WO2005067150A1 - Method for compressing audio signals without time constraint - Google Patents

Method for compressing audio signals without time constraint Download PDF

Info

Publication number
WO2005067150A1
WO2005067150A1 PCT/EP2004/053397 EP2004053397W WO2005067150A1 WO 2005067150 A1 WO2005067150 A1 WO 2005067150A1 EP 2004053397 W EP2004053397 W EP 2004053397W WO 2005067150 A1 WO2005067150 A1 WO 2005067150A1
Authority
WO
WIPO (PCT)
Prior art keywords
dictionary
message
signal
decompression
generic
Prior art date
Application number
PCT/EP2004/053397
Other languages
French (fr)
Inventor
François CAPMAN
Bertrand Ravera
Original Assignee
Thales
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 Thales filed Critical Thales
Publication of WO2005067150A1 publication Critical patent/WO2005067150A1/en

Links

Classifications

    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
    • H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
    • H03M7/3082—Vector coding

Definitions

  • the invention relates to a method for compressing audio signals without delay constraint. It applies for example for the compression of audio signals in a messaging system.
  • the standard compression scheme for audio signals is based on short-term modeling of this signal. This modeling makes it possible to obtain a set of optimal parameters which will be used during the decompression phase.
  • the time constraint is always associated with this type of model. It does not allow to take into account, during the modeling, the strong correlations which exist between different zones of the audio signal. These areas are generally not located over a time interval shorter than the modeling horizon, the latter being itself always less than the characteristic delay of the compression system.
  • the method according to the invention proposes a new method of compressing audio signals which in particular makes it possible to overcome the time constraint, or in other words, the limitation of the modeling horizon.
  • the invention relates to a compression / decompression method of a signal or message comprising one or more generic parameters, in a compression / decompression system comprising a generic dictionary characterized in that it comprises at least the following steps:
  • the method according to the invention has the following advantages in particular. It does not assume any strong limitation on the type of audio signal or message compression / decompression system in which it is implemented. It can thus be used in conjunction with most 'existing systems that have a with0 dictionary quantization stage of part or all of the model parameters, excluding the prediction quantification methods.
  • Figure 1 an example of a generation diagram of the adapted original dictionary to an audio signal
  • Figure 2 the diagram of the dictionary adapted to the audio signal
  • Figure 3 an associated compression / decompression diagram.
  • the example given below concerns the compression / decompression of an audio signal.
  • the compression / decompression system notably has a set of generic parameters, for example short-term envelope filters, pitch, energy, etc. and a generic dictionary which is, in this case, the basic coder of the compression / decompression system.
  • FIG. 1 shows an example of steps prior to the execution of the steps of the method according to the invention. They consist, for example, in coding (step 1) the audio signal by the generic coder of the compression system, in order to obtain the parameters representative of the signal (step 2). These parameters are then transmitted, for example, to a segmentation step known to those skilled in the art.
  • the implementation of the method according to the invention comprises, in relation to FIGS. 1, 2 and 3, at least the following steps:
  • the following step 5 consists, for example, in representing the original dictionary adapted to the message in the space of the generic parameters of the compression / decompression system.
  • the method creates an indexing table 6 between the original dictionary adapted to the message 5 and the generic dictionary 7 or DG.
  • This indexing table makes it possible to obtain the dictionary qualified as “dictionary adapted to the message” 8.
  • the vectors s (i) d ⁇ co_quant of the dictionary adapted to the message 8 are obtained by searching for the best representatives of the vectors s (o) of ⁇ co among the vectors s (j) C smell of generic dictionary 7. to do this, a distortion criterion may be used.
  • the method has the dictionary adapted to the message which it then uses to quantify the parameters of the audio signal as shown in FIG. 3. 3.
  • Implementation of a specific compression scheme i The following step 6 consists then quantifying the parameters of the audio signal with the dictionary adapted to the message 8 in order to obtain a set of indexes 9 (or indices) on the dictionary adapted to the message.
  • the indexes on the dictionary adapted to the signal are transmitted to the decompression part of the system.
  • the indexing table or index table is transmitted to the decompression device before the indexes relating to the parameters (FIG. 3).
  • the dictionary adapted to the message is obtained from the index table itself resulting from the generic dictionary and from the original dictionary adapted to the message. For the decompression part of the system, it is from the index table and the generic dictionary that the dictionary adapted to the message is reconstructed.
  • the first step of the process is to transform the indexes of the audio signal on the dictionary adapted to the message into index 10 on the generic dictionary of the system. This is achieved, for example, by reverse quantization using the dictionary adapted to the message reconstructed from the transmitted indexing table and the generic dictionary. These indexes on the generic dictionary are then used to produce generic parameters 1 1 necessary for the decompression system, for example by using a quantification method known to those skilled in the art. These generic parameters are then used to obtain the original audio signal.
  • the example which follows illustrates the implementation of the method for a signal resulting from a recording of a radio program broadcasting news.
  • the typology of the audio signal is not unique. We find undegraded speech (taken from his studio), degraded speech (telephone interview or disturbing sound environment), music (jingle) and a mixture of speech and music (sound marker).
  • the generic compression / decompression system is a proprietary system with a throughput of 3600 bits / s.
  • the audio test material consists of 15 minutes of recording.
  • the audio signal is sampled at 16 KHz.
  • the generic parameters retained for this example of implementation are the LSF (English abbreviation of Une Spectral Frequency).
  • the coding rate for these generic parameters is 2666bits / s (60 bits per 22.5 ms frame).
  • the original dictionary adapted to the message is obtained by applying vector quantization processing to LSF vectors calculated every 22.5 ms and has 1024 vectors (for 10-bit quantization).
  • the dictionary adapted to the message is constructed by looking for the best representatives of the original vectors adapted to the audio signal in the generic dictionary.
  • the indexing table is determined and known to the compression system.
  • the LSF parameters are then coded with the dictionary adapted to the message.
  • the method transmits to the decompression system, the indexing table and then the indexes on the dictionary adapted to the message (included in this case of application between 0 and 1023).
  • the other parameters of the audio signal are calculated according to the method specific to the generic system.
  • the decompression system receives the indexing table and the indexes on the dictionary adapted to the message. By inverse quantization, the indexes on the dictionary adapted to the message are transmitted as an index on the generic dictionary of the compression / decompression system.
  • the method according to the invention also applies to MMS message type signals, etc.

Abstract

The invention relates to a method for compressing/decompressing a signal or message comprising at least one generic parameter, in a compression/decompression system comprising a generic dictionary. Said method consists of the following steps: an original dictionary adapted to the message is determined from the totality or quasi-totality of the signal; an indexing table is created between the original dictionary adapted to the message and the generic dictionary of the compression/decompression system in order to obtain the dictionary adapted to the message; the parameters of the signal are quantified with the dictionary adapted to the message in order to obtain the indexes of the signal on the dictionary adapted to the message; the indexing table is transmitted to a decompression stage and the indexes of the signal are then transmitted to the dictionary adapted to the message; and the parameters of the signal are determined from the dictionary adapted to the reconstituted message and indexes of the signal transmitted to the decompression part. The invention can be applied to the compression/decompression of an audio signal.

Description

PROCEDE DE COMPRESSION DE SIGNAUX AUDIO SANS CONTRAINTE DE DELAI METHOD FOR COMPRESSING AUDIO SIGNALS WITHOUT DELAY CONSTRAINTS
L'invention concerne un procédé de compression de signaux audio sans contrainte de délai. Elle s'applique par exemple pour la compression de signaux audio dans un système de messagerie.The invention relates to a method for compressing audio signals without delay constraint. It applies for example for the compression of audio signals in a messaging system.
Le schéma de compression standard de signaux audio est basé sur la modélisation à court terme de ce signal. Cette modélisation permet d'obtenir un ensemble de paramètres optimaux qui seront utilisés durant la phase de décompression. La contrainte de délai est toujours associée à ce type de modèle. Elle ne permet pas de tenir compte, durant la modélisation, des fortes corrélations qui existent entre différentes zones du signal audio. Ces zones ne sont généralement pas localisées sur un intervalle temporel inférieur à l'horizon de modélisation, ce dernier étant quant à lui toujours inférieur au délai caractéristique du système de compression. Le procédé selon l'invention propose une nouvelle méthode de compression des signaux audio qui permet notamment de s'affranchir de la contrainte de délai, ou autrement dit, de la limitation de l'horizon de modélisation.The standard compression scheme for audio signals is based on short-term modeling of this signal. This modeling makes it possible to obtain a set of optimal parameters which will be used during the decompression phase. The time constraint is always associated with this type of model. It does not allow to take into account, during the modeling, the strong correlations which exist between different zones of the audio signal. These areas are generally not located over a time interval shorter than the modeling horizon, the latter being itself always less than the characteristic delay of the compression system. The method according to the invention proposes a new method of compressing audio signals which in particular makes it possible to overcome the time constraint, or in other words, the limitation of the modeling horizon.
L'invention concerne un procédé de compression/décompression d'un signal ou message comprenant un ou plusieurs paramètres génériques, dans un système de compression/décompression comprenant un dictionnaire générique caractérisé en ce qu'il comporte au moins les étapes suivantes :The invention relates to a compression / decompression method of a signal or message comprising one or more generic parameters, in a compression / decompression system comprising a generic dictionary characterized in that it comprises at least the following steps:
• Déterminer un dictionnaire original adapté au message, DOAM, à partir de l'ensemble du signal ou de sa quasi-totalité, • Créer une table d'indexation entre le DOAM et le dictionnaire générique DG, afin d'obtenir le dictionnaire adapté au message DAM, • Quantifier les paramètres du signal avec le dictionnaire adapté au message, afin d'obtenir les indices du signal sur le dictionnaire adapté au• Determine an original dictionary adapted to the message, DOAM, from the whole signal or almost all of it, • Create an indexing table between the DOAM and the generic dictionary DG, in order to obtain the dictionary adapted to the DAM message, • Quantify the parameters of the signal with the dictionary adapted to the message, in order to obtain the indices of the signal on the dictionary adapted to
5 message, • Transmettre la table d'indexation à une étape de décompression et après transmettre les indices du signal sur le dictionnaire adapté au message, • Déterminer le dictionnaire adapté au signal à partir du dictionnaire générique et de la table d'indexation pour l'étape de décompression et0 utiliser ce dictionnaire DAM pour quantifier les indices du signal afin d'obtenir les index du signal sur le dictionnaire générique, • Déterminer les paramètres du signal à partir des index sur le dictionnaire générique. 5 Le procédé selon l'invention présente notamment les avantages suivants. Il ne suppose pas de limitation forte sur le type de système de compression/décompression de signaux audio ou de message dans lequel il est mis en œuvre. Il est ainsi utilisable conjointement à la plupart des ' ' systèmes existants qui possèdent un étage de quantification avec0 dictionnaire d'une partie ou de la totalité des paramètres du modèle, en excluant les méthodes de quantification par prédiction. Ce ou ces dictionnaires sont qualifiés dans la description de « dictionnaires génériques ». 5 D'autres caractéristiques et avantages du procédé selon l'invention apparaîtront mieux à la lecture de l'exemple donné à titre indicatif et nullement limitatif, annexé des figures qui représentent : • La figure 1 un exemple de schéma de génération du dictionnaire original adapté à un signal audio,0 • La figure 2 le schéma du dictionnaire adapté au signal audio, • La figure 3 un schéma de compression/décompression associé. L'exemple donné ci-après concerne la compression/décompression d'un signal audio. Le système de compression/décompression possède notamment un ensemble de paramètres génériques, par exemple les filtres d'enveloppe court terme, le pitch, l'énergie, etc. et un dictionnaire générique qui est, dans ce cas, le codeur de base de référence du système de compression/décompression. L'idée mise en œuvre dans le procédé selon l'invention, consiste notamment, à transmettre de la partie compression du système vers la partie décompression du système, un premier flux qui comprend l'information nécessaire pour coder et décoder le signal audio et un deuxième flux qui comprend les paramètres du signal audio. La figure 1 montre un exemple d'étapes préalables à l'exécution des étapes du procédé selon l'invention. Elles consistent, par exemple, à coder (étape 1 ) le signal audio par le codeur générique du système de compression, afin d'obtenir les paramètres représentatifs du signal (étape 2). Ces paramètres sont ensuite transmis, par exemple, à une étape de segmentation connue de l'Homme du métier. La mise en œuvre du procédé selon l'invention comporte, en relation aux figures 1 , 2 et 3, au moins les étapes suivantes:5 message, • Transmit the indexing table to a decompression step and after transmitting the signal indices on the dictionary suitable for the message, • Determine the dictionary suitable for the signal from the generic dictionary and the indexing table for the decompression step and to use this DAM dictionary to quantify the signal indices in order to obtain the signal indexes on the generic dictionary, • Determine the signal parameters from the indexes on the generic dictionary. The method according to the invention has the following advantages in particular. It does not assume any strong limitation on the type of audio signal or message compression / decompression system in which it is implemented. It can thus be used in conjunction with most 'existing systems that have a with0 dictionary quantization stage of part or all of the model parameters, excluding the prediction quantification methods. This or these dictionaries are qualified in the description of "generic dictionaries". 5 Other characteristics and advantages of the method according to the invention will appear better on reading the example given by way of indication and in no way limiting, appended to the figures which represent: • Figure 1 an example of a generation diagram of the adapted original dictionary to an audio signal, 0 • Figure 2 the diagram of the dictionary adapted to the audio signal, • Figure 3 an associated compression / decompression diagram. The example given below concerns the compression / decompression of an audio signal. The compression / decompression system notably has a set of generic parameters, for example short-term envelope filters, pitch, energy, etc. and a generic dictionary which is, in this case, the basic coder of the compression / decompression system. The idea implemented in the method according to the invention, consists in particular, in transmitting from the compression part of the system to the decompression part of the system, a first stream which includes the information necessary for coding and decoding the audio signal and a second stream which includes the parameters of the audio signal. FIG. 1 shows an example of steps prior to the execution of the steps of the method according to the invention. They consist, for example, in coding (step 1) the audio signal by the generic coder of the compression system, in order to obtain the parameters representative of the signal (step 2). These parameters are then transmitted, for example, to a segmentation step known to those skilled in the art. The implementation of the method according to the invention comprises, in relation to FIGS. 1, 2 and 3, at least the following steps:
1. Création du dictionnaire original adapté au message. Partie compression du système (figure 1 ). A partir du signal audio, déterminer les paramètres du modèle à court terme, en tenant compte de l'ensemble ou de la quasi-totalité du signal audio. Cette étape 3 est réalisée, par exemple, par quantification des paramètres, en utilisant une méthode classique connue de l'Homme du métier. Le procédé dispose alors d'un ensemble de paramètres représentatifs du signal audio. La corrélation à long terme de l'ensemble des paramètres permet de créer un dictionnaire de paramètres (étape 4) représentatifs du signal audio. Ce dictionnaire de taille réduite est désigné par l'expression « dictionnaire original adapté au message » ou DOAM. 2. Obtention du dictionnaire adapté au message (figure 2). L'étape 5 suivante consiste, par exemple, à représenter le dictionnaire original adapté au message dans l'espace des paramètres génériques du système de compression/décompression. Pour cela, le procédé crée une table d'indexation 6 entre le dictionnaire original adapté au message 5 et le dictionnaire générique 7 ou DG. Cette table d'indexation permet d'obtenir le dictionnaire qualifié de « dictionnaire adapté au message » 8. Dans cet exemple d'application pour un signal audio, les vecteurs s(i)dιco_quant du dictionnaire adapté au message 8 sont obtenus en cherchant les meilleurs représentants des vecteurs s(o)dιco parmi les vecteurs s(j)Codeur du dictionnaire générique 7. Pour ce faire, un critère de distorsion pourra être utilisé. Seuls les index des vecteurs du dictionnaire générique sont sauvegardés dans le dictionnaire adapté au message. Le procédé dispose du dictionnaire adapté au message qu'il utilise alors pour quantifier les paramètres du signal audio tel qu'il est représenté à la figure 3. 3. Mise en place d'un schéma de compression spécifique i L'étape 6 suivante consiste alors à quantifier les paramètres du signal audio avec le dictionnaire adapté au message 8 afin d'obtenir un ensemble d'index 9 (ou indices) sur le dictionnaire adapté au message. Les index sur le dictionnaire adapté au signal sont transmis à la partie décompression du système. Afin de pouvoir appliquer l'opération inverse lors de l'étape de décompression, la table d'indexation ou table d'index est transmise au dispositif de décompression avant les index relatifs aux paramètres (figure 3). Au niveau de la partie compression du système, le dictionnaire adapté au message est obtenu à partir de la table d'index elle-même résultant du dictionnaire générique et du dictionnaire original adapté au message. Pour la partie décompression du système, c'est à partir de la table d'index et du dictionnaire générique que l'on reconstitue le dictionnaire adapté au message.1. Creation of the original dictionary adapted to the message. Compression part of the system (figure 1). From the audio signal, determine the parameters of the short-term model, taking into account all or almost all of the audio signal. This step 3 is carried out, for example, by quantifying the parameters, using a conventional method known to those skilled in the art. The method then has a set of parameters representative of the audio signal. The long-term correlation of all the parameters makes it possible to create a dictionary of parameters (step 4) representative of the audio signal. This reduced-size dictionary is designated by the expression “original dictionary adapted to the message” or DOAM. 2. Obtaining the dictionary adapted to the message (Figure 2). The following step 5 consists, for example, in representing the original dictionary adapted to the message in the space of the generic parameters of the compression / decompression system. For this, the method creates an indexing table 6 between the original dictionary adapted to the message 5 and the generic dictionary 7 or DG. This indexing table makes it possible to obtain the dictionary qualified as “dictionary adapted to the message” 8. In this example of application for an audio signal, the vectors s (i) dιco_quant of the dictionary adapted to the message 8 are obtained by searching for the best representatives of the vectors s (o) of ιco among the vectors s (j) C smell of generic dictionary 7. to do this, a distortion criterion may be used. Only the vector indexes of the generic dictionary are saved in the dictionary suitable for the message. The method has the dictionary adapted to the message which it then uses to quantify the parameters of the audio signal as shown in FIG. 3. 3. Implementation of a specific compression scheme i The following step 6 consists then quantifying the parameters of the audio signal with the dictionary adapted to the message 8 in order to obtain a set of indexes 9 (or indices) on the dictionary adapted to the message. The indexes on the dictionary adapted to the signal are transmitted to the decompression part of the system. In order to be able to apply the reverse operation during the decompression step, the indexing table or index table is transmitted to the decompression device before the indexes relating to the parameters (FIG. 3). At the compression part of the system, the dictionary adapted to the message is obtained from the index table itself resulting from the generic dictionary and from the original dictionary adapted to the message. For the decompression part of the system, it is from the index table and the generic dictionary that the dictionary adapted to the message is reconstructed.
4. Mise en place d'un schéma de décompression spécifique Au niveau de la décompression, la première étape du procédé est de transformer les index du signal audio sur le dictionnaire adapté au message en index 10 sur le dictionnaire générique du système. Ceci est réalisé, par exemple, par quantification inverse en utilisant le dictionnaire adapté au message reconstitué à partir de la table d'indexation transmise et du dictionnaire générique. Ces index sur le dictionnaire générique sont ensuite utilisés pour produire des paramètres génériques 1 1 nécessaires au système de décompression, par exemple en utilisant une méthode de quantification connue de l'Homme du métier. Ces paramètres génériques servent ensuite pour obtenir le signal audio d'origine.4. Implementation of a specific decompression scheme At the decompression level, the first step of the process is to transform the indexes of the audio signal on the dictionary adapted to the message into index 10 on the generic dictionary of the system. This is achieved, for example, by reverse quantization using the dictionary adapted to the message reconstructed from the transmitted indexing table and the generic dictionary. These indexes on the generic dictionary are then used to produce generic parameters 1 1 necessary for the decompression system, for example by using a quantification method known to those skilled in the art. These generic parameters are then used to obtain the original audio signal.
L'exemple qui suit illustre la mise en œuvre du procédé pour un signal issu d'un enregistrement d'un programme radio diffusant des nouvelles. La typologie du signal audio n'est pas unique. On y retrouve de la parole non dégradée (prise de son studio), de la parole dégradée ( interview téléphonique ou ambiance sonore perturbatrice), de la musique (jingle) et un mélange de parole et de musique (marqueur sonore). Le système de compression/décompression générique est un système propriétaire possédant un débit de 3600 bits/s. Le matériel audio de test est constitué de 15 minutes d'enregistrement. Le signal audio est échantillonné à 16 KHz. Les paramètres génériques retenus pour cet exemple de mise en œuvre, sont les LSF (abréviation anglo-saxonne de Une Spectral Frequency). Le débit de codage de ces paramètres génériques est de 2666bits/s (60 bits par trame de 22,5 ms). Le dictionnaire original adapté au message est obtenu en appliquant un traitement de quantification vectorielle aux vecteurs de LSF calculés tous les 22,5 ms et possède 1024 vecteurs ( pour une quantification sur 10 bits). Le dictionnaire adapté au message est construit en cherchant les meilleurs représentants des vecteurs originaux adaptés au signal audio dans le dictionnaire générique. A la fin de cette étape, la table d'indexation est déterminée et connue du système de compression. Les paramètres LSF sont ensuite codés avec le dictionnaire adapté au message. Le procédé transmet au système de décompression, la table d'indexation et ensuite les index sur le dictionnaire adapté au message (compris dans ce cas d'application entre 0 et 1023). Les autres paramètres du signal audio sont calculés selon la méthode propre au système générique. Le système de décompression reçoit la table d'indexation et les index sur le dictionnaire adapté au message. Par quantification inverse, les index sur le dictionnaire adapté au message sont transmis en index sur le dictionnaire générique du système de compression/décompression.The example which follows illustrates the implementation of the method for a signal resulting from a recording of a radio program broadcasting news. The typology of the audio signal is not unique. We find undegraded speech (taken from his studio), degraded speech (telephone interview or disturbing sound environment), music (jingle) and a mixture of speech and music (sound marker). The generic compression / decompression system is a proprietary system with a throughput of 3600 bits / s. The audio test material consists of 15 minutes of recording. The audio signal is sampled at 16 KHz. The generic parameters retained for this example of implementation are the LSF (English abbreviation of Une Spectral Frequency). The coding rate for these generic parameters is 2666bits / s (60 bits per 22.5 ms frame). The original dictionary adapted to the message is obtained by applying vector quantization processing to LSF vectors calculated every 22.5 ms and has 1024 vectors (for 10-bit quantization). The dictionary adapted to the message is constructed by looking for the best representatives of the original vectors adapted to the audio signal in the generic dictionary. At the end of this step, the indexing table is determined and known to the compression system. The LSF parameters are then coded with the dictionary adapted to the message. The method transmits to the decompression system, the indexing table and then the indexes on the dictionary adapted to the message (included in this case of application between 0 and 1023). The other parameters of the audio signal are calculated according to the method specific to the generic system. The decompression system receives the indexing table and the indexes on the dictionary adapted to the message. By inverse quantization, the indexes on the dictionary adapted to the message are transmitted as an index on the generic dictionary of the compression / decompression system.
,•4. Le procédé selon l'invention s'applique aussi pour des signaux de type message MMS, etc , • 4. The method according to the invention also applies to MMS message type signals, etc.

Claims

REVENDICATIONS
1- Procédé de compression/décompression d'un signal audio comprenant un ou plusieurs paramètres génériques, dans un système de compression/décompression comprenant un dictionnaire générique caractérisé en ce qu'il comporte au moins les étapes suivantes : • Déterminer un dictionnaire original adapté au message (5) à partir de l'ensemble ou de la quasi-totalité du signal,1- A method of compressing / decompressing an audio signal comprising one or more generic parameters, in a compression / decompression system comprising a generic dictionary characterized in that it comprises at least the following steps: • Determining an original dictionary adapted to the message (5) from all or almost all of the signal,
• Créer une table d'indexation entre le dictionnaire original adapté au message (5) et le dictionnaire générique (7) du système de compression/décompression, afin d'obtenir le dictionnaire adapté au message (8),• Create an indexing table between the original dictionary adapted to the message (5) and the generic dictionary (7) of the compression / decompression system, in order to obtain the dictionary adapted to the message (8),
• Quantifier les paramètres du signal avec le dictionnaire adapté au message (8), afin d'obtenir les indices (9) du signal sur le dictionnaire adapté au message,• Quantify the signal parameters with the dictionary adapted to the message (8), in order to obtain the indices (9) of the signal on the dictionary adapted to the message,
• Transmettre la table d'indexatio'ri (6) à une étape de décompression et ensuite transmettre les indices (9) du signal sur le dictionnaire adapté au message,• Transmit the index table (6) to a decompression step and then transmit the indices (9) of the signal on the dictionary adapted to the message,
• Déterminer les paramètres du signal à partir du dictionnaire adapté au message reconstitué et des index du signal transmis à la partie décompression.• Determine the parameters of the signal from the dictionary adapted to the reconstructed message and the indexes of the signal transmitted to the decompression part.
2 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce que la détermination des paramètres au niveau de la décompression comporte au moins les étapes suivantes :2 - Compression / decompression method according to claim 1 characterized in that the determination of the parameters at the decompression level comprises at least the following steps:
• Déterminer le dictionnaire adapté au message (5) à partir du dictionnaire générique (7) et de la table d'indexation (6) pour l'étape de décompression et utiliser ce dictionnaire adapté au message pour quantifier les index du signal afin d'obtenir les index du signal sur le dictionnaire générique, • Déterminer les paramètres du signal à partir des index sur le dictionnaire générique.• Determine the dictionary suitable for the message (5) from the generic dictionary (7) and the indexing table (6) for the decompression step and use this dictionary adapted to the message for quantify the signal indexes in order to obtain the signal indexes on the generic dictionary, • Determine the signal parameters from the indexes on the generic dictionary.
3 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce qu'il utilise un signal de parole où les paramètres génériques sont : les filtres d'enveloppe à court terme, le pitch, l'énergie, etc.3 - A compression / decompression method according to claim 1 characterized in that it uses a speech signal where the generic parameters are: short-term envelope filters, pitch, energy, etc.
4 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce que les paramètres génériques sont les LSF et l'étape de quantification est effectuée par traitement de quantification vectorielle des vecteurs LSF.4 - Method of compression / decompression according to claim 1 characterized in that the generic parameters are the LSF and the quantification step is carried out by vector quantization processing of the LSF vectors.
5 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce que le dictionnaire adapté au message (5) est construit en cherchant les meilleurs représentants des vecteurs originaux adaptés au signal audio dans le dictionnaire générique.5 - Method of compression / decompression according to claim 1 characterized in that the dictionary adapted to the message (5) is constructed by seeking the best representatives of the original vectors adapted to the audio signal in the generic dictionary.
6 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce que le message est un message de type MMS.6 - A compression / decompression method according to claim 1 characterized in that the message is an MMS type message.
7 - Procédé de compression/décompression selon la revendication 1 caractérisé en ce que la table d'indexation (6) et les indices du signal (9) sont transmis via un protocole de type MMS (Multimedia Messaging Service). 7 - A compression / decompression method according to claim 1 characterized in that the indexing table (6) and the signal indices (9) are transmitted via an MMS (Multimedia Messaging Service) type protocol.
PCT/EP2004/053397 2003-12-12 2004-12-10 Method for compressing audio signals without time constraint WO2005067150A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR0314599 2003-12-12
FR0314599A FR2863792B1 (en) 2003-12-12 2003-12-12 METHOD FOR COMPRESSING AUDIO SIGNALS WITHOUT TIME LIMIT

Publications (1)

Publication Number Publication Date
WO2005067150A1 true WO2005067150A1 (en) 2005-07-21

Family

ID=34610617

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2004/053397 WO2005067150A1 (en) 2003-12-12 2004-12-10 Method for compressing audio signals without time constraint

Country Status (2)

Country Link
FR (1) FR2863792B1 (en)
WO (1) WO2005067150A1 (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4975956A (en) * 1989-07-26 1990-12-04 Itt Corporation Low-bit-rate speech coder using LPC data reduction processing

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4975956A (en) * 1989-07-26 1990-12-04 Itt Corporation Low-bit-rate speech coder using LPC data reduction processing

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
GERSHO ET AL: "Vector Quantization and Signal Compression", 1991, KLUWER, BOSTON,US, XP002291034 *
TOMOYUKI OHYA ET AL: "REVISED TC-WVQ SPEECH CODER FOR MOBILE COMMUNICATION SYSTEM", PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON SPOKEN LANGUAGE PROCESSING (ICSLP). KOBE, NOV. 18 - 22, 1990, TOKYO, ASJ, JP, vol. VOL. 1, 18 November 1990 (1990-11-18), pages 125 - 128, XP000503330 *
WAI C. CHU: "Speech Coding Algorithms, Foundation and Evolution of Standardized Coders", April 2003, XP002291035 *

Also Published As

Publication number Publication date
FR2863792A1 (en) 2005-06-17
FR2863792B1 (en) 2006-05-12

Similar Documents

Publication Publication Date Title
US5479559A (en) Excitation synchronous time encoding vocoder and method
EP0768780B1 (en) Method and apparatus for wavelet based data compression having adaptive bit rate control for compression of digital audio or other sensory data
US8509931B2 (en) Progressive encoding of audio
EP1581930B1 (en) Method for encoding and decoding audio at a variable rate
US6732075B1 (en) Sound synthesizing apparatus and method, telephone apparatus, and program service medium
JP2005189886A (en) Method for improving coding efficiency of audio signal
JP5264901B2 (en) Hierarchical coding of digital audio signals
FR2929466A1 (en) DISSIMULATION OF TRANSMISSION ERROR IN A DIGITAL SIGNAL IN A HIERARCHICAL DECODING STRUCTURE
EP2183851A1 (en) Encoding/decoding by symbol planes with dynamic calculation of probability tables
US5504834A (en) Pitch epoch synchronous linear predictive coding vocoder and method
US7801732B2 (en) Audio codec system and audio signal encoding method using the same
US7508333B2 (en) Method and apparatus to quantize and dequantize input signal, and method and apparatus to encode and decode input signal
EP2795618B1 (en) Method of detecting a predetermined frequency band in an audio data signal, detection device and computer program corresponding thereto
JP2013084002A (en) Device and method for enhancing quality of speech codec
KR100972349B1 (en) System and method for determinig the pitch lag in an LTP encoding system
CA2300647A1 (en) Audio coding, decoding and transcoding procedures
FR2884989A1 (en) Digital multimedia signal e.g. voice signal, coding method, involves dynamically performing interpolation of linear predictive coding coefficients by selecting interpolation factor according to stationarity criteria
JP2006171751A (en) Speech coding apparatus and method therefor
WO2005067150A1 (en) Method for compressing audio signals without time constraint
EP2203915A1 (en) Transmission error dissimulation in a digital signal with complexity distribution
JP2024502287A (en) Speech enhancement method, speech enhancement device, electronic device, and computer program
EP2126904B1 (en) Audio encoding method and device
EP1756806A1 (en) Method for quantifying an ultra low-rate speech encoder
US20240153520A1 (en) Neutralizing distortion in audio data
Leslie et al. Joint source and channel coding for internet audio transmission

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase