Classifications

• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; 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
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
  • G10L19/06—Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients

Definitions

• the present invention relates to a method and a device for spectral reconstruction of a multi-channel audio signal, in particular of a stereophonic signal.
• the invention also relates to a decoding device comprising this reconstruction device, an associated coding device and a coding / decoding system.
• bit rate we mean the amount of information transmitted per unit of time, generally expressed in kbit / s.
• Known rate reduction coders are for example transform type coders, CELP type coders and even parametric type coders, such as a parametric MPEG4 type coder.
• the audio signal In bit rate reduction audio coding, the audio signal often has to undergo bandwidth limitation when the bit rate becomes low. This limitation of bandwidth is necessary to avoid the introduction of audible quantization noise into the encoded signal. It is therefore desirable to regenerate as far as possible the high frequency content of the original signal.
• This technique is based on an analysis in sub-bands and a complex harmonic duplication. It implements phase and amplitude adjustment methods which are costly in calculation. In addition, the spectral weighting factors only roughly model the spectral envelope.
• the stereophonic content is generally greatly altered. Indeed, if the transmission rate is insufficient, there is a tendency to transmit only signals with low stereophonic content.
• stereo M / S stereo mid-side
• stereo intensity in which, above a certain frequency, a monophonic audio signal is transmitted, generally corresponding to. a weighting of the left and right channels, with gain factors which describe the original energy ratios between the two channels.
• the selective application by frequency bands of different gains for each audio channel makes it possible to recreate an impression of stereophonic signal.
• the stereo part at low frequency may be of poor quality. If the frequency limit beyond which the stereo intensity technique is applied is lowered, the stereophonic content is degraded because the use of gain factors allows only a rough reconstruction of the stereo content. The cost of transmitting the gain factors becomes significant if a more detailed reconstruction of the stereo content is desired.
• the application of different gains by sub-bands tends to create discontinuities.
• the problem underlying the invention is to provide a method and a device for reconstructing a stereo signal, and more generally of a multi-channel audio signal, allowing a reconstruction of the stereophonic content, in particular for the high frequencies, and requiring only a small amount of data to transmit.
• spatial (Vj L ) characterized in that, for at least one component, it comprises:
• the composite signal comprises at least one monophonic component (M) in a first spectral band (Bi), the spectral whitening step provides a whitened monophonic signal and the shaping step uses a filter envelope having the characteristic spectral envelope of the channel to be reconstructed in said first band.
• the composite signal comprises several spatial components (Ni L ), each spatial component being associated with a channel and at least one spatial component having a spectrum limited to a second spectral band, the reconstruction of the channel associated with the limited spectrum space component comprising:
• the second spectral band (B) is a low frequency band and the third spectral band (B 3 ) is adjacent to the second.
• the components of the composite signal come from the decoding of a multi-channel source signal coded by a spectrum limiting coder.
• the characteristic of the envelope filter is obtained from information giving the spectral envelope of the corresponding channel of the source signal in the first and third bands.
• the invention is also defined by a device for reconstructing at least one channel of a multi-channel audio signal, in particular of a stereo signal, from a composite signal which may include monophonic components and spatial components , the device comprising means for implementing the steps of the method defined above.
• the invention is also defined by a device for coding a source audio signal with several channels, at least a first spectral band of said signal being coded in monophonic, the device further providing spectral envelope information for at least one channel. in said first band.
• the invention is also defined by a device for limiting the spectrum of a source audio signal with several channels, the spectrum of at least one channel being limited to a second spectral band by coding, the device further providing a spectral envelope information of said channel in a band distinct from said second band.
• the invention is also defined by a spectrum limitation coding device for a source audio signal with several channels, at least one first spectral band of said signal being coded in monophonic, the spectrum of at least one channel being limited by the coding at a second spectral band, distinct from the first, the device further providing information on the spectral envelope of said channel in the first band and in a third band distinct from said first and second bands.
• the spectral envelope information of the second channel is transmitted in the form of a difference with that of the first channel.
• the invention is also defined by a signal coming from a spectrum limitation coding device as defined above, the signal comprising at least for a first spectral band a coded monophonic component as well as a spectral envelope information. coded relative to the spectral envelope of an audio channel in said first band.
• the invention is also defined by a signal from a coding device with spectrum limitation as defined above, the signal comprising at least for a second spectral band a coded spatial component relating to an audio channel as well as a coded spectral envelope information relating to the spectral envelope of said audio channel in a band distinct from said second band.
• the invention is also defined by a signal originating from a spectrum limitation coding device as defined above, the signal comprising at least, for a first spectral band, a monophonic component and, for a second distinct spectral band of the first, a coded spatial component relating to an audio channel, as well as information of spectral envelope of said channel in the first band and in a third spectral band distinct from said first and second bands.
• the invention is also defined by a device for decoding a coded multi-channel audio signal, the device comprising a decoder adapted to supply from the coded signal a composite signal which may include monophonic components and spatial components, and a device reconstruction as defined above.
• the invention is also defined by a decoding device comprising a first decoder adapted to supply, from said signal defined above, a composite signal capable of comprising monophonic components and spatial components as well as a second decoder adapted to supply, to from said signal, spectral envelope information.
• the invention is further defined by a system for coding / decoding an audio signal with several channels, comprising a coding device and a decoding device as defined above.
• the stereophonic content in particular the high frequency stereophonic content can be reconstituted during decoding without or with a minimum transmission of the information linked to the high frequency content of the original signals.
• the spectral shape of high frequency stereophonic signals can be modeled through two filters, one filter for each channel.
• Envelope information can be transmitted at low cost, because one can easily measure the differences between two envelopes and thereby take advantage of possible redundancies between the spectral shapes modeled. Only one channel can be transmitted, the other channel can be reconstructed by bleaching the transmitted signal and applying an envelope filter.
• the envelope information relating to the non-transmitted channel has a very low transmission cost.
• FIG. . 1 schematically represents a device for reconstructing a stereophonic audio signal according to a first embodiment of the invention
• Figs. 2a to 2c illustrate the processing carried out by the reconstruction device of FIG. 1
• Fig. 3 schematically represents a device for reconstructing a stereophonic audio signal according to a second embodiment of the invention
• Figs. 4a to 4d illustrate the processing carried out by the reconstruction device of FIG. 3
• Fig. 5 schematically represents a device for reconstructing a stereophonic audio signal according to a third embodiment of the invention
• Figs. 6a to 6d illustrate the processing carried out by the reconstruction device of FIG. 5.
• the spectral reconstruction device can be applied to the spectral reconstruction of a stereophonic audio signal resulting from the decoding of a signal coded by an encoder with spectral band limitation. It can be any type of flow reduction encoder.
• the encoder can be of the transform type (MPEG1, MPEG2 or MPEG4-GA), of the CELP type (ITU G72X), or even of the parametric type (parametric MPEG4).
• the invention can also be applied to signals which have not previously been the subject of coding, for example, signals which have simply undergone subsampling and an alteration of their stereophonic content.
• Fig. 1 describes a first embodiment of the invention.
• the signal is coded by an encoder 100 and, after transmission of the coded signal by any means, the coded signal is decoded by a decoder 110.
• the reconstruction of the stereophonic content of an audio signal and more generally of the different channels of a multi-channel audio signal is carried out by the modules 150, 155, 150, 115, 175 and the summers 180 ,. For reasons of simplification, only one channel i has been shown
• the signal decoded by the module 110 has a monophonic component (M) and a spatial component N, L , associated with the channel i to be reconstructed, as illustrated in FIG. 2a.
• the monophonic component can be a component common to several channels, for example a sum of several channels or even the signal of a predominant channel among a set of channels.
• the component N, L will be a low frequency signal with limited band (B 2 ) and the monophonic signal (M) will occupy a band (Bl) adjacent to the first.
• the spectrum of the monophonic part (M) is whitened using a whitening filter 150. It is known that under certain stationary hypotheses, a signal can be modeled as the result of the filtering of an excitation signal by a spectral envelope filter. If a description of the spectral envelope of the signal is available, it is possible to whiten its spectrum by passing it through a whitening filter with transfer function (approximately) inverse to the envelope function. An approximation of the initial excitation signal is thus obtained, free of the influence of the spectral shape in the band considered.
• the module 155 is a spectral envelope estimation module for the monophonic signal in the Bi band. It can for example model the envelopes by an LPC analysis, as described in the article by J. Makhoul, entitled “Linear Prediction: a tutorial review”, Proceedings of the IEEE, Vol. 63, ⁇ ° 4, pp 561-580.
• the spatial component N, and the bleached monophonic component are represented in FIG. 2b.
• the bleached monophonic component is subjected to a spectral envelope step in the envelope filter 170,.
• This envelope filter has for characteristic the spectral envelope of the original channel i in the band Bi.
• spectral estimation means 105 associated with the coder performs a spectral estimation of the different channels in the band Bi. and provide information describing the envelopes of the different channels in this band.
• the envelopes are coded differently. In other words, the envelope of a first channel is coded and those of the other channels are coded by difference, so as to take advantage of the similarity of the envelopes to reduce the redundancy in the information to be transmitted.
• the information relating to the various envelopes is decoded in the module 175.
• the decoded information e (V,) are for example LPC coefficients. They are supplied to the envelope filter 170 ,.
• the spectral envelope of the channel i is obtained as the extrapolated, in the band Bi, of the spectral envelope of the spatial component N, L , in the band B 2 .
• This variant is symbolized in broken lines by the extrapolation module 115, receiving the component V, L and supplying the extrapolated envelope to the envelope filter 170,.
• the spatial component N, L is then added by means of the summator 180, to the monophonic component shaped to provide a reconstructed channel N ,.
• the spectrum of the reconstructed pathway is illustrated in Fig. 2c.
• This monophonic component can correspond to a channel or to a sum of channels, as seen above.
• the different channels are reconstructed thanks to the shaping of the monophonic signal whitened by their respective spectral envelopes.
• Fig. 3 describes a second embodiment of the invention.
• the modules bearing the same references as in FIG. 1 have a function identical to that already described. For reasons of simplification, only the reconstruction of a channel i has been shown.
• the decoder 110 supplies spatial components with limited spectra. This will typically be the case if the encoder 100 is a spectrum limiting encoder.
• a spatial component N of spectrum limited to the band B 2 , as shown in FIG. 4a.
• the module 160 is a spectral transposition module. Its function is to copy the spectral content of at least part of the band B 2 , called the source band, into a second band B 3 , called the target band.
• the transposition operation is for example a simple translation of spectrum in the target band or else the combination of a reversal and a translation.
• band B is a low frequency band and the target band is adjacent to the latter.
• the spectral transposition operation has been illustrated in FIG. 4b.
• the signal obtained at the output of 160 is a signal with spectrum limited to band B. It is subjected to spectral whitening in the whitening filter 150,.
• the characteristic of the whitening filter is the inverse of the spectral envelope of the spatial component transposed in the band B 3 .
• Module 1 15 estimates the coefficients of the whitening filter and supplies them to the latter.
• the coefficients of the filter 150 are obtained from the spectral envelope of the channel i in the source band. It should be noted that the order of the spectral bleaching (150,) and spectral transposition (160,) modules can be reversed. The order chosen depends in particular on the desired whitening precision. The result of spectral bleaching is illustrated in Fig. 4c.
• the spatial component with transposed and whitened spectrum is subjected to a spectral shaping step in the envelope filter 170,.
• the characteristic of this envelope filter is the spectral envelope of the original channel i in band B 3 .
• the information relating to the various envelopes is decoded in module 175.
• the decoded information e (N,) are for example LPC coefficients. They are supplied to the envelope filter 170 ,.
• the spectral envelope of channel i is obtained as the extrapolated, in band B 3 , of the spectral envelope of the spatial component N,, in band B.
• This variant is symbolized by the link in broken lines between the module 115, and the envelope filter 170,.
• the spatial component N, L is then added by means of the summator 181, to the signal from the filter 170, to provide a reconstructed channel N ,.
• the spectrum of the reconstructed pathway is illustrated in Fig. 4d.
• Fig. 5 describes a third embodiment of the invention.
• the modules bearing the same references as those in FIG. 1 or of FIG. 3 have a function identical to that already described. For reasons of simplification, only the reconstruction of a channel i has been shown.
• the decoder 110 supplies a monophonic component M as well as spatial components with limited spectra N,.
• the monophonic component can be common to several or to all channels. Illustrated in FIG. 6 has a spatial component with limited spectrum N, L as well as the monophonic component M.
• the spatial component (for example relating to a channel of a stereo signal) of the signal occupies a low frequency band. In the higher frequencies (spectral band Bl) the signal is coded in monophonic.
• the monophonic component M at the output of the decoder 110, the monophonic component M, on the one hand, and the spatial component V, on the other hand, undergo separate processing.
• the monophonic component is whitened by means of the whitening filter 150 as in FIG. 1.
• the spatial component for its part, is the subject of a transposition of all or part of its spectral content in a target band B 3 , typically contiguous to the monophonic band B 15 as shown in FIG. 6b.
• the source band of the spectral content is included in the band B 2 .
• the source band is included in the band B 2 + B ⁇ , in other words at least part of the spectral content of the monophonic band can also be transposed.
• the signal at the output of 160 has a spectrum limited to the band B 3 . It is bleached in the whitening filter 150, the transfer function of which is determined from the spectral estimation module 115, or, alternatively, by spectral envelope information of the original channel i in the source band given by the module. 175 decoding function.
• the transfer function of the filter 150 is given by the inverse of the spectral envelope in the source band.
• Fig. 6c illustrates the result of the laundering operations in bands B] and B 2 .
• the whitened monophonic and spatial components are combined in the summator 180, and the sum is filtered by the envelope filter 170 ,.
• the transfer function of this filter is given by spectral envelope information of the original channel i, in the band B1. + B 3 , provided by the envelope decoding module 175. As indicated more top, the envelopes of the different channels can be coded at 105 as differences.
• the transfer function of the filter casing 170 is extrapolated in the B band. 1 + B 3 , of the spectral envelope of the component N, L.
• the modules 105 and 175 are not necessary.
• the spatial component N is combined with the signal from 170, by means of the summator 181, to provide a reconstructed channel N, the spectrum of which is shown in FIG. 6d.
• the envelope filters can be applied in the time domain or in the frequency domain.
• the device according to the invention has been shown in the form of functional modules, it goes without saying that all or part of this device can be produced by means of a single processor or a plurality of dedicated or non-dedicated processors.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• Spectroscopy & Molecular Physics (AREA)
• Computational Linguistics (AREA)
• Signal Processing (AREA)
• Health & Medical Sciences (AREA)
• Audiology, Speech & Language Pathology (AREA)
• Human Computer Interaction (AREA)
• Acoustics & Sound (AREA)
• Multimedia (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=8860516

Family Applications (1)

Country Status (3)

Families Citing this family (5)

Family Cites Families (4)

• 2001
  • 2001-02-23 FR FR0102681A patent/FR2821475B1/fr not_active Expired - Fee Related
• 2002
  • 2002-02-18 WO PCT/FR2002/000619 patent/WO2002069326A1/fr not_active Application Discontinuation
  • 2002-02-18 EP EP02706874A patent/EP1362344A1/de not_active Withdrawn

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events