EP (1) EP0128065B1 (en)
DE (1) DE3467946D1 (en)
FR (1) FR2545966B1 (en)

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE1079118B (en) * 1958-04-11 1960-04-07 Siemens Ag Method for electrical communication under frequency band pressure
FR1602217A (en) * 1968-12-16 1970-10-26
DE2062680A1 (en) * 1970-12-19 1972-06-29 Licentia Gmbh Synthesizer for speech signals
DE2435057B2 (en) * 1973-10-18 1975-09-11 Hewlett-Packard Ltd., South Queensferry, West Lothian (Grossbritannien) Circuit arrangement for synchronizing and / or re-triggering a generator for generating a sequence of pseudo-random binary signals
FR2252799A5 (en) * 1973-11-26 1975-06-20 Commissariat Energie Atomique Automatic recording and synthesis of speech - uses time interval sectioner for speech amplitude signals with analogue-digital-analogue conversion

Also Published As

Publication number Publication date
DE3467946D1 (en) 1988-01-14
FR2545966A1 (en) 1984-11-16
EP0128065A1 (en) 1984-12-12
FR2545966B1 (en) 1986-02-21

Similar Documents

Publication Publication Date Title
EP0419337B1 (en) Digital signal encoding method, encoder and decoder for carrying out the method, regeneration method and regenerator therefore
EP0223667B1 (en) Process for marine seismic prospection using a coded vibratory signal, and device for carrying out such a method
EP0768770B1 (en) Method and arrangement for the creation of comfort noise in a digital transmission system
EP2277172B1 (en) Concealment of transmission error in a digital signal in a hierarchical decoding structure
CA1285071C (en) Voice coding process and device for implementing said process
FR2511170A1 (en) AUTOMATIC THRESHOLD TRACKING SYSTEM
FR2596936A1 (en) VOICE SIGNAL TRANSMISSION SYSTEM
EP0194902A1 (en) Method and system for spread spectrum transmission, especially for information transmission in an electric power distribution network
FR2520539A1 (en) METHOD AND SYSTEM FOR PROCESSING SILENCES IN A SPEECH SIGNAL
WO2008096084A1 (en) Synthesis of lost blocks of a digital audio signal, with pitch period correction
CA2029768C (en) Method and device for coding very low flow rate vocoder predictive filters
FR2707115A1 (en) A method for producing seismic wavelet forms having minimal side lobe energy.
FR2784218A1 (en) LOW-SPEED SPEECH CODING METHOD
FR2681488A1 (en) DISPERSE SPECTRUM COMMUNICATION DEVICE.
EP0128065B1 (en) Method of generating an excitation signal for a channel or linear prediction vocoder, and excitation signal generator therefor
EP0043056A1 (en) Process for the detection of speech in a telephone circuit signal, and speech detector therefor
FR2616918A1 (en) SYSTEM AND METHOD FOR RECORDING AND REPRODUCING RADAR SIGNAL INFORMATION
EP0573358B1 (en) Variable speed voice synthesizer method and apparatus
FR2888704A1 (en)
EP0275751B1 (en) Method and apparatus for the digital transmission of voice signals by way of a radio channel
EP0184953B1 (en) Method and device for digital information transmission using differential frequency shift keying
EP0311494B1 (en) Method and apparatus for digital transmission of speech between mobile stations
CA2019774C (en) Device for detecting in a binary signal a pointer consisting of a continuous sequence of binary elements having the same value
CA2079884A1 (en) Method and device for low-speed speech coding
FR2741743A1 (en) Speech intelligibility improvement method for low bit rate vocoder

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Designated state(s): DE GB IT NL

17P Request for examination filed

Effective date: 19850515

17Q First examination report despatched

Effective date: 19860711

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): DE GB IT NL

REF Corresponds to:

Ref document number: 3467946

Country of ref document: DE

Date of ref document: 19880114

GBT Gb: translation of ep patent filed (gb section 77(6)(a)/1977)
ITF It: translation for a ep patent filed

Owner name: SOCIETA' ITALIANA BREVETTI S.P.A.

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed
ITTA It: last paid annual fee
PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: GB

Payment date: 19960426

Year of fee payment: 13

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: DE

Payment date: 19960525

Year of fee payment: 13

PGFP Annual fee paid to national office [announced via postgrant information from national office to epo]

Ref country code: NL

Payment date: 19960625

Year of fee payment: 13

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Effective date: 19970504

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: NL

Effective date: 19971201

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 19970504

NLV4 Nl: lapsed or anulled due to non-payment of the annual fee

Effective date: 19971201

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 19980203