Data processing: speech signal processing – linguistics – language – Speech signal processing – Psychoacoustic
Reexamination Certificate
2000-06-16
2003-12-02
Dorvil, Richemond (Department: 2654)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Psychoacoustic
C704S226000, C704S229000, C704S214000
Reexamination Certificate
active
06658378
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method and apparatus for decoding encoding parameters selectively protected against errors produced in a transmission path, and a program furnishing medium.
2. Description of Related Art
There are known a variety of encoding methods for doing signal compression by exploiting statistic properties in the time domain and in the frequency domain of audio signals, including the speech and acoustic signals, and psychoacoustic properties of the human being. As these encoding methods, VSELP (Vector Sum Excited Linear Prediction) encoding system or the PSI-CELP (Pitch Synchronous Innovation—CELP), classified under the so-called CELP (Code Excited Linear Prediction) encoding system, is attracting attention as being a low bit rate speech encoding system.
In the waveform encoding system, such as this CELP encoding system, input speech signals are formed into blocks or frames, with a pre-set number of samples of the input speech signals as an encoding unit, and a closed loop search of an optimum vector is performed using an analysis by synthesis method on the block- or frame-based time-domain speech waveform to execute vector quantization of the waveform to output the vector index.
Meanwhile, the code bits, obtained with the speech encoding system with a low bitrate, such as 2 kbps or 4 kbps, are widely used in communication, computer or broadcasting as general audio not constrained by specified audio information contents. Therefore, these code bits need to be protected strongly against errors produced on the transmission path.
If an error is produced in succession on the transmission path, sound dropout occurs in succession for long time in speech decoding, thus lowering the speech quality.
The present inventors have clarified, in Japanese Laying-Open Patent H-11-122120, an encoding method and apparatus and a decoding method and apparatus which is strong against errors occurring on the transmission path to enable the sound quality to be improved appreciably.
In this encoding method and apparatus, the input speech signal is divided on the time axis every pre-set encoding unit and encoded from one encoding unit to another to output plural sorts of the encoding speech parameters. Of these plural sorts of the encoding speech parameters, a psychoacoustically crucial bit group is selected and, from this crucial bit group, an error check code is generated. This error check code and the crucial bit group are subjected to convolution coding to enable protection of the psychoacoustically crucial bit groups from errors occurring on the transmission path.
For decoding the transmitted encoded data, produced on convolution encoding of the error check code and the crucial bit group and on merging to the bit group excluding the crucial bit group, the decoding method and apparatus process the convolution coded output with coevolution decoding, convolution decode the crucial bit group, added to with the error check code, and the bit group excluding the crucial bit group, output the convolution decoded output, check the transmission error using the error check code appended to the convolution decoded output, adjust the convolution decoded output depending on the result of the error check and process the adjusted convolution decoded output with speech decoding, thus enabling decoding of the speech such as to suppress the lowering of the sound quality due to errors occurring on transmission paths.
Meanwhile, in the above-described decoding method and apparatus, it is necessary to maintain signal continuity as speech signals following the speech decoding when adjusting the decoded output responsive to the results of error detection as to transmission errors using the error check code appended to the convolution decoded output.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a decoding method and apparatus and a program furnishing medium whereby signal continuity as speech signals signal may be maintained to enable decoding of the high quality speech.
In one aspect, the present invention provides a decoding apparatus for decoding transmitted encoded parameters corresponding to input speech signals which have been encoded by an encoding device from one pre-set encoding unit on the time axis to another and which have been classed depending on the psychoacoustic sensitivity of bits to transmission path errors, wherein the decoding apparatus includes transmission path decoding means for detecting an error using an error check code appended to the encoding parameters of a pre-set class and providing for different frame masking processing depending on errors in the encoding parameters produced in the encoding unit.
Preferably, the encoding processing in the encoding device is performed using a plurality of vector quantizers, an index representing each code vector is classified according to psychoacoustic sensitivity of bits against transmission path errors and is transmitted in this classified state. The transmission path decoding means detects the error check codes appended to the index of a pre-set class to provide for different frame masking processing depending on an index error produced in the encoding unit.
Preferably, the encoding processing in the encoding device is performed using a multi-stage vector quantizer prepared by alternate learning, and an index representing each code vector is classified according to psychoacoustic sensitivity of bits against transmission path errors and is transmitted in this classified state. The transmission path decoding means detects the error check codes appended to the index of a pre-set class to provide for different frame masking processing depending on whether or not an error has been detected in plural indexes.
In another aspect, the present invention provides a decoding method for decoding transmitted encoded parameters corresponding to input speech signals which have been encoded by an encoding device from one pre-set encoding unit on the time axis to another and which have been classed depending on the psychoacoustic sensitivity of bits to transmission path errors, wherein the decoding method includes a transmission path decoding step of detecting an error using an error check code appended to the encoding parameters of a pre-set class and providing for different frame masking processing depending on errors in the encoding parameters produced in the encoding unit.
Preferably, the encoding processing in the encoding device is performed using a plurality of vector quantizers, an index representing each code vector is classified according to psychoacoustic sensitivity of bits against transmission path errors and is transmitted in this classified state. The transmission path decoding step detects the error check codes appended to the index of a pre-set class to provide for different frame masking processing depending on an index error produced in the encoding unit.
Preferably, the encoding processing in the encoding device is performed using a multi-stage vector quantizer prepared by alternate learning, and an index representing each code vector is classified according to psychoacoustic sensitivity of bits against transmission path errors and is transmitted in this classified state. The transmission path decoding means detects the error check codes appended to the index of a pre-set class to provide for different frame masking processing depending on whether or not an error has been detected in plural indexes.
In still another aspect, the present invention provides a program furnishing medium for furnishing a program for decoding transmitted encoded parameters corresponding to input speech signals which have been encoded by an encoding device from one pre-set encoding unit on the time axis to another and which have been classed depending on the psychoacoustic sensitivity of bits to transmission path errors, wherein the program includes a transmission path decoding step of detecting an error using an error check code appended to the encoding parameters
Azad Abul K.
Dorvil Richemond
Maioli Jay H.
Sony Corporation
LandOfFree
Decoding method and apparatus and program furnishing medium does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Decoding method and apparatus and program furnishing medium, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Decoding method and apparatus and program furnishing medium will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3113253