EP0124411A1 - Channel vocoder comprising means for suppressing parasitic modulation of the synthesized speech signal - Google Patents

Channel vocoder comprising means for suppressing parasitic modulation of the synthesized speech signal Download PDF

Info

Publication number
EP0124411A1
EP0124411A1 EP84400744A EP84400744A EP0124411A1 EP 0124411 A1 EP0124411 A1 EP 0124411A1 EP 84400744 A EP84400744 A EP 84400744A EP 84400744 A EP84400744 A EP 84400744A EP 0124411 A1 EP0124411 A1 EP 0124411A1
Authority
EP
European Patent Office
Prior art keywords
signal
synthesis
input
channel
subtractor
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.)
Granted
Application number
EP84400744A
Other languages
German (de)
French (fr)
Other versions
EP0124411B1 (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.)
ZURCHER JEAN FREDERIC
Original Assignee
ZURCHER JEAN FREDERIC
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 ZURCHER JEAN FREDERIC filed Critical ZURCHER JEAN FREDERIC
Publication of EP0124411A1 publication Critical patent/EP0124411A1/en
Application granted granted Critical
Publication of EP0124411B1 publication Critical patent/EP0124411B1/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
    • G10L19/02—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 channel vocoder provided with means for compensating for spurious modulations of the synthesized speech signal.
  • a vocoder is schematically represented. This can be broken down into three parts: an analysis sub-assembly 2, a transmission unit 4 and a synthesis sub-assembly 6.
  • the analysis sub-assembly 2 will transform a signal from analog speech 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 speed of the speech signal thus coded is of the order of 2 to 5 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 subset 2 analysis and code 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 means 8 and of the 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 coding the spectral envelope of the speech signal P is then applied to the input of a means 12 comprising a filter or a set of filters to restore this spectral envelope.
  • the second digital signal, coded by the means 10 is applied to the input of a means 14 for developing an excitation signal which reproduces the fine structure of the spectrum. This excitation signal is applied to the means 12 to restore the spectral envelope of the speech signal.
  • This general scheme of a vocoder applies in particular to channel vocoders. It will be recalled that there are three types of channel vocoders: the channel vocoders proper, the baseband vocoders and the voice excitation vocoders.
  • the means 8 encodes the entire spectral band analyzed, for example between 200 Hz and 6000 Hz, cut into sub-bands each treated by an analysis channel, and the means 10 codes a parameter allowing the generation of the excitation signal on 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; at reception, the signal of excitement is taken from this decoded baseband.
  • the high band is coded by means 8 as the total spectral band of the vocoding channels of the first type.
  • the processing carried out by the third type of vocoder differs from that carried out by the second in that the excitation signal is not deduced from the decoded baseband but is obtained from parameters provided by temporal analysis of the speech signal. .
  • the speech signal P from a microphone, a telephone line, a tape recorder or the like is applied to the input of an amplification means 16 to adapt the level of this speech signal to a compatible level with a good functioning of the vocoder.
  • the outgoing signal of this amplification means 16 is applied on the one hand to the input of a means 18 which codes the data necessary for the elaboration of the excitation signal of the synthesis filters and which fulfills the function of the means 10 of FIG. 1, and on the other hand on the input of a pre-emphasis means 20 which increases the gain (+ 6 dB / octave) of this signal in the frequency band between for example 600 Hz and 6000 Hz.
  • This means 20 has the role of compensating for the natural attenuation of the high frequencies in the speech signal and allows all the analysis channels to be used with all the available dynamics. This means 20 can however be omitted. This means 20 outputs a signal which attacks each of the n analysis channels 22.
  • each of the analysis channels 22 comprises in series a bandpass filter 24, a stage 26 for rectifying the signal and a lowpass filter 28. All of the filters pass -strip 24 1 , ... 24 n constitutes a bank of contiguous filters.
  • Each analysis channel 22 selects by means of the bandpass filter 24 a frequency band of the speech signal and will rectify the sound vibration signal into an energy signal thanks to stage 26.
  • the energy signal is then applied to the input of a low-pass filter 28 whose cut-off frequency is of the order of 20 to 50 Hz.
  • This low-pass filter makes it possible to have, at the output of each analysis channel 22, a signal representative of the energy of the speech signal contained in the passband of the channel in question and whose frequency of variation is low, (it is limited by the upper cut-off frequency of the low-pass filter) which limits the data rate transmitting.
  • each of the analysis channels 22 is connected, through a multiplexer 30, to a quantization means 32 which comprises an analog-digital converter (necessary only if the upstream processing is carried out in analog technology) and a quantizer, for example logarithmic, thus delivering digital signals coded in decibels.
  • the signal from this quantization means 32 is multiplexed with the signal from the means 18 to produce a frame which is sent in a transmission line L.
  • this transmission line L is the channel vocoder synthesis sub-assembly.
  • a synthesis subset of a known type of a channel vocoder is first describe with reference to Figure 3, a synthesis subset of a known type of a channel vocoder.
  • the input of the synthesis subset is constituted by a reception means 36 which extracts the data from the frame.
  • the multiplexed data from the analysis channels are applied to the input of a storage means 40 and the data describing the fine structure of the spectrum of the speech signal are applied to the input of a means 38 for developing the excitation signal.
  • the signal delivered by the storage means 40 is demultiplexed by a demultiplexer 42 whose outputs control the gains of the modulators 46 of the n synthesis channels 44.
  • Each synthesis channel comprises in series, downstream of its modulator 46, a pass filter band 48, which is similar to the bandpass filter 24 of the same rank analysis channel.
  • Each modulator 46 modulates the excitation signal as a function of the data item coming from the demultiplexer 42 and coming from the analysis channel of the same rank.
  • the output of each synthesis channel 44 is connected to an adder 49.
  • the signal delivered by this adder undergoes processing in a de-emphasis means 58, which attenuates the frequencies between, for example 600 Hz and 6000 Hz symmetrically to the action of the accentuating means 20 of the analysis part.
  • a de-emphasis means 58 which attenuates the frequencies between, for example 600 Hz and 6000 Hz symmetrically to the action of the accentuating means 20 of the analysis part.
  • a low-pass decoder-filter assembly 60 which restores the speech signal in the analyzed band, its cut-off frequency possibly being for example 6000 Hz.
  • the synthesized speech signal S delivered by the channel vocoder is never perfectly identical to the initial speech signal P. This is partly due to the imperfect restitution of the instantaneous spectral envelope of the speech signal. This imperfection has several causes.
  • each filter is not constant over the entire spectral band of this filter.
  • the gain is not negative and infinite.
  • the filters of two consecutive analysis or synthesis channels overlap with a certain negative gain (for example -6 dB).
  • A Denote A by the level (in decibels) of the frequency line f ex of the speech signal P at a determined time.
  • this line of frequency f ex has an intensity of A in -6 decibels.
  • the excitation signal from the means 38 modulated respectively by the data from the analysis channel p-1 from the analysis channel p has an intensity equal to A in -6 + A ex decibels.
  • These signals from the modulators of rank p-1 and p will be applied respectively to the inputs of the bandpass filters 48 p-1 and 48p. These filters will again decrease the gain of the frequency line f ex by 6 decibels.
  • the frequency line f ex therefore has an intensity equal to twice A at + A ex -12 decibels, that is to say an intensity from A to + A ex - 6 decibels since a weakening of 6 decibels corresponds to a halving of the signal intensity.
  • This ripple can be eliminated, or at least very attenuated, if an analysis chain is added behind the bandpass filter of each synthesis channel and if a modulator is controlled so that, for substantially stationary signals, the output of the synthesis band pass filter is identical to the output of the band pass filter of the corresponding analysis channel.
  • Each synthesis channel comprises in series a first bandpass filter, a modulator and a second bandpass filter identical to the first.
  • the first bandpass filter receives the excitation signal.
  • Each analysis chain includes an energy measurement means, consisting of a detector and a low-pass filter in series, a subtractor and a direct current amplifier.
  • the energy measurement means receives the output signal from the synthesis channel and delivers a signal to an input of the subtractor. This receives on another input the signal delivered by the analysis channel of the same rank as the synthesis channel and delivers on the gain control input of the modulator, through the DC amplifier, a signal expressing the signed difference between the signals received on its inputs.
  • This chain of analysis can only effectively correct undulations of low frequency because of the reaction time inherent in the closed loop structure.
  • reaction speed of this analysis chain comes from the nature of the energy measurement means which includes a low-pass filter with non-zero time constant.
  • this analysis chain cannot effectively correct a rapid unjustified increase in the signal of the output bandpass filter of the synthesis channel. This results in an unpleasant subjective impression when listening to the synthesized speech signal.
  • the object of the invention is precisely to eliminate these defects. For this, it is advisable to associate with each synthesis channel an analysis chain capable of rapidly correcting the signal of the output bandpass filter of this synthesis channel.
  • This analysis chain includes a peak detector receiving the signal delivered by the bandpass filter of the synthesis channel, a subtractor receiving the signal delivered by the peak detector and the excitation signal and a modulator which receives on its input. signal the output signal of the bandpass filter of the synthesis channel and on its gain input the signal delivered by the subtractor.
  • a low-pass filter can be used in the device of the invention to measure the energy. This somewhat limits the performance of the analysis chain, which however remains clearly superior thanks to the open loop structure of the analysis chain of the cited patent.
  • the vocoder according to the invention comprises n signal processing means, numbered from 1 to n, each signal processing means being associated with a synthesis channel.
  • the vocoder comprises a single signal processing means whose input of the quantization and coding circuit is connected to a multiplexer whose n inputs receive the energy signal from each synthesis channel and the modulator signal input of which is connected to another multiplexer of which n inputs receive the output signal of the bandpass filter of each synthesis channel.
  • the vocoder further comprises a smoothing means located between the subtractor and the modulator, said smoothing means performing in each channel a smoothing so that the difference in intensity between two successive data from the same channel remains below a predetermined value.
  • the smoothing means comprises a comparator, one input of which receives the signal from the subtractor and the other input of which receives another signal, a means for calculating a gain signal, means for storing this gain signal , said calculation means determining for a given frame said gain as a function of the output signal of the comparator and of the stored gain signal of the previous frame, said storage means supplying said other signal.
  • each subtractor comprises on its inverting input tilting means making it possible to receive on ladi the inverting input is either the signal from the associated coding means, or a predetermined fixed level signal.
  • each energy measurement means comprises in series a rectifier and a peak detector.
  • Channels to the vocoder according to the invention the subset of synthesis is shown schematically in Figure 5, receives a receiving means 36, a data frame comprising a repre - sentation of a digital speech signal analyzed by the analysis subset of said vocoder.
  • This means delivers a first signal to a means for generating the excitation signal 38 and a second signal to a storage means 40.
  • the vocoder synthesis sub-assembly according to the invention is identical to the subset of vocoder synthesis of a known type described in FIG. 3.
  • the excitation signal produced by the means 38 is applied to the inputs of the n synthesis channels 45, respectively identical to the analysis channels of the same rank. They each include a band-pass filter 48, a rectifier 50 and a low-pass filter 52.
  • the latter can advantageously be replaced by a peak detector which follows the possible abrupt increases in speed more quickly than a low-pass filter. energy localized in frequency of the excitation signal, so that these increases are compensated as quickly as possible by the processing in the synthesis chain according to the invention.
  • the signals delivered by each of these synthesis channels 45 are then quantified and coded by a quantization and coding means 56 to which they are connected through an output multiplexer 54 S l .
  • the means 56 performs the same coding as the quantization and coding means of the analysis subset.
  • the energy signal extracted from the synthesis channels 45 and from the means 56 is applied to the inverting input of a subtractor 62 whose other input receives the signal delivered by the storage means 40 which contains the data from the channels d 'analysis.
  • the subtractor 62 delivers at output the signed difference of the signals applied to each of its inputs.
  • This signal is processed by a smoothing means 64 which will be described in more detail with reference to the following figure.
  • the smoothing means 64 performs smoothing per channel. It ensures that two consecutive data from the same channel do not differ by more than a certain predetermined value, for example 1 decibel. This plea can be omitted.
  • the output of the smoothing means is connected to the input controlling the gain of a modulator 46.
  • a signal from the output S. 2 of a multiplexer 66 and which receives on each of its inputs a signal taken at the output of the bandpass filters 48 of each of the n synthesis channels.
  • the signal delivered by the modulator 46 after processing by a deactivation means 58 and a decoder-low-pass filter assembly 60 identical to those used in the vocoder of known type described in FIG. 3, represents the synthesized speech signal S.
  • the assembly constituted by the means 56, the subtractor 62, the optional smoothing means 64 and the modulator 46 constitutes a signal processing means 63, a first input of which is connected to the output S 1 and a second input of which is connected to the exit S -2 .
  • the vocoder according to the invention has a structure which makes it possible to eliminate parasitic ripples from the synthesized speech signal appearing in vocoders according to known art. Indeed, let’s take the digital example used to illustrate the imperfections of vocoders according to known art.
  • a line of frequency f ex which is at the intersection of two bandpass filters of analysis channels or consecutive synthesis p-1 and p. In the initial speech signal, this line has an intensity A and at the output of the analysis filters, it has an intensity of A in -6 dB. The intensity of this line is contained in the signal which is stored by the storage means 40 of the synthesis part of the vocoder.
  • the means 38 for generating the excitation signal will provide a line of frequency f ex of intensity A, as in the vocoder described with reference to FIG. 3.
  • this excitation signal will be processed by the synthesis channels of rank p-1 and p identical to the analysis channels of the same rank.
  • the intensity of the frequency line f ex will be A - 6 dB.
  • the subtractor 62 For the channel of rank p-1, there is therefore at the input of the subtractor 62 on the inverting input, a number representing an intensity A - 6 dB and on the non-inverting input, a number representing an intensity A - 6 dB from the pl rank analysis channel. At the output of this subtractor 62, there is therefore for the frequency line f ex a difference of A - A ex dB.
  • This number modulates in the modulator 46 the gain of the signal delivered by the band-pass filter 48 p-1 of the synthesis channel of rank p-1 and whose frequency line f ex has an intensity of A ex - 6 decibels. At the output of this modulator 46, there is therefore a signal whose frequency line f ex has an intensity of (A in -A ex ) + (A ex -6) dB, that is to say A - 6 dB .
  • the frequency line f ex is processed by channels p-1 and p. There is therefore also, at the output of the modulator 46, a signal coming from the channel of rank p containing a line of frequency f ex of intensity A - 6 dB. By summing on the two channels p-1 and p, the frequency line f ex therefore has an intensity of twice A - 6 dB, that is to say an intensity of the order of A decibels.
  • the intensity of the lines whose frequency is at the intersection of two consecutive bandpass filters of analysis or synthesis channels is restored without weakening.
  • the parasitic ripple which appeared in the vocoders according to the prior art has disappeared.
  • the line of frequency f ex is restored with an intensity A independently of the intensity A ex of the line of frequency f ex of the excitation signal . This results in a correction of the possible ripple of the spectral envelope of the excitation signal. This is particularly interesting for baseband or voice excitation vocoders.
  • This smoothing means 64 makes it possible to have, for two consecutive data originating from the same channel, only a difference at most equal to a value predetermined, for example 1 decibel.
  • This smoothing means 64 comprises in series a comparator 66 and a calculation means 68, for calculating the corrected gain of the data applied at the input of the smoothing means 64. It also includes means 70, for storing the corrected gain calculated by the means 68, which is connected to comparator 66.
  • the data from a synthesis channel are processed by the smoothing means in the following manner.
  • the comparator 66 receives on a non-inverting input the data coming from the subtractor 62 and which are extracted from the frame of rank i, and on a inverting input a signal G i-1 which represents the gain of the data coming from the frame of rank i-1 and corresponding to the same channel.
  • the calculation means 68 will output a gain signal G i which is stored in the means 70 and which is applied to the input of the modulator 46.
  • FIG. 6 shows a particular embodiment of this smoothing means.
  • the means 68 comprises in series a means 72 for determining a coefficient ⁇ i of the gain, a multiplier 74 and an adder 76.
  • the means 70 comprises a memory 78 and a quantization means 80.
  • the means 72 receives from the digital comparator 66 a signal indicating whether the difference between the signals applied to the input of said digital comparator is positive, zero or negative.
  • the coefficient ⁇ i delivered by this means 72 can be equal, for example to 2- m if the difference is positive, 0 if it is zero and -2 -m if it is negative, the choice of the value of m predetermined defining the filtering time constant of the smoothing means 64.
  • This coefficient ⁇ i is multiplied by the multiplier 74 with the gain G i-1 stored in memory 78.
  • the result obtained is added in the adder 76 to the gain G i-1 to give a gain G i equal to G i-1 ( 1 + i ).
  • This gain G i stored in memory 78 is quantified by the quantization means 80 so that it can be compared to the signal delivered by the subtractor 62.
  • a switching means can be provided at the inverting input of the subtractor 62 to receive, depending on the state of the switching means, either the signal from the means 56 (for an operation according to the invention), ie a predetermined fixed level signal.
  • the operation of the synthesis subset is analogous to that of the conventional channel vocoder.

Abstract

1. Synthesis subassembly of a channel vocoder including a means (36) of reception and extraction of the data in a frame, a means (38) of producing an excitation signal, the said synthesis subassembly including : - _n synthesis channels (45) numbered from 1 to _n, each including a bandpass filter (48) and a means of energy measurement (50, 52) for measuring the energy contained in the said band, the said synthesis channels receiving on input the excitation signal coming from the means (38) of producing the excitation signal, - at least one subtractor (62) receiving on a first input the energy signal delivered by a synthesis channel and on a second input the signal coming from the means of reception (36) and representing the energy signal delivered by the corresponding analysis channel and delivering a signal expressing the signed difference between the two signals applied on input, - at least one modulator (46) associated with the subtractor, of which the input which controls the gain receives the signal delivered by the subtractor, the said synthesis subassembly being characterized in that the synthesis channels (45), the subtractor or subtractors (62) and the associated modulator or modulators (46) are arranged in cascade in this order and constitute an open loop circuit, the synthesis subassembly being arranged such that it includes at least one means of signal processing (63) including a subtractor (62) and its associated modulator (46), the said means of processing (63) including a quantification and coding circuit (56) which receives on input the energy signal from a synthesis channel (45) in order to deliver to one of the inputs of the subtractor a quantified and coded energy signal, the other input of the subtractor receiving the energy signal coming from the means of reception and delivered by the corresponding analysis channel, the modulator receiving on its signal to be modulated input the output signal from the band pass filter of the synthesis channel.

Description

L'invention concerne un vocodeur à canaux muni de moyens de compensation des modulations parasites du signal de parole synthétisé.The invention relates to a channel vocoder provided with means for compensating for spurious modulations of the synthesized speech signal.

On connaît plusieurs types de dispositif 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 le premier type de vocodeur. Ces trois types de vocodeurs ont une structure générale identique qui va être décrite en référence à la figure 1.Several types of device 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 first type of vocoder. These three types of vocoders have an identical general structure which will be described with reference to FIG. 1.

Sur cette figure 1, on à représenté schématiquement un vocodeur. Celui-ci 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 5 kbit/s, ce qui est très inférieur au codage MIC qui nécessite 64 kbit/s.In this figure 1, a vocoder is schematically represented. This can be broken down into three parts: an analysis sub-assembly 2, a transmission unit 4 and a synthesis sub-assembly 6. We have seen that the analysis sub-assembly 2 will transform a signal from analog speech 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 speed of the speech signal thus coded is of the order of 2 to 5 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 subset 2, analysis and code 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 moyen 8 et du 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 means 8 and of the 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 comportant un filtre ou un ensemble de filtres pour restituer cette enveloppe spectrale. Le deuxième signal numérique, codé par le moyen 10 est appliqué sur l'entrée d'un moyen 14 d'élaboration d'un signal d'excitation qui reproduit la structure fine du spectre. Ce signal d'excitation est appliqué au moyen 12 pour restituer l'enveloppe spectrale du signal de parole.The digital signal coding the spectral envelope of the speech signal P, is then applied to the input of a means 12 comprising a filter or a set of filters to restore this spectral envelope. The second digital signal, coded by the means 10 is applied to the input of a means 14 for developing an excitation signal which reproduces the fine structure of the spectrum. This excitation signal is applied to the means 12 to restore the spectral envelope of the speech signal.

Ce schéma général d'un vocodeur s'applique en particulier aux vocodeurs à canaux. On rappelle qu'il y a trois types de vocodeurs à canaux : les vocodeurs à canaux proprement dits, les vocodeurs à bande de base et les vocodeurs à excitation vocale. Dans le premier et le troisième types de vocodeur, le moyen 8 code l'ensemble de la bande spectrale analysée, comprise par exemple entre 200 Hz et 6000 Hz, découpée en sous-bandes traitées chacune par un canal d'analyse, 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 1000 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 cette bande de base décodée. La bande haute est codée par le moyen 8 comme la bande spectrale totale des vocodeurs à canaux 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 analyse temporelle du signal de parole.This general scheme of a vocoder applies in particular to channel vocoders. It will be recalled that there are three types of channel vocoders: the channel vocoders proper, the baseband vocoders and the voice excitation vocoders. In the first and third types of vocoder, the means 8 encodes the entire spectral band analyzed, for example between 200 Hz and 6000 Hz, cut into sub-bands each treated by an analysis channel, and the means 10 codes a parameter allowing the generation of the excitation signal on 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; at reception, the signal of excitement is taken from this decoded baseband. The high band is coded by means 8 as the total spectral band of the vocoding channels of the first type. The processing carried out by the third type of vocoder differs from that carried out by the second in that the excitation signal is not deduced from the decoded baseband but is obtained from parameters provided by temporal analysis of the speech signal. .

On va maintenant décrire, plus en détail, le sous-ensemble d'analyse d'un vocodeur à canaux en référence à la figure 2. La structure décrite est une structure connue de l'homme de l'art ; elle est utilisée dans le sous-ensemble d'analyse du vocodeur à canaux selon l'invention.We will now describe, in more detail, the analysis subset of a channel vocoder with reference to FIG. 2. The structure described is a structure known to those skilled in the art; it is used in the subset of analysis of the channel vocoder according to the invention.

Le signal de parole P issu d'un microphone, d'une ligne téléphonique, d'un magnétophone ou autre est appliqué à l'entrée d'un moyen d'amplification 16 pour adapter le niveau de ce signal de parole à un niveau compatible avec un bon fonctionnement du vocodeur. Le signal sortant de ce moyen d'amplification 16 est appliqué d'une part sur l'entrée d'un moyen 18 qui code les données nécessaires à l'élaboration du signal d'excitation des filtres de synthèse et qui remplit la fonction du moyen 10 de la figure 1, et d'autre part sur l'entrée d'un moyen de pré-accentuation 20 qui ac- croit le gain (+ 6 dB/octave) de ce signal dans la bande de fréquence comprise entre par exemple 600 Hz et 6000 Hz. Ce moyen 20 a pour rôle de compenser l'atténuation naturelle des hautes fréquences dans le signal de parole et permet à tous les canaux d'analyse d'être utilisés avec toute la dynamique disponible. Ce moyen 20 peut cependant être omis. Ce moyen 20 délivre en sortie un signal qui attaque chacun des n canaux d'analyse 22.The speech signal P from a microphone, a telephone line, a tape recorder or the like is applied to the input of an amplification means 16 to adapt the level of this speech signal to a compatible level with a good functioning of the vocoder. The outgoing signal of this amplification means 16 is applied on the one hand to the input of a means 18 which codes the data necessary for the elaboration of the excitation signal of the synthesis filters and which fulfills the function of the means 10 of FIG. 1, and on the other hand on the input of a pre-emphasis means 20 which increases the gain (+ 6 dB / octave) of this signal in the frequency band between for example 600 Hz and 6000 Hz. This means 20 has the role of compensating for the natural attenuation of the high frequencies in the speech signal and allows all the analysis channels to be used with all the available dynamics. This means 20 can however be omitted. This means 20 outputs a signal which attacks each of the n analysis channels 22.

Pour les trois types de vocodeurs à canaux décrits plus haut, chacun des canaux d'analyse 22 comprend en série un filtre passe-bande 24, un étage 26 pour redresser le signal et un filtre passe-bas 28. L'ensemble des filtres passe-bande 241, ... 24n constitue un banc de filtres contigus.For the three types of channel vocoders described above, each of the analysis channels 22 comprises in series a bandpass filter 24, a stage 26 for rectifying the signal and a lowpass filter 28. All of the filters pass -strip 24 1 , ... 24 n constitutes a bank of contiguous filters.

Chaque canal d'analyse 22 sélectionne au moyen du filtre passe-bande 24 une bande de fréquence du signal de parole et va redresser le signal de vibration sonore en un signal d'énergie grâce à l'étage 26. Le signal d'énergie est ensuite appliqué à l'entrée d'un filtre passe-bas 28 dont la fréquence de coupure est de l'ordre de 20 à 50 Hz. Ce filtre passe-bas permet d'avoir en sortie de chaque canal d'analyse 22, un signal représentatif de l'énergie du signal de parole contenu dans la bande passante du canal considéré et dont la fréquence de variation est faible, (elle est limitée par la fréquence supérieure de coupure du filtre passe-bas) ce qui limite le débit des données à transmettre.Each analysis channel 22 selects by means of the bandpass filter 24 a frequency band of the speech signal and will rectify the sound vibration signal into an energy signal thanks to stage 26. The energy signal is then applied to the input of a low-pass filter 28 whose cut-off frequency is of the order of 20 to 50 Hz. This low-pass filter makes it possible to have, at the output of each analysis channel 22, a signal representative of the energy of the speech signal contained in the passband of the channel in question and whose frequency of variation is low, (it is limited by the upper cut-off frequency of the low-pass filter) which limits the data rate transmitting.

La sortie de chacun des canaux d'analyse 22 est connectée, à travers un multiplexeur 30, à un moyen de quantification 32 qui comprend un convertisseur analogique-numérique (nécessaire seulement si le traitement en amont est réalisé en technologie analogique) et un quantificateur, par exemple logarithmique, délivrant ainsi des signaux numériques codés en décibels. Le signal issu de ce moyen de quantification 32 est multiplexé avec le signal issu du moyen 18 pour réaliser une trame qui est envoyée dans une ligne de transmission L.The output of each of the analysis channels 22 is connected, through a multiplexer 30, to a quantization means 32 which comprises an analog-digital converter (necessary only if the upstream processing is carried out in analog technology) and a quantizer, for example logarithmic, thus delivering digital signals coded in decibels. The signal from this quantization means 32 is multiplexed with the signal from the means 18 to produce a frame which is sent in a transmission line L.

A l'autre extrémité de cette ligne de transmission L, se trouve le sous-ensemble de synthèse du vocodeur à canaux. On va tout d'abord décrire en référence à la figure 3, un sous-ensemble de synthèse d'un type connu d'un vocodeur à canaux.At the other end of this transmission line L, is the channel vocoder synthesis sub-assembly. We will first describe with reference to Figure 3, a synthesis subset of a known type of a channel vocoder.

L'entrée du sous-ensemble de synthèse est constituée par un moyen 36 de réception qui extrait les données de la trame. Les données multiplexées issues des canaux d'analyse sont appliquées à l'entrée d'un moyen de mémorisation 40 et les données décrivant la structure fine du spectre du signal de parole sont appliquées à l'entrée d'un moyen 38 d'élaboration du signal d'excitation.The input of the synthesis subset is constituted by a reception means 36 which extracts the data from the frame. The multiplexed data from the analysis channels are applied to the input of a storage means 40 and the data describing the fine structure of the spectrum of the speech signal are applied to the input of a means 38 for developing the excitation signal.

Le signal délivré par le moyen de mémorisation 40 est démultiplexé par un démultiplexeur 42 dont les sorties commandent les gains des modulateurs 46 des n canaux de synthèse 44. Chaque canal de synthèse comprend en série, en aval de son modulateur 46, un filtre passe-bande 48, qui est semblable au filtre passe-bande 24 du canal d'analyse de même rang. Chaque modulateur 46 module le signal d'excitation en fonction de la donnée issue du démultiplexeur 42 et provenant du canal d'analyse de même rang. La sortie de chaque canal de synthèse 44 est reliée à un additionneur 49. Le signal délivré par cet additionneur subit un traitement dans un moyen désaccentuateur 58, qui atténue les fréquences comprises entre par exemple 600 Hz et 6000 Hz de manière symétrique à l'action du moyen accentuateur 20 de la partie analyse. Dans le cas d'un traitement totalement numérique, il est suivi par un ensemble décodeur-filtre passe-bas 60 qui restitue le signal de parole dans la bande analysée, sa fréquence de coupure pouvant être par exemple de 6000 Hz.The signal delivered by the storage means 40 is demultiplexed by a demultiplexer 42 whose outputs control the gains of the modulators 46 of the n synthesis channels 44. Each synthesis channel comprises in series, downstream of its modulator 46, a pass filter band 48, which is similar to the bandpass filter 24 of the same rank analysis channel. Each modulator 46 modulates the excitation signal as a function of the data item coming from the demultiplexer 42 and coming from the analysis channel of the same rank. The output of each synthesis channel 44 is connected to an adder 49. The signal delivered by this adder undergoes processing in a de-emphasis means 58, which attenuates the frequencies between, for example 600 Hz and 6000 Hz symmetrically to the action of the accentuating means 20 of the analysis part. In the case of fully digital processing, it is followed by a low-pass decoder-filter assembly 60 which restores the speech signal in the analyzed band, its cut-off frequency possibly being for example 6000 Hz.

Le signal de parole synthétisé S délivré par le vocodeur à canaux n'est jamais parfaitement identique au signal de parole P initial. Ceci est en partie dû à la restitution imparfaite de l'enveloppe spectrale instantanée du signal de parole. Cette imperfection a plusieurs causes.The synthesized speech signal S delivered by the channel vocoder is never perfectly identical to the initial speech signal P. This is partly due to the imperfect restitution of the instantaneous spectral envelope of the speech signal. This imperfection has several causes.

Certains défauts sont inhérents à la nature même des vocodeurs à canaux et ne peuvent être atténués que par une forte augmentation du débit consacré au codage du signal de parole. Ces défauts peuvent être atténués, notamment en diminuant la largeur des filtres passe-bandes d'analyse et de synthèse, ce qui conduit à en augmenter le nombre, et en augmentant le nombre de niveaux de quantification de l'amplitude de l'enveloppe spectrale instantanée dans chacun des canaux d'analyse.Certain defects are inherent in the very nature of channel vocoders and can only be mitigated by a large increase in the bit rate devoted to coding the speech signal. These defects can be mitigated, in particular by reducing the width of the analysis and synthesis bandpass filters, which leads to increasing the number thereof, and by increasing the number of quantification levels of the amplitude of the spectral envelope. instant in each of the analysis channels.

Par ailleurs, on observe dans les vocodeurs à canaux connus, une ondulation parasite sur le spectre instantané du signal de synthèse qui vient s'ajouter aux bruits de quantification inhérents à ce type de vocodeur, et qui est partiellement à l'origine de l'impression subjective de réverbération ou de résonance présentée par la parole synthétique. Cette ondulation est notamment due, dans le cas d'un vocodeur à bande de base, au fait que l'enveloppe spectrale du signal d'excitation n'est jamais plate, donc neutre. Il s'ensuit des modifications fluctuantes des enveloppes spectrales instantanées reconstituées par chaque canal de synthèse, qui s'effectuent au rythme des modifications spectrales affectant la bande de base.In addition, we observe in vocoders with known channels, a parasitic ripple on the instantaneous spectrum of the synthesis signal which is added to the quantization noises inherent in this type of vocoder, and which is partially at the origin of the subjective impression of reverberation or resonance presented by synthetic speech. This ripple is notably due, in the case of a baseband vocoder, to the fact that the spectral envelope of the excitation signal is never flat, therefore neutral. It follows from the fluctuating modifications of the instantaneous spectral envelopes reconstituted by each synthesis channel, which take place at the rate of the spectral modifications affecting the baseband.

En outre, dans les vocodeurs à canaux classiques, il existe une modulation parasite des enveloppes spectrales des signaux de synthèse due à la structure des filtres passe-bande d'analyse et de synthèse. Ainsi, sur la figure 4, on a représenté des diagrammes théoriques et réels de gain des filtres passe-bande des canaux d'analyse ou de synthèse. Le cas théorique est illustré sur la figure 4a. On a représenté les gains des filtres successifs p-1, p et p+1 centrés sur les fréquences fp-l, fp et fp+1, extraits du banc de filtre constitué par les filtres passe-bandes des n canaux d'analyse ou de synthèse du vocodeur. Dans cette représentation théorique, dans chaque bande spectrale, le gain est nul (en décibel) sur toute la largeur de la bande et négatif et infini en dehors.In addition, in conventional channel vocoders, there is parasitic modulation of the spectral envelopes of the synthesis signals due to the structure of the analysis and synthesis bandpass filters. Thus, in FIG. 4, theoretical and real gain diagrams of the bandpass filters of the analysis or synthesis channels have been represented. The theoretical case is illustrated in Figure 4a. We have represented the gains of successive filters p-1, p and p + 1 centered on the frequencies fp -l , fp and f p + 1 , extracted from the filter bank constituted by the bandpass filters of n vocoder analysis or synthesis channels. In this theoretical representation, in each spectral band, the gain is zero (in decibels) over the entire width of the band and negative and infinite outside.

En réalité, comme on l'a représenté sur la figure 4b, le gain de chaque filtre n'est pas constant sur toute la bande spectrale de ce filtre. En particulier, aux bornes inférieure et supérieure de fréquence de chaque canal, le gain n'est pas négatif et infini. Les filtres de deux canaux consécutifs d'analyse ou de synthèse se recoupent avec un certain gain négatif (par exemple -6 dB).In reality, as shown in FIG. 4b, the gain of each filter is not constant over the entire spectral band of this filter. In particular, at the lower and upper frequency limits of each channel, the gain is not negative and infinite. The filters of two consecutive analysis or synthesis channels overlap with a certain negative gain (for example -6 dB).

On cherche à se rapprocher du cas idéal où les canaux ne se recoupent pas. Dans ce but, on peut augmenter la sélectivité de chaque filtre, ce qui en augmente l'étroitesse, donc le nombre et par conséquent le débit. Le gain à l'intersection de deux canaux est donc limité en valeur inférieure par le nombre de canaux que l'on peut utiliser.We are trying to get closer to the ideal case where the channels do not overlap. For this purpose, it is possible to increase the selectivity of each filter, which increases its narrowness, therefore the number and therefore the flow rate. The gain at the intersection of two channels is therefore limited to a lower value by the number of channels that can be used.

A titre d'exemple, on suppose que le gain de la raie de fréquence fex située à l'intersection de deux filtres consécutifs centrés sur les fréquences fp-1 et fp est de -6 décibels. On va alors montrer sur cet exemple numérique que la structure des moyens de synthèse d'un vocodeur selon l'art connu induit une ondulation parasite du signal de parole synthétisé.By way of example, it is assumed that the gain of the line of frequency f ex situated at the intersection of two consecutive filters centered on the frequencies fp -1 and fp is -6 decibels. We will then show on this digital example that the structure of the means of synthesis of a vocoder according to the known art induces a parasitic ripple of the synthesized speech signal.

Notons A en le niveau (en décibels) de la raie de fréquence fex du signal de parole P à un instant déterminé. A la sortie du filtre passe-bande du canal d'analyse 22 de rang p-1 (figure 2), cette raie de fréquence fex a une intensité de Aen-6 décibels. A la sortie du filtre passe-bande du canal d'analyse 22 de rang p, elle a le même niveau d'intensité de Aen-6 décibels. Notons par ailleurs Aex le niveau en décibels de cette raie de fréquence fex dans le signal d'excitation délivré par le moyen d'élaboration d'un signal d'excitation 38 de la figure 3. Toujours en référence à cette figure 3, à la sortie des modulateurs 46, des canaux de synthèses 44 de rang p-1 et p respectivement, le signal d'excitation issu du moyen 38 modulé respectivement par les données issues du canal d'analyse p-1 du canal d'analyse p, a une intensité égale à Aen-6+Aex décibels. Ces signaux issus des modulateurs de rang p-1 et p vont être appliqués respectivement sur les entrées des filtres passe-bande 48p-1 et 48p. Ces filtres vont diminuer à nouveau le gain de la raie de fréquence fex de 6 décibels. A la sortie de chacun des canaux de synthèse p-1 et p, on a donc un signal dont l'intensité sur la raie de fréquence fex est de A en + Aex -12 décibels. Dans le signal de parole synthétisé par l'additionneur, la raie de fréquence fex a donc une intensité égale à deux fois Aen+ Aex-12 décibels, c'est-à-dire une intensité de Aen+ Aex-6 décibels puisque un affaiblissement de 6 décibels correspond à une diminution de moitié de l'intensité du signal.Denote A by the level (in decibels) of the frequency line f ex of the speech signal P at a determined time. At the output of the bandpass filter of the analysis channel 22 of rank p-1 (FIG. 2), this line of frequency f ex has an intensity of A in -6 decibels. At the output of the bandpass filter of the analysis channel 22 of rank p, it has the same intensity level of Ae n -6 decibels. Note also A ex the level in decibels of this frequency line f ex in the excitation signal delivered by the means for generating an excitation signal 38 of FIG. 3. Still with reference to this FIG. 3, at the output of the modulators 46, synthesis channels 44 of rank p-1 and p respectively, the excitation signal from the means 38 modulated respectively by the data from the analysis channel p-1 from the analysis channel p, has an intensity equal to A in -6 + A ex decibels. These signals from the modulators of rank p-1 and p will be applied respectively to the inputs of the bandpass filters 48 p-1 and 48p. These filters will again decrease the gain of the frequency line f ex by 6 decibels. At the output of each of the synthesis channels p-1 and p, there is therefore a signal whose intensity on the frequency line f ex is from A to + A ex -12 decibels. In the speech signal synthesized by the adder, the frequency line f ex therefore has an intensity equal to twice A at + A ex -12 decibels, that is to say an intensity from A to + A ex - 6 decibels since a weakening of 6 decibels corresponds to a halving of the signal intensity.

On voit donc qu'un signal de parole dont la raie de fréquence fex située à l'intersection de deux filtres consécutifs et ayant une intensité de Aen + Aex décibels est transformée, par le vocodeur, en un signal de parole synthétisé dont la même raie a une intensité de Aen+ Aex -6 décibels, c'est-à-dire une intensité deux fois moindre. Cette diminution d'intensité apparait pour toutes les raies qui se trouvent à l'intersection de deux filtres passe-bande consécutifs des canaux d'analyse ou de synthèse. Sur toute la largeur de la bande spectrale couverte par le banc de filtres des canaux d'analyse et de synthèse, on a donc une modulation du gain qui est maximum pour les raies situées au milieu d'une bande d'un filtre passe-bande et qui est minimum pour les raies situées à l'intersection de deux filtres passe-bande consécutifs. Cette modulation est l'une des causes principales des défauts mentionnés précédemment.We therefore see that a speech signal whose frequency line f ex located at the intersection of two consecutive filters and having an intensity of A to + A ex decibels is transformed, by the vocoder, into a synthesized speech signal whose the same line has an intensity of A in + A ex -6 decibels, that is to say an intensity twice less. This reduction in intensity appears for all the lines which are at the intersection of two consecutive bandpass filters of the analysis or synthesis channels. Over the entire width of the spectral band covered by the filter bank of the analysis and synthesis channels, there is therefore a gain modulation which is maximum for the lines located in the middle of a band of a filter. bandpass and which is minimum for the lines located at the intersection of two consecutive bandpass filters. This modulation is one of the main causes of the faults mentioned above.

Cette ondulation peut être éliminée, ou au moins très atténuée, si on ajoute une chaîne d'analyse derrière le filtre passe-bande de chaque canal de synthèse et si on commande un modulateur de telle sorte que, pour des signaux sensiblement stationnaires, la sortie modulée du filtre passe-bande du canal de synthèse soit identique à la sortie du filtre passe-bande du canal d'analyse correspondant.This ripple can be eliminated, or at least very attenuated, if an analysis chain is added behind the bandpass filter of each synthesis channel and if a modulator is controlled so that, for substantially stationary signals, the output of the synthesis band pass filter is identical to the output of the band pass filter of the corresponding analysis channel.

Une telle chaîne d'analyse est décrite dans le brevet américain n°2996576 délivré le 15 août 1961 et intitulé "Feedback Vocoder". Selon ce brevet, la correction du signal de sortie du filtre passe-bande d'un canal de synthèse se fait en boucle fermée.Such an analysis chain is described in American patent n ° 2996576 issued on August 15, 1961 and entitled "Feedback Vocoder". According to this patent, the correction of the output signal of the bandpass filter of a synthesis channel is done in closed loop.

Chaque canal de synthèse comprend en série un premier filtre passe-bande, un modulateur et un second filtre passe-bande identique au premier. Le premier filtre passe-bande reçoit le signal d'excitation.Each synthesis channel comprises in series a first bandpass filter, a modulator and a second bandpass filter identical to the first. The first bandpass filter receives the excitation signal.

Une chaîne d'analyse est associée à chaque canal de synthèse. Chaque chaîne d'analyse comprend un moyen de mesure d'énergie, constitué d'un détecteur et d'un filtre passe-bas en série, un soustracteur et un amplificateur de courant continu. Le moyen de mesure d'énergie reçoit le signal de sortie du canal de synthèse et délivre un signal sur une entrée du soustracteur. Celui-ci reçoit sur une autre entrée le signal délivré par le canal d'analyse de même rang que le canal de synthèse et délivre sur l'entrée de commande de gain du modulateur, à travers l'amplificateur de courant continu, un signal exprimant la différence signée entre les signaux reçus sur ses entrées.An analysis chain is associated with each synthesis channel. Each analysis chain includes an energy measurement means, consisting of a detector and a low-pass filter in series, a subtractor and a direct current amplifier. The energy measurement means receives the output signal from the synthesis channel and delivers a signal to an input of the subtractor. This receives on another input the signal delivered by the analysis channel of the same rank as the synthesis channel and delivers on the gain control input of the modulator, through the DC amplifier, a signal expressing the signed difference between the signals received on its inputs.

Cette chaîne d'analyse ne peut corriger efficacement que les ondulations de faible fréquence à cause du temps de réaction inhérent à la structure en boucle fermée.This chain of analysis can only effectively correct undulations of low frequency because of the reaction time inherent in the closed loop structure.

Une autre limitation dans la vitesse de réaction de cette chaîne d'analyse provient de la nature du moyen de mesure d'énergie qui comprend un filtre passe-bas à constante de temps non nulle.Another limitation in the reaction speed of this analysis chain comes from the nature of the energy measurement means which includes a low-pass filter with non-zero time constant.

Ainsi, cette chaîne d'analyse ne peut pas corriger efficacement un accroissement injustifié rapide du signal du filtre passe-bande de sortie du canal de synthèse. Il en résulte une impression subjective désagréable à l'écoute du signal de parole synthétisé.Thus, this analysis chain cannot effectively correct a rapid unjustified increase in the signal of the output bandpass filter of the synthesis channel. This results in an unpleasant subjective impression when listening to the synthesized speech signal.

Le but de l'invention est précisément de supprimer ces défauts. Pour cela, il convient d'associer à chaque canal de synthèse une chaîne d'analyse apte à corriger rapidement le signal du filtre passe-bande de sortie de ce canal de synthèse.The object of the invention is precisely to eliminate these defects. For this, it is advisable to associate with each synthesis channel an analysis chain capable of rapidly correcting the signal of the output bandpass filter of this synthesis channel.

A cette fin, on se propose d'utiliser une chaîne d'analyse en boucle ouverte ce qui permet de supprimer le temps de réaction. Cette chaîne d'analyse comprend un détecteur de crête recevant le signal délivré par le filtre passe-bande du canal de synthèse, un soustracteur recevant le signal délivré par le détecteur de crête et le signal d'excitation et un modulateur qui reçoit sur son entrée de signal le signal de sortie du filtre passe-bande du canal de synthèse et sur son entrée de gain le signal délivré par le soustracteur.To this end, it is proposed to use an open loop analysis chain which makes it possible to eliminate the reaction time. This analysis chain includes a peak detector receiving the signal delivered by the bandpass filter of the synthesis channel, a subtractor receiving the signal delivered by the peak detector and the excitation signal and a modulator which receives on its input. signal the output signal of the bandpass filter of the synthesis channel and on its gain input the signal delivered by the subtractor.

Cette chaîne d'analyse réagit plus vite que la chaîne d'analyse du brevet cité car la correction de gain appliquée au modulateur ne modifie pas l'analyse en cours dans le canal de synthèse. La convergence entre les niveaux des signaux d'analyse et de synthèse dépend donc moins des constantes de temps dues aux éléments de la chaîne d'analyse.This analysis chain reacts faster than the analysis chain of the cited patent because the gain correction applied to the modulator does not modify the analysis in progress in the synthesis channel. Convergence between the levels of analysis and syn signals thesis therefore depends less on time constants due to the elements of the analysis chain.

D'autre part, le remplacement du filtre passe-bas par un détecteur de crête, dont la constante de temps est nulle à la montée, permet également de corriger très rapidement tout accroissement injustifié du signal du filtre passe-bande du canal de synthèse.On the other hand, replacing the low-pass filter with a peak detector, whose time constant is zero on the rise, also makes it possible to very quickly correct any unjustified increase in the signal of the band-pass filter of the synthesis channel.

On peut toutefois utiliser un filtre passe-bas dans le dispositif de l'invention pour mesurer l'énergie. Ceci limite un peu les performances de la chaîne d'analyse, celles-ci restant toutefois nettement supérieures grâce à la structure en boucle ouverte à la chaîne d'analyse du brevet cité.However, a low-pass filter can be used in the device of the invention to measure the energy. This somewhat limits the performance of the analysis chain, which however remains clearly superior thanks to the open loop structure of the analysis chain of the cited patent.

De manière plus précise, l'invention a pour objet un sous-ensemble de synthèse d'un vocodeur à canaux comprenant un moyen de réception et d'extraction des données d'une trame, un moyen d'élaboration d'un signal d'excitation et comprenant en outre :

  • - n canaux de synthèse numérotés de 1 à n, comportant chacun un filtre passe-bande et un moyen de mesure d'énergie pour mesurer l'énergie contenue dans ladite bande, lesdits canaux de synthèse recevant en entrée le signal d'excitation issu du moyen d'élaboration du signal d'excitation,
  • - au moins un soustracteur recevant sur une première entrée le signal d'énergie délivré par un canal de synthèse et sur une deuxième entrée le signal issu du moyen de réception et représentant le signal d'énergie délivré par le canal d'analyse correspondant et délivrant un signal exprimant la différence signée entre les deux signaux appliqués en entrée,
  • - au moins un modulateur associé au soustracteur dont l'entrée qui commande le gain reçoit le signal délivré par le soustracteur, ledit sous-ensemble de synthèse étant tel que les canaux de synthèse, le (ou les) soustracteur (s) et le (ou les) modulateur(s) associés sont disposés en cascade dans cet ordre et constituent un circuit en boucle ouverte, le sous-ensemble de synthèse étant arrangé de manière à comporter au moins un moyen de traitement de signal incluant un soustracteur et son modulateur associé, ledit moyen de traitement comportant un circuit de quantification et codage qui reçoit en entrée le signal d'énergie d'un canal de synthèse pour délivrer à l'une des entrées du soustracteur un signal d'énergie quantifié et codé, l'autre entrée du soustracteur recevant le signal d'énergie issu du moyen de réception et délivré par le canal d'analyse correspondant, le modulateur recevant sur son entrée de signal à moduler le signal de sortie du filtre passe-bande du canal de synthèse.
More specifically, the subject of the invention is a sub-assembly for synthesizing a channel vocoder comprising means for receiving and extracting data from a frame, means for generating a signal excitement and further comprising:
  • n synthesis channels numbered from 1 to n, each comprising a bandpass filter and an energy measurement means for measuring the energy contained in said band, said synthesis channels receiving as input the excitation signal from the means of generating the excitation signal,
  • - at least one subtractor receiving on a first input the energy signal delivered by a synthesis channel and on a second input the signal coming from the reception means and representing the energy signal delivered by the corresponding analysis channel and delivering a signal expressing the signed difference between the two signals applied as input,
  • - at least one modulator associated with the subtractor whose input which controls the gain receives the signal delivered by the subtractor, said synthesis subset being such that the synthesis channels, the subtractor (s) and the associated modulator (s) are arranged in cascade in this order and constitute an open loop circuit, the synthesis sub-assembly being arranged so as to comprise at least one signal processing means including a subtractor and its associated modulator, said processing means comprising a quantization and coding circuit which receives as input the energy signal of a synthesis channel for delivering a quantized and coded energy signal to one of the inputs of the subtractor, the other input of the subtractor receiving the energy signal from the reception means and delivered by the corresponding analysis channel, the modulator receiving on its signal input to modulate the output signal of the bandpass filter of the synthesis channel.

L'invention a aussi pour objet un vocodeur à canaux comprenant un sous-ensemble d'analyse d'un signal de parole et un sous-ensemble de synthèse du signal analysé, ces deux sous-ensembles étant reliés par une ligne de transmission, ledit sous-ensemble d'analyse comprenant un moyen de détection et de mesure d'un signal d'excitation, n canaux d'analyse numérotés de 1 à n constitués chacun d'un filtre passe-bande et d'un moyen pour mesurer l'énergie contenue dans ladite bande, l'ensemble des filtres passe-bandes des n canaux d'analyse formant un banc de filtres contigus, au moins un moyen de codage pour coder numériquement les données issues du moyen de mesure et un moyen pour mettre en trame et transmettre les données issues du moyen de détection et des moyens de codage, ledit sous-ensemble de synthèse comprenant un moyen de réception et d'extraction des données de la trame, un moyen d'élaboration du signal d'excitation, et dans lequel le sous-ensemble de synthèse comprend en outre :

  • - n canaux de synthèse numérotés de 1 à n, constitués chacun d'un filtre passe-bande identique au filtre passe-bande du canal d'analyse de même rang et d'un moyen de mesure d'énergie pour mesurer l'énergie contenue dans ladite bande, lesdits canaux de synthèse recevant en entrée le signal d'excitation issu du moyen d'élaboration du signal d'excitation,
  • - au moins un soustracteur recevant sur une première entrée le signal d'énergie délivré par un canal de synthèse et sur une deuxième entrée le signal issu du moyen de réception et représentant le signal d'énergie délivré par le canal d'analyse correspondant et délivrant un signal exprimant la différence signée entre les deux signaux appliqués en entrée,
  • - au moins un modulateur associé au soustracteur dont l'entrée qui commande le gain reçoit le signal délivré par le soustracteur, ledit sous-ensemble de synthèse étant tel que les canaux de synthèse, le (ou les) soustracteur(s) et le (ou les) modulateur(s) associés sont disposés en cascade dans cet ordre et constituent un circuit en boucle ouverte, le sous-ensemble de synthèse étant arrangé de manière à comporter au moins un moyen de traitement de signal incluant un soustracteur et son modulateur associé, ledit moyen de traitement comportant un circuit de quantification et codage qui reçoit en entrée le signal d'énergie d'un canal de synthèse pour délivrer à l'une des entrées du soustracteur un signal d'énergie quantifié et codé, l'autre entrée du soustracteur recevant le signal d'énergie issu du moyen de réception et délivré par le canal d'analyse correspondant, le modulateur recevant sur son entrée de signal à moduler le signal de sortie du filtre passe-bande du canal de synthèse.
The subject of the invention is also a channel vocoder comprising a sub-assembly for analyzing a speech signal and a sub-assembly for synthesizing the analyzed signal, these two sub-assemblies being connected by a transmission line, said analysis sub-assembly comprising means for detecting and measuring an excitation signal, n analysis channels numbered from 1 to n each consisting of a bandpass filter and a means for measuring the energy contained in said band, all the band-pass filters of the n analysis channels forming a bank of contiguous filters, at least one coding means for digitally coding the data coming from the measurement means and a means for framing and transmit the data coming from the detection means and from the coding means, said synthesis sub-assembly comprising a means for receiving and extracting the data from the frame, a means for generating the excitation signal, and in which the synthesis subset further comprises:
  • - n synthesis channels numbered from 1 to n, each consisting of a bandpass filter identical to the bandpass filter of the analysis channel of the same rank and of an energy measurement means for measuring the energy contained in said band, said synthesis channels receiving as input the excitation signal originating from the means for generating the excitation signal,
  • - At least one subtractor receiving on a first input the energy signal delivered by a synthesis channel and on a second input the signal coming from the reception means and representing the energy signal delivered by the corresponding analysis channel and delivering a signal expressing the signed difference between the two signals applied as input,
  • - at least one modulator associated with the subtractor whose input which controls the gain receives the signal delivered by the subtractor, said synthesis subset being such that the synthesis channels, the (or) subtractor (s) and the ( or the) associated modulator (s) are arranged in cascade in this order and constitute an open loop circuit, the synthesis sub-assembly being arranged so as to comprise at least one signal processing means including a subtractor and its associated modulator , said processing means comprising a quantization and coding circuit which receives as an input the energy signal of a synthesis channel to deliver to one of the inputs of the subtractor a quantized and coded energy signal, the other input of the subtractor receiving the energy signal from the reception means and delivered by the corresponding analysis channel, the modulator receiving on its signal input to modulate the output signal of the bandpass filter of the synthesis channel.

Selon un mode de réalisation avantageux, le vocodeur selon l'invention comprend n moyens de traitement de signal, numérotés de 1 à n, chaque moyen de traitement de signal étant associé à un canal de synthèse.According to an advantageous embodiment, the vocoder according to the invention comprises n signal processing means, numbered from 1 to n, each signal processing means being associated with a synthesis channel.

Selon un mode de réalisation préféré, le vocodeur selon l'invention comprend un unique moyen de traitement de signal dont l'entrée du circuit de quantification et codage est reliée à un multiplexeur dont n entrées reçoivent le signal d'énergie de chaque canal de synthèse et dont l'entrée de signal du modulateur est reliée à un autre multiplexeur dont n entrées reçoivent le signal de sortie du filtre passe-bande de chaque canal de synthèse.According to a preferred embodiment, the vocoder according to the invention comprises a single signal processing means whose input of the quantization and coding circuit is connected to a multiplexer whose n inputs receive the energy signal from each synthesis channel and the modulator signal input of which is connected to another multiplexer of which n inputs receive the output signal of the bandpass filter of each synthesis channel.

Selon un mode de réalisation particulier, le vocodeur comprend en outre un moyen de lissage situé entre le soustracteur et le modulateur, ledit moyen de lissage effectuant dans chaque canal un lissage de telle sorte que la différence d'intensité entre deux données successives issues d'un même canal reste inférieure à une valeur prédéterminée.According to a particular embodiment, the vocoder further comprises a smoothing means located between the subtractor and the modulator, said smoothing means performing in each channel a smoothing so that the difference in intensity between two successive data from the same channel remains below a predetermined value.

Selon une caractéristique secondaire, le moyen de lissage comprend un comparateur dont une entrée reçoit le signal issu du soustracteur et dont l'autre entrée reçoit un autre signal, un moyen de calcul de signal de gain, des moyens de mémorisation de ce signal de gain, ledit moyen de calcul déterminant pour une trame donnée ledit gain en fonction du signal de sortie du comparateur et du signal de gain mémorisé de la trame précédente, lesdits moyens de mémorisation fournissant ledit autre signal.According to a secondary characteristic, the smoothing means comprises a comparator, one input of which receives the signal from the subtractor and the other input of which receives another signal, a means for calculating a gain signal, means for storing this gain signal , said calculation means determining for a given frame said gain as a function of the output signal of the comparator and of the stored gain signal of the previous frame, said storage means supplying said other signal.

Selon une autre caractéristique secondaire, chaque soustracteur comporte sur son entrée inverseuse des moyens basculeurs permettant de recevoir sur ladite entrée inverseuse soit le signal issu du moyen de codage assoc.ié, soit un signal de niveau fixe prédéterminé.According to another secondary characteristic, each subtractor comprises on its inverting input tilting means making it possible to receive on ladi the inverting input is either the signal from the associated coding means, or a predetermined fixed level signal.

Selon une autre caractéristique secondaire, chaque moyen de mesure d'énergie comprend en série un redresseur et un détecteur de crête.According to another secondary characteristic, each energy measurement means comprises in series a rectifier and a peak detector.

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

  • - la figure 1 déjà décrite, représente la structure générale d'un vocodeur ;
  • - la figure 2 déjà décrite, représente schématiquement la structure de la partie analyse d'un vocodeur à canaux de type connu utilisée dans le vocodeur de l'invention
  • - la figure 3 déjà décrite, représente schématiquement la partie synthèse d'un vocodeur à canaux d'un type connu ;
  • - les figures 4a et 4b déjà décrites, sont des diagrammes illustrant le gain théorique et réel dans chaque canal d'un banc de filtres passe-bande ;
  • - la figure 5 illustre schématiquement un mode de réalisation de la partie synthèse d'un vocodeur à canaux selon l'invention ;
  • - la figure 6 illustre schématiquement un moyen de lissage de type numérique utilisé dans un deuxième mode de réalisation d'un vocodeur à canaux selon l'invention.
Other characteristics and advantages of the invention will emerge more clearly from the description which follows, given by way of illustration and not limitation, with reference to the appended figures in which:
  • - Figure 1 already described, shows the general structure of a vocoder;
  • - Figure 2 already described, schematically shows the structure of the analysis part of a vocoder with known type of channels used in the vocoder of the invention
  • - Figure 3 already described, schematically shows the synthesis part of a vocoder with channels of a known type;
  • - Figures 4a and 4b already described, are diagrams illustrating the theoretical and real gain in each channel of a bank of bandpass filters;
  • - Figure 5 schematically illustrates an embodiment of the synthesis part of a channel vocoder according to the invention;
  • - Figure 6 schematically illustrates a digital type smoothing means used in a second embodiment of a channel vocoder according to the invention.

Le vocodeur à canaux selon l'invention dont le sous-ensemble de synthèse est représenté schématiquement sur la figure 5, reçoit sur un moyen de réception 36, une trame de données constituant une repré- sentation numérique d'un signal de parole analysé par le sous-ensemble d'analyse dudit vocodeur. Ce moyen délivre un premier signal à un moyen d'élaboration du signal d'excitation 38 et un deuxième signal à un moyen de mémorisation 40. Jusqu'à ce niveau, le sous-ensemble de synthèse du vocodeur selon l'invention est identique au sous-ensemble de synthèse de vocodeur d'un type connu décrit à la figure 3.Channels to the vocoder according to the invention, the subset of synthesis is shown schematically in Figure 5, receives a receiving means 36, a data frame comprising a repre - sentation of a digital speech signal analyzed by the analysis subset of said vocoder. This means delivers a first signal to a means for generating the excitation signal 38 and a second signal to a storage means 40. Up to this level, the vocoder synthesis sub-assembly according to the invention is identical to the subset of vocoder synthesis of a known type described in FIG. 3.

Le signal d'excitation élaboré par le moyen 38 est appliqué sur les entrées des n canaux de synthèse 45, respectivement identiques aux canaux d'analyse de même rang. Ils comprennent chacun un filtre passe-bande 48, un redresseur 50 et un filtre passe-bas 52. Ce dernier peut-être avantageusement remplacé par un détecteur de crête qui suit plus rapidement qu'un filtre passe-bas les possibles augmentations brutale d'énergie localisée en fréquence du signal d'excitation, de manière à ce que ces augmentations soient compensées le plus rapidement possible par le traitement dans la chaîne de synthèse selon l'invention.The excitation signal produced by the means 38 is applied to the inputs of the n synthesis channels 45, respectively identical to the analysis channels of the same rank. They each include a band-pass filter 48, a rectifier 50 and a low-pass filter 52. The latter can advantageously be replaced by a peak detector which follows the possible abrupt increases in speed more quickly than a low-pass filter. energy localized in frequency of the excitation signal, so that these increases are compensated as quickly as possible by the processing in the synthesis chain according to the invention.

Les signaux délivrés par chacun de ces canaux de synthèse 45 sont ensuite quantifiés et codés par un moyen de quantification et codage 56 auquel ils sont reliés au travers d'un multiplexeur 54 de sortie Sl. Le moyen 56 effectue le même codage que le moyen de quantification et codage du sous-ensemble d'analyse.The signals delivered by each of these synthesis channels 45 are then quantified and coded by a quantization and coding means 56 to which they are connected through an output multiplexer 54 S l . The means 56 performs the same coding as the quantization and coding means of the analysis subset.

Le signal d'énergie extrait des canaux de synthèse 45 et du moyen 56 est appliqué sur l'entrée inverseuse d'un soustracteur 62 dont l'autre entrée reçoit le signal délivré par le moyen de mémorisation 40 qui contient les données issues des canaux d'analyse. Le soustracteur 62 délivre en sortie la- différence signée des signaux appliqués sur chacune de ses entrées.The energy signal extracted from the synthesis channels 45 and from the means 56 is applied to the inverting input of a subtractor 62 whose other input receives the signal delivered by the storage means 40 which contains the data from the channels d 'analysis. The subtractor 62 delivers at output the signed difference of the signals applied to each of its inputs.

Ce signal est traité par un moyen de lissage 64 qui sera décrit plus en détail en référence à la figure suivante. Le moyen de lissage 64 réalise un lissage par canal. Il assure que deux données consécutives issues du même canal ne diffèrent pas de plus d'une certaine valeur prédéterminée par exemple 1 décibel. Ce moyen peut être omis.This signal is processed by a smoothing means 64 which will be described in more detail with reference to the following figure. The smoothing means 64 performs smoothing per channel. It ensures that two consecutive data from the same channel do not differ by more than a certain predetermined value, for example 1 decibel. This plea can be omitted.

La sortie du moyen de lissage est reliée à l'entrée commandant le gain d'un modulateur 46. Sur l'autre entrée du modulateur 46, on va appliquer un signal issu de la sortie S.2 d'un multiplexeur 66 et qui reçoit sur chacune de ses entrées un signal prélevé à la sortie des filtres passe-bande 48 de chacun des n canaux de synthèse. Le signal délivré par le modulateur 46, après traitement par un moyen de désac- centuation 58 et un ensemble décodeur-filtre passe-bas 60 identique à ceux utilisés dans le vocodeur de type connu décrit à la figure 3, représente le signal de parole synthétisé S.The output of the smoothing means is connected to the input controlling the gain of a modulator 46. On the other input of the modulator 46, we will apply a signal from the output S. 2 of a multiplexer 66 and which receives on each of its inputs a signal taken at the output of the bandpass filters 48 of each of the n synthesis channels. The signal delivered by the modulator 46, after processing by a deactivation means 58 and a decoder-low-pass filter assembly 60 identical to those used in the vocoder of known type described in FIG. 3, represents the synthesized speech signal S.

L'ensemble constitué par le moyen 56, le soustracteur 62, le moyen de lissage facultatif 64 et le modulateur 46 constitue un moyen de traitement de signal 63 dont une première entrée est reliée à la sortie S1 et une deuxième entrée est reliée à la sortie S-2.The assembly constituted by the means 56, the subtractor 62, the optional smoothing means 64 and the modulator 46 constitutes a signal processing means 63, a first input of which is connected to the output S 1 and a second input of which is connected to the exit S -2 .

Le vocodeur selon l'invention dont le sous-ensemble de synthèse vient d'être décrit a une structure qui permet d'éliminer les ondulations parasites du signal de parole synthétisé apparaissant dans les vocodeurs selon l'art connu. En effet, reprenons l'exemple numérique utilisé pour illustrer les imperfections des vocodeurs selon l'art connu. On considère une raie de fréquence fex qui se trouve à l'intersection de deux filtres passe-bande de canaux d'analyse ou de synthèse consécutifs p-1 et p. Dans le signal de parole initial, cette raie a une intensité A et à la sortie des filtres d'analyse, elle a une intensité de Aen-6 dB. L'intensité de cette raie est contenue dans le signal qui est mémorisé par le moyen de mémorisation 40 de la partie de synthèse du vocodeur. D'autre part, le moyen 38 d'élaboration du signal d'excitation va fournir une raie de fréquence fex d'intensité A , comme dans le vocodeur décrit en référence à la figure 3. Avec le vocodeur selon l'invention décrit à la figure 5, ce signal d'excitation va être traité par les canaux de synthèse de rang p-1 et p identiques aux canaux d'analyse de même rang. A la sortie de chacun de ces filtres de synthèse, l'intensité de la raie de fréquence fex sera de A - 6 dB.The vocoder according to the invention, the synthesis subset of which has just been described, has a structure which makes it possible to eliminate parasitic ripples from the synthesized speech signal appearing in vocoders according to known art. Indeed, let’s take the digital example used to illustrate the imperfections of vocoders according to known art. We consider a line of frequency f ex which is at the intersection of two bandpass filters of analysis channels or consecutive synthesis p-1 and p. In the initial speech signal, this line has an intensity A and at the output of the analysis filters, it has an intensity of A in -6 dB. The intensity of this line is contained in the signal which is stored by the storage means 40 of the synthesis part of the vocoder. On the other hand, the means 38 for generating the excitation signal will provide a line of frequency f ex of intensity A, as in the vocoder described with reference to FIG. 3. With the vocoder according to the invention described in FIG. 5, this excitation signal will be processed by the synthesis channels of rank p-1 and p identical to the analysis channels of the same rank. At the output of each of these synthesis filters, the intensity of the frequency line f ex will be A - 6 dB.

Pour le canal de rang p-1, on a donc à l'entrée du soustracteur 62 sur l'entrée inverseuse, un nombre représentant une intensité A - 6 dB et sur l'entrée non inverseuse, un nombre représentant une intensité A - 6 dB issu du canal d'analyse de rang p-l. A la sortie de ce soustracteur 62, on a donc pour la raie de fréquence fex une différence de A - Aex dB. Ce nombre module dans le modulateur 46 le gain du signal délivré par le filtre passe-bande 48p-1 du canal de synthèse de rang p-1 et dont la raie de fréquence fex a une intensité de Aex - 6 décibels. En sortie de ce modulateur 46, on a donc un signal dont la raie de fréquence f ex a une intensité de (Aen-Aex)+(Aex-6)dB, c'est-à-dire A - 6 dB.For the channel of rank p-1, there is therefore at the input of the subtractor 62 on the inverting input, a number representing an intensity A - 6 dB and on the non-inverting input, a number representing an intensity A - 6 dB from the pl rank analysis channel. At the output of this subtractor 62, there is therefore for the frequency line f ex a difference of A - A ex dB. This number modulates in the modulator 46 the gain of the signal delivered by the band-pass filter 48 p-1 of the synthesis channel of rank p-1 and whose frequency line f ex has an intensity of A ex - 6 decibels. At the output of this modulator 46, there is therefore a signal whose frequency line f ex has an intensity of (A in -A ex ) + (A ex -6) dB, that is to say A - 6 dB .

en Le calcul précédent a été effectué pour le canal p-1. En fait, la raie de fréquence fex est traitée par les canaux p-1 et p. On a donc aussi, en sortie du modulateur 46, un signal issu du canal de rang p contenant une raie de fréquence fex d'intensité A - 6 dB. En sommant sur les deux canaux p-1 et p, la raie de fréquence fex a donc une intensité de deux fois A - 6 dB, c'est-à-dire une intensité de l'ordre en A décibels.en The previous calculation was performed for channel p- 1. In fact, the frequency line f ex is processed by channels p-1 and p. There is therefore also, at the output of the modulator 46, a signal coming from the channel of rank p containing a line of frequency f ex of intensity A - 6 dB. By summing on the two channels p-1 and p, the frequency line f ex therefore has an intensity of twice A - 6 dB, that is to say an intensity of the order of A decibels.

Ainsi, dans le signal de parole synthétisé S délivré par le vocodeur à canaux selon l'invention, l'intensité des raies dont la fréquence est à l'intersection de deux filtres passe-bande consécutifs de canaux d'analyse ou de synthèse, est restituée sans affaiblissement. L'ondulation parasite qui apparaissait dans les vocodeurs selon l'art antérieur a disparu. De plus, contrairement à ce qui se produit dans les vocodeurs de l'art antérieur, la raie de fréquence fex est restituée avec une intensité A en indépendante de l'intensité Aex de la raie de fréquence fex du signal d'excitation. Il en résulte une correction de l'ondulation éventuelle de l'enveloppe spectrale du signal d'excitation. Ceci est particulièrement intéressant pour les vocodeurs à bande de base ou à excitation vocale.Thus, in the synthesized speech signal S delivered by the channel vocoder according to the invention, the intensity of the lines whose frequency is at the intersection of two consecutive bandpass filters of analysis or synthesis channels, is restored without weakening. The parasitic ripple which appeared in the vocoders according to the prior art has disappeared. In addition, contrary to what occurs in the vocoders of the prior art, the line of frequency f ex is restored with an intensity A independently of the intensity A ex of the line of frequency f ex of the excitation signal . This results in a correction of the possible ripple of the spectral envelope of the excitation signal. This is particularly interesting for baseband or voice excitation vocoders.

On va maintenant décrire en détail en référence à la figure 6, le moyen de lissage 64. Ce moyen de lissage 64 permet de n'avoir, pour deux données consécutives issues d'un même canal qu'une différence au plus égale à une valeur prédéterminée, par exemple 1 décibel. Ce moyen de lissage 64 comprend en série un comparateur 66 et un moyen de calcul 68, pour calculer le gain corrigé de la donnée appliquée en entrée du moyen de lissage 64. Il comprend aussi un moyen 70, pour mémoriser le gain corrigé calculé par le moyen 68, qui est relié au comparateur 66.We will now describe in detail with reference to FIG. 6, the smoothing means 64. This smoothing means 64 makes it possible to have, for two consecutive data originating from the same channel, only a difference at most equal to a value predetermined, for example 1 decibel. This smoothing means 64 comprises in series a comparator 66 and a calculation means 68, for calculating the corrected gain of the data applied at the input of the smoothing means 64. It also includes means 70, for storing the corrected gain calculated by the means 68, which is connected to comparator 66.

Les données issues d'un canal de synthèse sont traitées par le moyen de lissage de la manière suivante. Le comparateur 66 reçoit sur une entrée non inverseuse les données issues du soustracteur 62 et qui sont extraites de la trame de rang i, et sur une entrée inverseuse un signal Gi-1 qui représente le gain des données issues de la trame de rang i-1 et correspondant au même canal. Le moyen de calcul 68 va délivrer en sortie un signal de gain Gi qui est mémorisé dans le moyen 70 et qui est appliqué sur l'entrée du modulateur 46.The data from a synthesis channel are processed by the smoothing means in the following manner. The comparator 66 receives on a non-inverting input the data coming from the subtractor 62 and which are extracted from the frame of rank i, and on a inverting input a signal G i-1 which represents the gain of the data coming from the frame of rank i-1 and corresponding to the same channel. The calculation means 68 will output a gain signal G i which is stored in the means 70 and which is applied to the input of the modulator 46.

On a représenté sur la figure 6, un mode de réalisation particulier de ce moyen de lissage. Le moyen 68 comprend en série un moyen 72 pour déterminer un coefficient εi du gain, un multiplicateur 74 et un additionneur 76. Le moyen 70 comprend une mémoire 78 et un moyen de quantification 80.FIG. 6 shows a particular embodiment of this smoothing means. The means 68 comprises in series a means 72 for determining a coefficient ε i of the gain, a multiplier 74 and an adder 76. The means 70 comprises a memory 78 and a quantization means 80.

Le moyen 72 reçoit du comparateur numérique 66 un signal indiquant si la différence entre les signaux appliqués à l'entrée dudit comparateur numérique est positive, nulle ou négative. Le coefficient εi délivré par ce moyen 72 peut être égal, à titre d'exemple à 2-m si la différence est positive, 0 si elle et nulle et -2-m si elle est négative, le choix de la valeur de m prédéterminé définissant la constante de temps de filtrage du moyen de lissage 64. Ce coefficient εi est multiplié par le multiplicateur 74 avec le gain Gi-1 mémorisé dans la mémoire 78. Le résultat obtenu est ajouté dans l'additionneur 76 au gain Gi-1 pour donner un gain Gi égal à Gi-1 (1 + i). Ce gain Gi mémorisé dans la mémoire 78 est quantifié par le moyen de quantification 80 pour pouvoir être comparé au signal délivré par le soustracteur 62.The means 72 receives from the digital comparator 66 a signal indicating whether the difference between the signals applied to the input of said digital comparator is positive, zero or negative. The coefficient ε i delivered by this means 72 can be equal, for example to 2- m if the difference is positive, 0 if it is zero and -2 -m if it is negative, the choice of the value of m predetermined defining the filtering time constant of the smoothing means 64. This coefficient ε i is multiplied by the multiplier 74 with the gain G i-1 stored in memory 78. The result obtained is added in the adder 76 to the gain G i-1 to give a gain G i equal to G i-1 ( 1 + i ). This gain G i stored in memory 78 is quantified by the quantization means 80 so that it can be compared to the signal delivered by the subtractor 62.

Selon une variante de réalisation de l'ensemble de synthèse, un moyen de commutation peut être prévu à l'entrée inverseuse du soustracteur 62 pour recevoir, en fonction de l'état du moyen de commutation, soit le signal provenant du moyen 56 (pour un fonctionnement selon l'invention), soit un signal de niveau fixe prédéterminé. Dans ce dernier cas, le fonctionnement du sous-ensemble de synthèse est analogue à celui du vocodeur à canaux classique.According to an alternative embodiment of the synthesis assembly, a switching means can be provided at the inverting input of the subtractor 62 to receive, depending on the state of the switching means, either the signal from the means 56 (for an operation according to the invention), ie a predetermined fixed level signal. In the latter case, the operation of the synthesis subset is analogous to that of the conventional channel vocoder.

Claims (8)

1. Sous-ensemble de synthèse d'un vocodeur à canaux comprenant un moyen (36) de réception et d'extraction des données d'une trame, un moyen (38) d'élaboration d'un signal d'excitation, ledit sous-ensemble de synthèse comprenant : - n canaux de synthèse (45) numérotés de 1 à n, comportant chacun un filtre passe-bande (48) et un moyen de mesure d'énergie (50, 52) pour mesurer l'énergie contenue dans ladite bande, lesdits canaux de synthèse recevant en entrée le signal d'excitation issu du moyen (38) d'élaboration du signal d'excitation, - au moins un soustracteur (62) recevant sur une première entrée le signal d'énergie délivré par un canal de synthèse et sur une deuxième entrée le signal issu du moyen de réception (36) et représentant le signal d'énergie délivré par le canal d'analyse correspondant et délivrant un signal exprimant la différence signée entre les deux signaux appliqués en entrée, - au moins un modulateur (46) associé au soustracteur, dont l'entrée qui commande le gain reçoit le signal délivré par le soustracteur, ledit sous-ensemble de synthèse étant caractérisé en ce que les canaux de synthèse (45), le (ou les) soustracteur(s) (62) et le (ou les) modulateur(s) (46) associés sont disposés en cascade dans cet ordre et constituent un circuit en boucle ouverte, le sous-ensemble de synthèse étant arrangé de manière à comporter au moins un moyen de traitement (63) de signal incluant un soustracteur (62) et son modulateur (46) associé, ledit moyen de traitement (63) comportant un circuit de quantification et codage (56) qui reçoit en entrée le signal d'énergie d'un canal de synthèse (45) pour délivrer à l'une des entrées du soustracteur un signal d'énergie quantifié et codé, l'autre entrée du soustracteur recevant le signal d'énergie issu du moyen de réception et délivré par le canal d'analyse correspondant, le modulateur recevant sur son entrée de signal à moduler le signal de sortie du filtre passe-bande du canal de synthèse. 1. Sub-assembly for synthesizing a channel vocoder comprising means (36) for receiving and extracting data from a frame, means (38) for generating an excitation signal, said sub - synthesis package including: - n synthesis channels (45) numbered from 1 to n, each comprising a bandpass filter (48) and an energy measurement means (50, 52) for measuring the energy contained in said band, said channels synthesis receiving as input the excitation signal from the means (38) for developing the excitation signal, - at least one subtractor (62) receiving on a first input the energy signal delivered by a synthesis channel and on a second input the signal coming from the reception means (36) and representing the energy signal delivered by the channel corresponding analysis and delivering a signal expressing the difference signed between the two signals applied as input, - at least one modulator (46) associated with the subtractor, the input of which controls the gain receives the signal delivered by the subtractor, said synthesis sub-assembly being characterized in that the synthesis channels (45), the (or the) subtractor (s) (62) and the associated modulator (s) (46) are arranged in cascade in this order and constitute an open loop circuit, the synthesis sub-assembly being arranged so as to comprise at least one signal processing means (63) including a subtractor (62) and its associated modulator (46), said processing means (63) comprising a quantization and coding circuit (56) which receives in input the energy signal of a synthesis channel (45) to deliver to one of the inputs of the subtractor a quantized and coded energy signal, the other input of the subtractor receiving the energy signal from the means reception and delivered by the corresponding analysis channel, the modulator receiving on its signal input to modulate the output signal of the bandpass filter of the synthesis channel. 2. Vocodeur à canaux comprenant un sous-ensemble d'analyse (2) d'un signal de parole et un sous-ensemble de synthèse (6) du signal analysé, ces deux sous-ensembles étant reliés par un organe de transmission (4), ledit sous-ensemble d'analyse comprenant un moyen (18) de détection et de mesure d'un signal décrivant la structure fine du spectre du signal de parole, n canaux d'analyse (22) numérotés de 1 à n constitués chacun d'un filtre passe-bande (24) et d'un moyen (26, 28) pour mesurer l'énergie instantanée contenue dans ladite bande, l'ensemble des filtres passe-bande desdits n canaux d'analyse formant un banc de filtres contigus, au moins un moyen de quantification et codage (32) pour quantifier et coder les données issues des canaux d'analyse, et un moyen pour mettre en trame et transmettre les données issues dudit moyen de détection et de mesure et dudit moyen de codage, ledit sous-ensemble de synthèse comprenant un moyen (36) de réception et d'extraction des données de la trame, un moyen (38) d'élaboration du signal d'excitation, ledit sous-ensemble de synthèse étant caractérisé en ce qu'il comprend en outre : - n canaux de synthèse (45) numérotés de 1 à n, constitués chacun d'un filtre passe-bande (48) identique au filtre passe-bande du canal d'analyse de même rang et d'un moyen de mesure d'énergie (50, 52) pour mesurer l'énergie contenue dans ladite bande, lesdits canaux de synthèse recevant en entrée le signal d'excitation issu du moyen (38) d'élaboration du signal d'excitation, - au moins un soustracteur (62) recevant sur une première entrée le signal d'énergie délivré par un canal de synthèse et sur une deuxième entrée le signal issu du moyen de réception (36) et représentant le signal d'énergie délivré par le canal d'analyse correspondant et délivrant un signal exprimant la différence signée entre les deux signaux appliqués en entrée, - au moins un modulateur (46) associé au soustracteur dont l'entrée qui commande le gain reçoit le signal délivré par le soustracteur, ledit sous-ensemble de synthèse étant caractérisé en ce que les canaux de synthèse (45), le (ou les) soustracteur(s) (62) et le (ou les) modulateur (s) (46) associés sont disposés en cascade dans cet ordre et constituent un circuit en boucle ouverte, le sous-ensemble de synthèse étant arrangé de manière à comporter au moins un moyen de traitement (63) de signal incluant un soustracteur (62) et son modulateur (46) associé, ledit moyen de traitement (63) comportant un circuit de quantification et codage (56) qui reçoit en entrée le signal d'énergie d'un canal de synthèse (45) pour délivrer à l'une des entrées du soustracteur un signal d'énergie quantifié et codé, l'autre entrée du soustracteur recevant le signal d'énergie issu du moyen de réception et délivré par le canal d'analyse correspondant, le modulateur recevant sur son entrée de signal à moduler le signal de sortie du filtre passe-bande du canal de synthèse. 2. Channel vocoder comprising a sub-assembly for analyzing (2) a speech signal and a sub-assembly for synthesizing (6) the analyzed signal, these two sub-assemblies being connected by a transmission member (4 ), said analysis sub-assembly comprising means (18) for detecting and measuring a signal describing the fine structure of the spectrum of the speech signal, n analysis channels (22) numbered from 1 to n each constituted a bandpass filter (24) and a means (26, 28) for measuring the instantaneous energy contained in said band, the set of bandpass filters of said n analysis channels forming a filter bank contiguous, at least one quantization and coding means (32) for quantifying and coding the data from the analysis channels, and a means for framing and transmitting the data from said detection and measurement means and said coding means , said synthesis subset comprising means (36) for receiving and extracting data from the frame, u n means (38) for generating the excitation signal, said synthesis sub-assembly being characterized in that it further comprises: - n synthesis channels (45) numbered from 1 to n, each consisting of a bandpass filter (48) identical to the bandpass filter of the analysis channel of the same rank and of an energy measurement means (50, 52) to measure the energy contained in said strip, said synthesis channels receiving as input the excitation signal coming from the means (38) for generating the excitation signal, - at least one subtractor (62) receiving on a first input the energy signal delivered by a synthesis channel and on a second input the signal coming from the reception means (36) and representing the energy signal delivered by the channel corresponding analysis and delivering a signal expressing the difference signed between the two signals applied as input, - at least one modulator (46) associated with the subtractor whose input which controls the gain receives the signal delivered by the subtractor, said synthesis sub-assembly being characterized in that the synthesis channels (45), the (or ) subtractor (s) (62) and the associated modulator (s) (46) are arranged in cascade in this order and constitute an open loop circuit, the synthesis sub-assembly being arranged so as to include at at least one signal processing means (63) including a subtractor (62) and its associated modulator (46), said processing means (63) comprising a quantization and coding circuit (56) which receives the energy signal as an input a synthesis channel (45) for delivering to one of the inputs of the subtractor a quantized and coded energy signal, the other input of the subtractor receiving the energy signal from the reception means and delivered by the channel corresponding analysis, the modulator receiving at its signal input to be modulated l e output signal of the bandpass filter of the synthesis channel. 3. Vocodeur selon la revendication 2, caractérisé en ce qu'il comprend n moyens de traitement de signal (63), numérotés de 1 à n, chaque moyen de traitement de signal étant associé à un canal de synthèse.3. Vocoder according to claim 2, characterized in that it comprises n signal processing means (63), numbered from 1 to n, each means of signal processing being associated with a synthesis channel. 4. Vocodeur selon la revendication 2, caractérisé en ce qu'il comprend un unique moyen de traitement de signal dont l'entrée du circuit de quantification et codage (56) est reliée à un multiplexeur (54) dont n entrées reçoivent le signal d'énergie de chaque canal de synthèse et dont l'entrée de signal du modulateur (46) est reliée à un autre multiplexeur (66) dont n entrées reçoivent le signal de sortie du filtre passe-bande de chaque canal de synthèse.4. Vocoder according to claim 2, characterized in that it comprises a single signal processing means whose input of the quantization and coding circuit (56) is connected to a multiplexer (54) of which n inputs receive the signal d energy of each synthesis channel and of which the signal input of the modulator (46) is connected to another multiplexer (66) whose n inputs receive the output signal of the bandpass filter of each synthesis channel. 5. Vocodeur selon la revendication 4, caractérisé en ce qu'il comprend en outre un moyen de lissage entre le soustracteur (62) et le modulateur (46), ledit moyen de lissage effectuant dans chaque canal un lissage de telle sorte que la différence d'intensité entre deux données successives issues d'un même canal reste inférieure à une valeur prédéterminée.5. Vocoder according to claim 4, characterized in that it further comprises a smoothing means between the subtractor (62) and the modulator (46), said smoothing means performing in each channel a smoothing so that the difference intensity between two successive data from the same channel remains below a predetermined value. 6. Vocodeur selon la revendication 5, caractérisé en ce que le moyen de lissage (64) comprend un comparateur (66) dont une entrée reçoit le signal issu du soustracteur (62) et dont l'autre entrée reçoit un autre signal, un moyen de calcul (68) de signal de gain, des moyens de mémorisation (70) de ce signal de gain (G.), ledit moyen de calcul déterminant pour une trame donnée ledit gain en fonction du signal de sortie du comparateur et du signal du gain mémorisé (Gi-1) de la trame précédente, lesdits moyens de mémorisation fournissant ledit autre signal.6. Vocoder according to claim 5, characterized in that the smoothing means (64) comprises a comparator (66), one input of which receives the signal from the subtractor (62) and the other input of which receives another signal, a means computation (68) of gain signal, storage means (70) of this gain signal (G.), said computation means determining for a given frame said gain as a function of the output signal of the comparator and of the signal of the stored gain (G i-1 ) of the previous frame, said storage means providing said other signal. 7. Vocodeur selon l'une quelconque des revendications 2 à 6, caractérisé en ce que chaque soustracteur (62) comporte sur son entrée inverseuse des moyens de commutation permettant de recevoir sur ladite entrée inverseuse soit le signal issu du moyen de codage (56) associé, soit un signal de niveau fixe prédéterminé.7. Vocoder according to any one of claims 2 to 6, characterized in that each subtractor (62) comprises on its inverting input switching means making it possible to receive on said inverting input either the signal coming from the means associated coding code (56), ie a predetermined fixed level signal. 8. Vocodeur selon l'une quelconque des revendications 2 à 7, caractérisé en ce que chaque moyen de mesure d'énergie (50, 52) du canal de synthèse comprend en série un redresseur et un détecteur de crête.8. Vocoder according to any one of claims 2 to 7, characterized in that each energy measurement means (50, 52) of the synthesis channel comprises in series a rectifier and a peak detector.
EP19840400744 1983-04-20 1984-04-13 Channel vocoder comprising means for suppressing parasitic modulation of the synthesized speech signal Expired EP0124411B1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR8306471 1983-04-20
FR8306471A FR2544901B1 (en) 1983-04-20 1983-04-20 CHANNEL VOCODER PROVIDED WITH MEANS FOR COMPENSATING FOR PARASITIC MODULATIONS OF THE SYNTHETIC SPEECH SIGNAL

Publications (2)

Publication Number Publication Date
EP0124411A1 true EP0124411A1 (en) 1984-11-07
EP0124411B1 EP0124411B1 (en) 1987-11-19

Family

ID=9288045

Family Applications (1)

Application Number Title Priority Date Filing Date
EP19840400744 Expired EP0124411B1 (en) 1983-04-20 1984-04-13 Channel vocoder comprising means for suppressing parasitic modulation of the synthesized speech signal

Country Status (3)

Country Link
EP (1) EP0124411B1 (en)
DE (1) DE3467650D1 (en)
FR (1) FR2544901B1 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0637013A1 (en) * 1993-07-28 1995-02-01 Siemens Aktiengesellschaft Process and apparatus for outputting of digital-speech-signals
EP0676744A1 (en) * 1994-04-04 1995-10-11 Digital Voice Systems, Inc. Estimation of excitation parameters

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US2672512A (en) * 1949-02-02 1954-03-16 Bell Telephone Labor Inc System for analyzing and synthesizing speech
US2996579A (en) * 1960-01-13 1961-08-15 Gen Dynamics Corp Feedback vocoder
BE670406A (en) * 1964-10-01 1966-01-31
US3252105A (en) * 1962-06-07 1966-05-17 Honeywell Inc Rate limiting apparatus including active elements
FR1577550A (en) * 1967-08-03 1969-08-08
GB2002611A (en) * 1977-08-05 1979-02-21 Brown Ltd S Electric circuit device and pendulum unit incorporating such devices

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3346695A (en) * 1963-05-07 1967-10-10 Gunnar Fant Vocoder system

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US2672512A (en) * 1949-02-02 1954-03-16 Bell Telephone Labor Inc System for analyzing and synthesizing speech
US2996579A (en) * 1960-01-13 1961-08-15 Gen Dynamics Corp Feedback vocoder
US3252105A (en) * 1962-06-07 1966-05-17 Honeywell Inc Rate limiting apparatus including active elements
BE670406A (en) * 1964-10-01 1966-01-31
FR1577550A (en) * 1967-08-03 1969-08-08
GB2002611A (en) * 1977-08-05 1979-02-21 Brown Ltd S Electric circuit device and pendulum unit incorporating such devices

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
IRE TRANSACTIONS ON VEHICULAR COMMUNICATIONS, vol. VIC-11, no. 1, août 1962, pages 65-72, New York, US; A.L. HOPPER: "An experimental fast acting AGC circuit" *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0637013A1 (en) * 1993-07-28 1995-02-01 Siemens Aktiengesellschaft Process and apparatus for outputting of digital-speech-signals
EP0676744A1 (en) * 1994-04-04 1995-10-11 Digital Voice Systems, Inc. Estimation of excitation parameters

Also Published As

Publication number Publication date
FR2544901A1 (en) 1984-10-26
EP0124411B1 (en) 1987-11-19
FR2544901B1 (en) 1986-02-21
DE3467650D1 (en) 1987-12-23

Similar Documents

Publication Publication Date Title
EP0111612B1 (en) Speech signal coding method and apparatus
EP0002998B1 (en) Method and system for speech data compression
EP0768770B1 (en) Method and arrangement for the creation of comfort noise in a digital transmission system
EP2113913B1 (en) Method and system for reconstituting low frequencies in an audio signal
EP0064119B1 (en) Speech coding methods and apparatus for carrying out the method
EP0093219B1 (en) Digital coding method and device for carrying out the method
EP2419900B1 (en) Method and device for the objective evaluation of the voice quality of a speech signal taking into account the classification of the background noise contained in the signal
EP2002428B1 (en) Method for trained discrimination and attenuation of echoes of a digital signal in a decoder and corresponding device
BE897704A (en) ANALOG AND DIGITAL SIGNAL APPARATUS
FR2526608A1 (en) APPARATUS FOR TRANSMITTING ANALOG AND DIGITAL SIGNALS
FR2596936A1 (en) VOICE SIGNAL TRANSMISSION SYSTEM
EP0994464A1 (en) Method and apparatus for generating a wide-band signal from a narrow-band signal and telephone equipment comprising such an apparatus
FR2741217A1 (en) METHOD AND DEVICE FOR REMOVING PARASITE NOISE IN A COMMUNICATION SYSTEM
EP0085820B1 (en) Method for multi-speed digital transmission and apparatus for carrying out said method
EP0078581B1 (en) Differential pcm transmission system with adaptive prediction
EP0124411B1 (en) Channel vocoder comprising means for suppressing parasitic modulation of the synthesized speech signal
FR2722631A1 (en) METHOD AND SYSTEM FOR ADAPTIVE FILTERING BY BLIND EQUALIZATION OF A DIGITAL TELEPHONE SIGNAL AND THEIR APPLICATIONS
EP0334714A1 (en) Differential coder with a self-adaptive filter, and decoder for said coder
FR2793589A1 (en) Acoustic speech signal phase information processing having frequency periodic signals critical bandwidth calculator/local phase frequency plane change/phase significant discriminator producing digital phase outputs.
EP0989544A1 (en) Device and method for filtering a speech signal, receiver and telephone communications system
EP0084997A1 (en) Modulation device for a single sideband type modulation chain
WO1981002658A1 (en) Automatic corrector for the frequency transmission characteristics of an electro acoustic channel
EP0812070B1 (en) Method and device for compression encoding of a digital signal
EP0686964B1 (en) Method and arrangement to generate an analog synthesized signal
FR2707443A1 (en) Method of underwater acoustic transmission and device for improving the intelligibility of such transmissions.

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: 19850502

17Q First examination report despatched

Effective date: 19870128

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

GBT Gb: translation of ep patent filed (gb section 77(6)(a)/1977)
REF Corresponds to:

Ref document number: 3467650

Country of ref document: DE

Date of ref document: 19871223

ITF It: translation for a ep patent filed

Owner name: JACOBACCI & PERANI 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: 19960409

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: 19960419

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: 19960529

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: 19970413

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

Ref country code: NL

Effective date: 19971101

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

Effective date: 19970413

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: 19980101

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

Effective date: 19971101