CN101115051B - Audio signal processing method, system and audio signal transmitting/receiving device - Google Patents

Abstract

The invention discloses an audio signal processing method and system, and a transceiver device of audio signals. At the transmitting end of audio signals, the residue signals and mask threshold signals obtained by the processing of audio signals are respectively coded to residue signals multiple description and multiple description; and every residue signal description and one path of multiplex mask threshold signal description are combined to form a multiplex description that contains the residue signals and mask threshold signals. At the receiving end of the audio signals, the residue signal description and the mask threshold signal description contained in every one of the descriptions received are decommuted, all residue signal descriptions created after decommutation are decoded to one path of residue signals, and then all mask threshold signals created after decommutation are decoded to one path of mask threshold signals. The invention can effectively enhance audio quality in the process of communication and enhance user satisfaction degree.

Description

Acoustic signal processing method, system and audio signal transmitting/receiving device

Technical field

The present invention relates to the communications field, be specifically related to a kind of acoustic signal processing method, system and audio signal transmitting/receiving device.

Background technology

Along with the fast development of the communication technology, channel width and transmission rate are greatly improved, and the network equipment and communication terminal disposal ability also constantly strengthen; Simultaneously, various audio coding algorithms have all obtained suitable raising on coding quality and code efficiency, and this makes real-time high quality audio business also promptly merge to various Modern Communication System.Yet, in the reality operation, owing to reasons such as network congestion, channel disturbance and noises, communication network based on packet switch all inevitably faces packet loss and long problem of delaying, and this causes can being subjected to packet loss and having a strong impact on of delaying undoubtedly by the audio signal quality of Internet protocol (IP) network and mobile communication system transmission.

At present, the acoustic signal processing method that is used to reduce because of packet loss causes audio quality to descend has two kinds substantially, below these two kinds of methods is introduced respectively.

First kind of acoustic signal processing method is: information source is divided into the different level of priority by certain criterion, according to communication channel characteristics each layer is carried out hierarchy optimization again; The information source that to finish hierarchy optimization is afterwards carried out packing data respectively by pairing different levels separately and is handled, and forms the multiple description coded bit stream of layering, comprises basic multiple description coded bit stream of layer and the multiple description coded bit stream of enhancement layer; At last, packing data is handled formed above-mentioned bit stream and sent to receiving terminal.

When receiving the multiple description coded bit stream of basic layer, receiving terminal can recover the essential information from information source; When receiving when arrive basic layer multiple description coded bit stream and the multiple description coded bit stream of enhancement layer simultaneously, receiving terminal then can better recover information from information source with less distortion.

The purpose of first kind of acoustic signal processing method is to make receiving terminal recover information from information source with the little distortion of trying one's best, but its mode of operation can be introduced following point inevitably:

1, only under situation about receiving from the multiple description coded bit stream of basic layer of information source, receiving terminal just can carry out normal decoder; In case the multiple description coded bit stream generation packet loss of basic layer, receiving terminal can't normally carry out decode procedure so;

2, receiving terminal can only recover the enhancement layer partial information from information source when only receiving the multiple description coded bit stream of enhancement layer, can't reconstruct from the main information of information source;

3, in order to make receiving terminal can normally receive the multiple description coded bit stream of basic layer, and prevent the generation of error code or packet loss, when transmission, will carry out processing such as forward error correction or packet loss re-transmission to the multiple description coded bit stream of basic layer usually; This will obviously reduce communication efficiency, and increase communication is delayed;

4, because information source has been carried out layering according to priority, therefore each the hierarchical coding bit stream that generates also will have different priority, and being correctly decoded of low priority hierarchical coding bit stream must being correctly decoded based on high priority hierarchical coding bit stream; When the decode procedure of high priority hierarchical coding bit stream went wrong, being correctly decoded at all of low priority hierarchical coding bit stream do not known where to begin.

In sum, because first kind of acoustic signal processing method carried out layering to information source according to priority, therefore the reliability that improves communication process sound intermediate frequency quality is lower, can't effectively reduce the problem that descends because of packet loss, the communication process sound intermediate frequency quality of delaying causing; And the general who has surrendered obviously reduces user satisfaction under the audio quality.

Second kind of acoustic signal processing method is: utilize the preprocessor by psychoacoustic model control of present widespread usage that audio signal is handled, obtain having removed the residual signal with the incoherent time domain albefaction of the sense of hearing, and the residual signal that obtains is carried out multiple description codedly describing coding to obtain two-way or multichannel; Again the description coding that obtains is carried out undistorted entropy coding to remove the redundance of information source, at last the coded bit stream that is generated behind the undistorted entropy coding is sent into send channel.

Second kind of acoustic signal processing method tries hard to improve audio signal quality with psychoacoustic model, but its mode of operation can be introduced following point inevitably:

1, second kind of acoustic signal processing method mainly adopts time domain prediction and time domain coding method, and frequency resolution is lower, can not remove the correlation between the irrelevant degree of the sense of hearing and each frequency component of audio signal preferably;

2, just be transferred to receiving terminal by the signal after the psychoacoustic model processing, and be not to transmit as a part of describing coding as side information; In case so the parameter in the psychoacoustic model is lost, receiving terminal can't be correctly decoded the audio signal of receiving, thereby anti-packet loss poor-performing.

In sum, because second kind of acoustic signal processing method is the basis of Audio Signal Processing with the psychoacoustic model of anti-packet loss poor-performing just, therefore the reliability that improves communication process sound intermediate frequency quality is lower, can't effectively reduce the problem that descends because of packet loss, the communication process sound intermediate frequency quality of delaying causing; And the general who has surrendered obviously reduces user satisfaction under the audio quality.

Summary of the invention

In view of this, main purpose of the present invention is to provide a kind of acoustic signal processing method and system, with the audio quality in effective raising communication process, improves user satisfaction.

Another object of the present invention is to provide a kind of audio signal transmitting/receiving device,, improve user satisfaction with the audio quality in effective raising communication process.

For achieving the above object, technical scheme of the present invention is achieved in that

A kind of acoustic signal processing method comprises:

At the audio signal transmitting terminal, to describe and multichannel masking threshold signal description for the multichannel residual signal at the residual signal and the masking threshold signal encoding of Audio Signal Processing gained respectively, respectively with road He Lu in each road residual signal description and the multichannel masking threshold signal description, generate the description that multichannel all includes residual signal description and masking threshold signal description again;

At the audio signal receiving terminal, all that receive are included all that residual signal is described and the description of masking threshold signal description in each residual signal that is comprised in describing describe and the shunt of masking threshold signal description, all residual signals that the shunt back is generated are described and are decoded as one road residual signal again, also all masking threshold signal descriptions that the shunt back is generated are decoded as one road masking threshold signal, residual signal and the masking threshold signal that obtains after the decoding carried out parameter reconstruct and the synthetic processing of time-frequency, generate the audio signal of reconstruct.

The method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the odd even separating treatment, and to finishing multichannel residual signal description the carrying out entropy coding processing that this processing generates;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the odd even separating treatment, and carry out entropy coding and handle finishing multichannel masking threshold signal description that this processing generates.

All residual signals are described the method that is decoded as one road residual signal is:

All residual signals are described carry out the entropy decoding processing, and describe and carry out that odd even is synthetic to be handled finishing all residual signals after the decoding processing, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and carrying out to finishing all masking threshold signal descriptions after the decoding processing that odd even is synthetic to be handled, all the masking threshold signal descriptions after will carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

The method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the signal branch to handling, and describe, again the multichannel residual signal that generates is described and carried out entropy coding and handle finishing the multichannel residual signal that multichannel residual signal that processing generates carries out dualistic transformation respectively and generate corresponding way;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the signal branch to handling, and, again the multichannel masking threshold signal description that generates is carried out entropy coding and handle finishing the multichannel masking threshold signal description that multichannel masking threshold signal that processing generates carries out dualistic transformation respectively and generates corresponding way.

All residual signals are described the method that is decoded as one road residual signal is:

All residual signals are described carry out the entropy decoding processing, and describe and carry out the antithesis inverse transformation and handle finishing all residual signals after the decoding processing, all residual signals after will finishing the antithesis inverse transformation again and handling are described and are synthesized one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and carry out the antithesis inverse transformation and handle finishing all masking threshold signal descriptions after the decoding processing, all the masking threshold signal descriptions that will finish again after the antithesis inverse transformation is handled synthesize one road masking threshold signal.

Described multichannel is described as two-way and describes.

The method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the scalar quantization encoding process of describing more, and to finishing multichannel residual signal description the carrying out entropy coding processing that processing generates;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the scalar quantization encoding process of describing more, and carry out entropy coding and handle finishing multichannel masking threshold signal description that processing generates.

All residual signals are described the method that is decoded as one road residual signal is:

The entropy decoding processing is carried out in all residual signals descriptions, and describe and carry out the scalar quantization decoding processing of describing more finishing all residual signals after the decoding processing, will carry out all residual signals of describing after the scalar quantization decoding processing more and describe and be decoded as one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and all masking threshold signal descriptions of finishing after the decoding processing are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

Described multichannel is described as the above description of two-way.

Described residual signal is to carry out after time frequency analysis, the residual signal analysis resulting to original audio signal;

Described time frequency analysis method is: described original audio signal is comprised the processing of revising discrete cosine transform MDCT, obtain the time-frequency conversion parameter;

The method of described residual signal analysis is: remove irrelevant information of the sense of hearing or irrelevant degree in the described time-frequency conversion parameter.

Described masking threshold signal is to carry out after the psychoacoustic model analysis resulting to original audio signal.

Further description of multichannel residual signal and the multichannel masking threshold signal description that transmitting terminal generated carried out undistorted coding and audio pack processing.

Before receiving terminal carries out shunt, further carry out audio pack and unpack and undistorted decoding processing.

A kind of audio signal processing, this system comprises the multiple description coded device that is positioned at the audio signal transmitting terminal, is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal; Also comprise the multiple description encoding device that is positioned at the audio signal receiving terminal, form by the residual signal multiple description encoding device that all links to each other, masking threshold signal multiple description encoding device with splitter;

Wherein, residual signal/multiple description coded the device of masking threshold signal, the residual signal that is used for receiving is encoded to the multichannel residual signal and describes, and is multichannel masking threshold signal description with the masking threshold signal encoding that receives, and each road description that the back of will encoding is generated sends to mixer;

Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the description that multichannel all includes residual signal description and masking threshold signal description;

Splitter, the residual signal that all that are used for receiving include all that residual signal is described and each of the description of masking threshold signal description comprised in describing describe and the masking threshold signal description along separate routes, and all residual signals descriptions and the masking threshold signal description that are generated after will be along separate routes send to residual signal/masking threshold signal multiple description encoding device;

Residual signal/masking threshold signal multiple description encoding device, all residual signals descriptions that are used for receiving are decoded as one road residual signal, and all masking threshold signal descriptions that receive are decoded as one road masking threshold signal;

Described multiple description encoding device further links to each other with the parameter reconstruct module that is connected with the time-frequency synthesis module;

Wherein, the parameter reconstruct module is used to receive residual signal and the masking threshold signal that the multiple description encoding device is generated, and the signal of receiving is carried out parameter reconstruct handle, and will finish the time-frequency conversion parameter that is generated after handling again and send to the time-frequency synthesis module;

The time-frequency synthesis module is used for the time-frequency conversion parameter of receiving is carried out the synthetic processing of time-frequency, generates the audio signal of reconstruct.

Described residual signal/multiple description coded the device of masking threshold signal comprises continuous odd even separation module, entropy coder;

Wherein, the odd even separation module, be used for described residual signal is carried out the odd even separating treatment, and will finish multichannel residual signal that this processing generates and describe and send to described entropy coder, described masking threshold signal is carried out the odd even separating treatment, and will finish the multichannel masking threshold signal description that this processing generates and send to described entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

Described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, odd even synthesis module;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals descriptions that will finish after the decoding processing send to the odd even synthesis module, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the odd even synthesis module;

The odd even synthesis module, be used for the synthetic processing of odd even is carried out in all residual signals descriptions of receiving, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal, all masking threshold signal descriptions of receiving are carried out odd even is synthetic to be handled, all the masking threshold signal descriptions after will carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

Described residual signal/multiple description coded the device of masking threshold signal comprises that the residual signal/masking threshold signal branch that links to each other successively is to module, dualistic transformation module and entropy coder;

Wherein, residual signal/masking threshold signal divides module, be used for described residual signal is carried out the signal branch to handling, and will finish the multichannel residual signal that processing generates and send to the dualistic transformation modules of describing more, described masking threshold signal is carried out the signal branch to handling, and will finish the multichannel masking threshold signal that processing generates and send to the dualistic transformation modules of describing more;

The dualistic transformation modules of describing more, be used for the multichannel residual signal that the multichannel residual signal of receiving carries out dualistic transformation respectively and generates corresponding way is described, again the multichannel residual signal that is generated is described and sent to entropy coder, the multichannel masking threshold signal of receiving is carried out dualistic transformation respectively and generates the multichannel masking threshold signal description of corresponding way, again the multichannel masking threshold signal description that is generated is sent to entropy coder;

Entropy coder is used for receiving that the multichannel residual signal is described and multichannel masking threshold signal description carries out the entropy coding processing.

Described residual signal/masking threshold signal multiple description encoding device comprises entropy decoder, antithesis inverse transformation decoder and the residual signal/masking threshold signal synthesizing module that links to each other successively;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and will finish all residual signals that generated after the decoding processing and describe to send to describe antithesis inverse transformation decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the antithesis inverse transformation decoders of describing more;

The antithesis inverse transformation decoders of describing more, being used for that the antithesis inverse transformation is carried out in all residual signals descriptions of receiving handles, and all residual signals descriptions that will finish after the processing of antithesis inverse transformation send to residual signal/masking threshold signal synthesizing module, all masking threshold signal descriptions of receiving are carried out the antithesis inverse transformation handle, and all the masking threshold signal descriptions that will finish after the antithesis inverse transformation is handled send to residual signal/masking threshold signal synthesizing module;

Residual signal/masking threshold signal synthesizing module, all residual signals descriptions that are used for receiving synthesize one road residual signal, and all masking threshold signal descriptions that receive are synthesized one road masking threshold signal.

Described residual signal/multiple description coded the device of masking threshold signal comprises continuous many descriptions scalar quantizer, entropy coder;

Wherein, the scalar quantizer of describing more, be used for described residual signal is carried out the scalar quantization encoding process of describing more, and will finish multichannel residual signal that processing generates and describe and send to entropy coder, described masking threshold signal is carried out describing the scalar quantization encoding process more, and will finish the multichannel masking threshold signal description that processing generates and send to entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

Described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, describes the scalar quantization decoder more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals that will finish after the decoding processing describe to send to describe the scalar quantization decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the scalar quantization decoders of describing more;

The scalar quantization decoders of describing more, be used for all residual signals of receiving are described to carry out describing the scalar quantization decoding processing more, all residual signals after the scalar quantization of description more than will the carrying out decoding processing are described and are decoded as one road residual signal, all masking threshold signal descriptions of receiving are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

The described scalar quantizer of describing is to handle two scalar quantizer of describing described above more more, and the described scalar quantization decoders of describing are to handle two scalar quantization decoders of describing described above more more.

Described multiple description coded device further links to each other with the residual signal analysis module that is connected with the time frequency analysis module, also links to each other with psycho-acoustic module;

Wherein, the time frequency analysis module is used for original audio signal is carried out time frequency analysis, and will finish and analyze the time-frequency conversion parameter that the back generated and send to the residual signal analysis module;

The residual signal analysis module is used for time-frequency conversion parameter and the masking threshold signal received are carried out the residual signal analysis, and the residual signal that will be generated after will analyzing sends to described multiple description coded device;

Psycho-acoustic module is used for original audio signal is carried out the psychoacoustic model analysis, and the masking threshold signal that is generated after analyzing is sent to described multiple description coded device and residual signal analysis module.

Described multiple description coded device further links to each other with undistorted coding and audio pack processing module;

Undistorted coding and audio pack processing module are used to receive the description that multiple description coded device generates, and undistorted coding and audio pack processing are carried out in the description of receiving.

Described multiple description encoding device further unpacks with audio pack and undistorted decoder module links to each other;

Audio pack unpacks and undistorted decoder module, is used for that audio pack is carried out in the multichannel description of receiving and unpacks and undistorted decoding processing.

A kind of audio signal dispensing device, this device is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal;

Residual signal/multiple description coded the device of masking threshold signal, the residual signal that is used for receiving is encoded to the multichannel residual signal and describes, with the masking threshold signal encoding that receives is multichannel masking threshold signal description, and each road description that the back of will encoding is generated sends to mixer;

Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the description that multichannel all includes residual signal description and masking threshold signal description.

Described residual signal/multiple description coded the device of masking threshold signal comprises continuous odd even separation module, entropy coder;

Wherein, the odd even separation module, be used for described residual signal is carried out the odd even separating treatment, and will finish multichannel residual signal that this processing generates and describe and send to described entropy coder, described masking threshold signal is carried out the odd even separating treatment, and will finish the multichannel masking threshold signal description that this processing generates and send to described entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

Described residual signal/multiple description coded the device of masking threshold signal, comprise successively residual signal/masking threshold signal branch of linking to each other to module, dualistic transformation module and entropy coder are described more;

Wherein, residual signal/masking threshold signal divides module, be used for described residual signal is carried out the signal branch to handling, and will finish the multichannel residual signal that processing generates and send to the dualistic transformation modules of describing more, described masking threshold signal is carried out the signal branch to handling, and will finish the multichannel masking threshold signal that processing generates and send to the dualistic transformation modules of describing more;

The dualistic transformation modules of describing more, be used for the multichannel residual signal that the multichannel residual signal of receiving carries out dualistic transformation respectively and generates corresponding way is described, again the multichannel residual signal that is generated is described and sent to entropy coder, the multichannel masking threshold signal of receiving is carried out dualistic transformation respectively and generates the multichannel masking threshold signal description of corresponding way, again the multichannel masking threshold signal description that is generated is sent to entropy coder;

Entropy coder is used for receiving that the multichannel residual signal is described and multichannel masking threshold signal description carries out the entropy coding processing.

Described residual signal/multiple description coded the device of masking threshold signal comprises continuous many descriptions scalar quantizer, entropy coder;

Wherein, the scalar quantizer of describing more, be used for described residual signal is carried out the scalar quantization encoding process of describing more, and will finish multichannel residual signal that processing generates and describe and send to entropy coder, described masking threshold signal is carried out describing the scalar quantization encoding process more, and will finish the multichannel masking threshold signal description that processing generates and send to entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

The described scalar quantizer of describing is to handle two scalar quantizer of describing described above more more.

Described multiple description coded device further links to each other with the residual signal analysis module that is connected with the time frequency analysis module, also links to each other with psycho-acoustic module;

Wherein, the time frequency analysis module is used for original audio signal is carried out time frequency analysis, and will finish and analyze the time-frequency conversion parameter that the back generated and send to the residual signal analysis module;

The residual signal analysis module is used for time-frequency conversion parameter and the masking threshold signal received are carried out the residual signal analysis, and the residual signal that will be generated after will analyzing sends to described multiple description coded device;

Psycho-acoustic module is used for original audio signal is carried out the psychoacoustic model analysis, and the masking threshold signal that is generated after analyzing is sent to described multiple description coded device and residual signal analysis module.

Described multiple description coded device further links to each other with undistorted coding and audio pack processing module;

Undistorted coding and audio pack processing module are used to receive the multichannel description that multiple description coded device generates, and undistorted coding and audio pack processing are carried out in the multichannel description of receiving.

A kind of audio signal receiving system, this device is made up of the residual signal multiple description encoding device that all links to each other with splitter, masking threshold signal multiple description encoding device;

Wherein, splitter, the residual signal that all that are used for receiving include all that residual signal is described and each of the description of masking threshold signal description comprised in describing describe and the masking threshold signal description along separate routes, and all residual signals descriptions and the masking threshold signal description that are generated after will be along separate routes send to residual signal/masking threshold signal multiple description encoding device;

Residual signal/masking threshold signal multiple description encoding device, all residual signals descriptions that are used for receiving are decoded as one road residual signal, and all masking threshold signal descriptions that receive are decoded as one road masking threshold signal;

Described multiple description encoding device further links to each other with the parameter reconstruct module that is connected with the time-frequency synthesis module;

Wherein, the parameter reconstruct module is used to receive residual signal and the masking threshold signal that the multiple description encoding device is generated, and the signal of receiving is carried out parameter reconstruct handle, and will finish the time-frequency conversion parameter that is generated after handling again and send to the time-frequency synthesis module;

The time-frequency synthesis module is used for the time-frequency conversion parameter of receiving is carried out the synthetic processing of time-frequency, generates the audio signal of reconstruct.

Described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, odd even synthesis module;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals descriptions that will finish after the decoding processing send to the odd even synthesis module, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the odd even synthesis module;

The odd even synthesis module, be used for the synthetic processing of odd even is carried out in all residual signals descriptions of receiving, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal, all masking threshold signal descriptions of receiving are carried out odd even is synthetic to be handled, all masking threshold signal descriptions of respecting after carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

Description of drawings

Described residual signal/masking threshold signal multiple description encoding device comprises continuous successively entropy decoder, describes antithesis inverse transformation decoder and residual signal/masking threshold signal synthesizing module more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and will finish all residual signals that generated after the decoding processing and describe to send to describe antithesis inverse transformation decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the antithesis inverse transformation decoders of describing more;

The antithesis inverse transformation decoders of describing more, being used for that the antithesis inverse transformation is carried out in all residual signals descriptions of receiving handles, and all residual signals descriptions that will finish after the processing of antithesis inverse transformation send to residual signal/masking threshold signal synthesizing module, all masking threshold signal descriptions of receiving are carried out the antithesis inverse transformation handle, and all the masking threshold signal descriptions that will finish after the antithesis inverse transformation is handled send to residual signal/masking threshold signal synthesizing module;

Residual signal/masking threshold signal synthesizing module, all residual signals descriptions that are used for receiving synthesize one road residual signal, and all masking threshold signal descriptions that receive are synthesized one road masking threshold signal.

Described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, describes the scalar quantization decoder more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals that will finish after the decoding processing describe to send to describe the scalar quantization decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the scalar quantization decoders of describing more;

The scalar quantization decoders of describing more, be used for all residual signals of receiving are described to carry out describing the scalar quantization decoding processing more, all residual signals after the scalar quantization of description more than will the carrying out decoding processing are described and are decoded as one road residual signal, all masking threshold signal descriptions of receiving are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

The described scalar quantization decoder of describing is to handle two scalar quantization decoders of describing described above more more.

Described multiple description encoding device further unpacks with audio pack and undistorted decoder module links to each other;

Audio pack unpacks and undistorted decoder module, is used for that audio pack is carried out in the description of receiving and unpacks and undistorted decoding processing.

Compared with prior art, acoustic signal processing method provided by the present invention, system and audio signal transmitting/receiving device, at the audio signal transmitting terminal, to be that many descriptions of residual signal and masking threshold signal are described more at the residual signal and the masking threshold signal encoding of Audio Signal Processing gained respectively, respectively with road He Lu in each road residual signal description and the multichannel masking threshold signal description, generate the multichannel that all includes residual signal and masking threshold signal and describe again; At the audio signal receiving terminal, with each describes residual signal description and the shunt of masking threshold signal description that is comprised in all descriptions that receive, all residual signals that the shunt back is generated are described and are decoded as one road residual signal again, also all masking threshold signal descriptions that generated after the shunt are decoded as one road masking threshold signal.

As seen, the present invention can effectively improve the audio quality in the communication process, improves user satisfaction.

Fig. 1 is the audio signal encoder structure and the schematic diagram of a preferred embodiment of the present invention;

Fig. 2 is the audio signal decoder structure and the schematic diagram of a preferred embodiment of the present invention;

Fig. 3 is multiple description coded device structure of the present invention and schematic diagram;

Fig. 4 is multiple description encoding device structure of the present invention and schematic diagram;

Fig. 5 is residual signal multiple description encoding device structure of the present invention and schematic diagram;

Fig. 6 is masking threshold signal multiple description encoding device structure of the present invention and schematic diagram;

Fig. 7 is multiple description coded device structure of residual signal of the present invention and schematic diagram;

Fig. 8 is multiple description encoding device structure and the schematic diagram that matches with multiple description coded device among Fig. 7;

Fig. 9 is multiple description coded device structure of residual signal of the present invention and schematic diagram;

Figure 10 is multiple description encoding device structure and the schematic diagram that matches with multiple description coded device among Fig. 9;

Figure 11 is the Audio Signal Processing schematic diagram of a preferred embodiment of the present invention;

Figure 12 is the quantized interval schematic diagram of a preferred embodiment of the present invention;

Figure 13 is that the single of a preferred embodiment of the present invention describes and the quantized interval contrast schematic diagram of describing more;

Figure 14 is multiple description coded device structure and the schematic diagram based on Figure 11;

Figure 15 is multiple description encoding device structure and the schematic diagram based on Figure 11;

Figure 16 is the Audio Signal Processing schematic diagram of another preferred embodiment of the present invention;

Figure 17 is multiple description coded device structure and the schematic diagram based on Figure 16;

Figure 18 is multiple description encoding device structure and the schematic diagram based on Figure 16;

Figure 19 is the three-dimensional charts of many descriptions scalar quantizer encoding and decoding of a preferred embodiment of the present invention.

Embodiment below in conjunction with drawings and the specific embodiments to the detailed description of the invention.

Acoustic signal processing method provided by the present invention, at the audio signal transmitting terminal, to be that many descriptions of residual signal and masking threshold signal are described more at the residual signal and the masking threshold signal encoding of Audio Signal Processing gained respectively, respectively with road He Lu in each road residual signal description and the multichannel masking threshold signal description, generate the multichannel that all includes residual signal and masking threshold signal and describe again; At the audio signal receiving terminal, with each describes residual signal description and the shunt of masking threshold signal description that is comprised in all descriptions that receive, all residual signals that the shunt back is generated are described and are decoded as one road residual signal again, also all masking threshold signal descriptions that generated after the shunt are decoded as one road masking threshold signal.

Audio signal processing provided by the present invention comprises the multiple description coded device that is positioned at the audio signal transmitting terminal, is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal; Also comprise the multiple description encoding device that is positioned at the audio signal receiving terminal, form by the residual signal multiple description encoding device that all links to each other, masking threshold signal multiple description encoding device with splitter;

Wherein, the residual signal/multiple description coded device of masking threshold signal, the residual signal/masking threshold signal encoding that is used for receiving is that residual signal/masking threshold signal is described more, and many descriptions that the back of will encoding is generated send to mixer; Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the many descriptions that all include residual signal and masking threshold signal; Splitter, the residual signal that each description of all descriptions that is used for receiving is comprised is described and the shunt of masking threshold signal description, and the many descriptions of residual signal/masking threshold signal that generated after the shunt are sent to residual signal/masking threshold signal multiple description encoding device; Residual signal/masking threshold signal multiple description encoding device, the residual signal/masking threshold signal multiple description encoding that is used for receiving is one road residual signal/masking threshold signal.

Audio signal dispensing device provided by the present invention is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal; Wherein, the residual signal/multiple description coded device of masking threshold signal, the residual signal/masking threshold signal encoding that is used for receiving is that residual signal/masking threshold signal is described more, and many descriptions that the back of will encoding is generated send to mixer; Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the multichannel that all includes residual signal and masking threshold signal and describes.

Audio signal receiving system provided by the present invention is made up of the residual signal multiple description encoding device that all links to each other with splitter, masking threshold signal multiple description encoding device; Wherein, splitter, the residual signal that each description of all descriptions that is used for receiving is comprised is described and the shunt of masking threshold signal description, and the many descriptions of residual signal/masking threshold signal that generated after the shunt are sent to residual signal/masking threshold signal multiple description encoding device; Residual signal/masking threshold signal multiple description encoding device, the residual signal/masking threshold signal multiple description encoding that is used for receiving is one road residual signal/masking threshold signal.

In principle, the present invention can roughly be divided into two aspects, in the Audio Signal Processing aspect: need carry out the analyses of describing with synthetic to audio signal more, as: audio signal is decomposed into mutual incoherent masking threshold signal and residual signal; Quantizing and the coding aspect: respectively masking threshold signal and residual signal are carried out the multi-description coding-decodings processing of describing many decoders more.Have again, when the channel packet loss is serious, can also carry out error concealment to packet loss according to the historical record that difference is described.

Referring to Fig. 1, Fig. 1 is the audio signal encoder structure and the schematic diagram of a preferred embodiment of the present invention.Among Fig. 1, the original audio signal that selects through window is divided into two-way, and wherein one tunnel input time frequency analysis module 110 is imported psycho-acoustic module 120 with another road.

120 pairs of audio signals of receiving of psycho-acoustic module are carried out the psychoacoustic model analysis, obtain the masking threshold signal relevant, and the masking threshold signal that obtains is sent to residual signal analysis module 130, multiple description coded device 140 and Bit Allocation in Discrete module 150 respectively with the current audio frame of receiving.110 pairs of audio signals of receiving of time frequency analysis module are revised discrete cosine transform time frequency analysis such as (MDCT) and are handled, and time frequency analysis are handled time-frequency conversion parameter such as resulting frequency domain MDCT coefficient and send to residual signal analysis module 130; The masking threshold signal that 130 utilizations of residual signal analysis module are received is removed irrelevant information of the sense of hearing or the irrelevant degree in the time-frequency conversion parameter of receiving, has obtained removing the residual signal of sense of hearing incoherence and has sent to multiple description coded device 140.

Residual signal that can characterize current audio signals information and masking threshold signal that 140 pairs of multiple description coded devices are received carry out multiple description coded, obtaining two can be separately or the description of Combined Treatment: describe 1 and describe 2, and will describe 1 and description 2 send to undistorted coding and audio pack processing module 160.Bit Allocation in Discrete module 150 is with the control information as Bit Allocation in Discrete of the masking threshold signal received, determine the Bit Allocation in Discrete mode, describe side informations such as sign and quantizer sign according to this masking threshold signal, and the side information of determining is sent to undistorted coding and audio pack processing module 160.

Undistorted encoding process such as Huffman coding, arithmetic coding or run length coding, RLC are carried out in 160 pairs of descriptions of receiving of undistorted coding and audio pack processing module 1 and description 2, to eliminate the information source redundance, further reduce bit rate; Afterwards described side information is added description 1 and the description 2 of finishing encoding process and carry out bit-envelope, again the coded bit stream of packaged description 1 and description 2 is sent into channel.

Referring to Fig. 2, Fig. 2 is the audio signal decoder structure and the schematic diagram of a preferred embodiment of the present invention.Among Fig. 2, audio pack unpacks and undistorted decoder module 210, multiple description encoding device 220, parameter reconstruct module 230 and time-frequency synthesis module 240 link to each other successively.Wherein, audio pack unpacks and 210 pairs of descriptions of receiving of undistorted decoder module 1 unpack and decode operation with description 2, and the description 1 that will obtain, description 2 and side information send to multiple description encoding device 220; 220 pairs of descriptions of receiving 1 of multiple description encoding device, description 2 and side information carry out multiple description encoding, and will decode resulting masking threshold signal and residual signal send to parameter reconstruct module 230; Parameter reconstruct module 230 is utilized masking threshold signal and the residual signal reconstruct time-frequency conversion parameter of receiving, and will finish the resulting time-frequency conversion parameter of reconstruct and send to time-frequency synthesis module 240, go out audio signal by time-frequency synthesis module 240 according to the time-frequency conversion parameter reconstruct.

In above-mentioned Fig. 1, Fig. 2, be provided with psychoacoustic model in the psycho-acoustic module 120, this model description has the apperceive characteristic of people's ear to audio signal, and this description is mainly reflected in masking characteristics; From the angle of audio compression and coding, psychoacoustic model has determined the energy of the imperceptible maximum quantizing noise of people's ear in the critical band, in other words conj.or perhaps the masking by noise thresholding on sense of hearing perception meaning.Particularly, described psychoacoustic model can have different implementation methods, as: the model 1 that the MPEG-1 that employing is commonly used at present and first and second layer audio coding of MPEG-2 are adopted, or adopt the model 2 that is adopted as the 3rd layer MP3 audio coding, or adopt substrate curve (Floor) in the Ogg Vorbis audio coding, also can adopt the psychoacoustic model of other kinds such as psychoacoustic model in the AC3 audio coding.

The function of time frequency analysis module 110 is that time-domain audio signal is carried out conversion or filtering, the redundance of being brought with the correlation of removing in the original audio signal.Time frequency analysis based on conversion can adopt MDCT, modulated lapped transform (mlt) (MLT) or wavelet transform time-frequency conversion modes such as (DWT), and what obtain based on the time frequency analysis of conversion is the audio frequency parameter of transform domain or frequency domain; Time frequency analysis based on filtering can adopt the sub-band filter algorithm that is similar to MPEG-1 and the employing of MPEG-2 audio coding, and carries out handling as the time-frequency conversion of MDCT, MLT or DWT etc. in each subband, and what obtain also is the audio frequency parameter of transform domain or frequency domain.In addition, the function of the time-frequency synthesis module 240 among Fig. 2 and the functions reversed of time frequency analysis module 110 that is: utilize the frequency domain audio frequency parameter to carry out inverse transformation, obtain the audio signal after the reconstruct.

The major function of residual signal analysis module 130 is to reject the uncorrelated degree of the sense of hearing that remains in through in the frequency-domain audio signals behind the time frequency analysis.If this processing is carried out at linear domain, can obtain residual signal divided by the masking threshold signal with the audio frequency frequency domain parameter; If carry out, can deduct the masking threshold signal with the audio frequency frequency domain parameter and obtain residual signal at log-domain.In addition, the function of the parameter reconstruct module 230 among Fig. 2 and the functions reversed of residual signal analysis module 130, that is: residual signal and the masking threshold signal reconstruction audio frequency frequency domain parameter that obtains with multiple description encoding.If linear domain can obtain the time-frequency conversion parameter with sense of hearing residual signal and masking threshold signal multiplication; If log-domain can obtain the time-frequency conversion parameter with residual signal and masking threshold signal plus.

The function of Bit Allocation in Discrete module 150 is according to the multiple description coded quantizer quantified precision of receiving of masking threshold signal controlling, according to available bit number dynamic bit being carried out in the formation of undistorted coding and audio pack simultaneously distributes, adjust quantified precision and Bit Allocation in Discrete repeatedly with alternative manner, use up up to available bit number, perhaps reached default coding quality.In actual applications, Bit Allocation in Discrete module 150 is supported fixed rate (CBR), variable bit rate (VBR) and Mean Speed coding modes such as (ABR).

Undistorted coding and audio pack processing module 160 are used for respectively describing 1 and describe 2 and carry out undistorted entropy coding, add side information afterwards and form two codings and describe bit streams output.Described two description bit streams of output can be that be equal to or incoordinate aspect importance, and the required bit number of the coding of above-mentioned two descriptions can be identical or different, and these two code rates of describing bit streams can be identical or different.In addition, audio pack unpacks and the function of undistorted decoder module 210 and the functions reversed of undistorted coding and audio pack processing module 160, that is: describe bit stream to two and unpack and undistorted decoding, obtains two audio description information.

In actual applications, multiple description coded device 140 shown in Figure 1 can be as shown in Figure 3, and 220 of multiple description encoding devices shown in Figure 2 can be as shown in Figure 4.

Referring to Fig. 3, Fig. 3 is multiple description coded device structure of the present invention and schematic diagram.Among Fig. 3, the multiple description coded device 310 of residual signal links to each other with mixer 330, mixer 340 respectively with the multiple description coded device 320 of masking threshold signal.During practical application, 310 pairs of residual signals of receiving of the multiple description coded device of residual signal carry out encoding process, and with the formed residual signal of encoding process describe 1 and residual signal describe 2 and send to mixer 330, mixer 340 respectively; 320 pairs of masking threshold signals of receiving of the multiple description coded device of masking threshold signal carry out encoding process, and formed masking threshold signal description 1 of encoding process and masking threshold signal description 2 are sent to mixer 330, mixer 340 respectively.330 pairs of residual signals of receiving of mixer describe 1 and masking threshold signal description 1 close the road and handle, and will finish and close the description 1 that is generated behind the road and send; 340 pairs of residual signals of receiving of mixer describe 2 and masking threshold signal description 2 close the road and handle, and will finish and close the description 2 that is generated behind the road and send.

The encryption algorithm that multiple description coded device 310 of residual signal and the multiple description coded device 320 of masking threshold signal can adopt has multiple, as: many descriptions scalar quantization algorithm (MDSQ) at present commonly used, describe transition coding algorithm (MDTC) more or describe vector quantization (VQ) method more.It should be noted that: for residual signal, because the masking threshold signal may only comprise small amount of data, so the pairing multiple description coded mode of masking threshold signal can also be direct copy.

Certainly, multiple description coded device 310 of residual signal and the multiple description coded device 320 of masking threshold signal can also receive and handle side information, the side information that to finish processing again sends to mixer 330 and mixer 340, closes the road processing by limit signal and other description that mixer 330,340 pairs of mixers are received.

Referring to Fig. 4, Fig. 4 is multiple description encoding device structure of the present invention and schematic diagram.Among Fig. 4, splitter 410, splitter 420 link to each other with masking threshold signal multiple description encoding device 440 with residual signal multiple description encoding device 430 respectively.During practical application, 410 pairs of descriptions of receiving of splitter 1 are carried out shunt and are handled, and will handle along separate routes formed residual signal describe 1 and masking threshold signal description 1 send to residual signal multiple description encoding device 430, masking threshold signal multiple description encoding device 440 respectively; 420 pairs of descriptions of receiving of splitter 2 are carried out shunt and are handled, and will handle along separate routes formed residual signal describe 2 and masking threshold signal description 2 send to residual signal multiple description encoding device 430, masking threshold signal multiple description encoding device 440 respectively.430 pairs of residual signals of receiving of residual signal multiple description encoding device describe 1 and residual signal describe 2 and carry out decoding processing, and will finish the residual signal of finishing reconstruct that the decoding back generated and send; 440 pairs of masking threshold signal descriptions of receiving 1 of masking threshold signal multiple description encoding device and masking threshold signal description 2 carry out decoding processing, and will finish the masking threshold signal of finishing reconstruct that decoding back generated and send.

Certainly, splitter 410 and splitter 420 can also receive and handle side information, the side information that to finish processing again sends to residual signal multiple description encoding device 430, masking threshold signal multiple description encoding device 440, carries out decoding processing by residual signal multiple description encoding device 430,440 pairs of limit signals of receiving of masking threshold signal multiple description encoding device and other description.Have, in actual applications, what send to described splitter may have only description 1 or describe 2 again; In this case, the splitter that receives described description can normally carry out subsequent treatment such as shunt to this description, finishes description along separate routes and also can normally be carried out follow-up decoding processing.

In actual applications, residual signal multiple description encoding device 430 shown in Figure 4 can be as shown in Figure 5, and masking threshold signal multiple description encoding device 440 then can be as shown in Figure 6.

Referring to Fig. 5, Fig. 5 is residual signal multiple description encoding device structure of the present invention and schematic diagram, and decoder shown in Figure 5 is to adopt three decoder architectures that multiple description encoding is carried out in the description that receives.Particularly, if receive only a description, just with residual signal multiple description encoding device a 510 or 530 pairs of described description decodings of residual signal multiple description encoding device c as the limit decoder of receiving this description; If receive two descriptions, just with the 520 pairs of described description decodings of residual signal multiple description encoding device b as the center decoder of receiving this description.

Referring to Fig. 6, Fig. 6 is masking threshold signal multiple description encoding device structure of the present invention and schematic diagram, and decoder shown in Figure 6 is to adopt three decoder architectures that multiple description encoding is carried out in the description that receives.Particularly, if receive only a description, just with masking threshold signal multiple description encoding device a 610 or 630 pairs of described description decodings of masking threshold signal multiple description encoding device c as the limit decoder of receiving this description; If receive two descriptions, just with the 620 pairs of described description decodings of masking threshold signal multiple description encoding device b as the center decoder of receiving this description.

The position of the switch of the multiple description encoding device output signal shown in Fig. 5, Fig. 6 can be selected automatically according to the situation that receives description.

In actual applications, residual signal among Fig. 3, Fig. 4 and masking threshold signal multi-description coding-decoding device can be realized by many descriptions odd even separation algorithm, as Fig. 7, shown in Figure 8; Also can realize, as Fig. 9, shown in Figure 10 by many descriptions dualistic transformation algorithm; Can also realize by many descriptions scalar quantization algorithm, extremely shown in Figure 180 as Figure 11.

Below, be example with the accompanying drawing, be described at above-mentioned different multi-description coding-decoding algorithms respectively.Need to prove that in the following description, the object of multi-description coding-decoding mainly is a residual signal; In actual applications, also can use identical multi-description coding-decoding algorithm handles masking threshold signal or other audio signal components.

At the situation that is realized multi-description coding-decoding by many descriptions odd even separation algorithm, its operating principle is: time domain or frequency domain audio frequency parameter are separated according to its index value or natural order odd even.Because two descriptions that odd even is separated are uncorrelated fully; Therefore the redundancy of introducing between two descriptions is zero, and therefore total code rate can not increase yet.

Referring to Fig. 7, Fig. 7 is multiple description coded device structure of residual signal of the present invention and schematic diagram.Among Fig. 7,710 pairs of residual signals of receiving of odd even separation module carry out the odd even separating treatment, and will finish description 1 that processing generates, describe 2 and send to entropy coder 720, entropy coder 730 respectively; Encoding process is carried out in 720 pairs of descriptions of receiving of entropy coder 1, and the bit stream that will finish the description 1 that is generated after the encoding process sends; Encoding process is carried out in 730 pairs of descriptions of receiving of entropy coder 2, and the bit stream that will finish the description 2 that is generated after the encoding process sends.

Particularly, residual signal is represented with R (k): R (k), and k=1,2,3 ... N

Wherein, N is the number (being generally even number) of residual signal, also is simultaneously half of audio analysis window length.

The algorithm signal of describing two of residual signal is described M with two respectively more ₁(k ₁) and M ₂(k ₂) expression:

1:M is described ₁(k ₁), k ₁=1,2,3 ... N/2

2:M is described ₂(k ₂), k ₂=1,2,3 ... N/2

Odd even is described mapping algorithm and the result is as follows more so:

When k is an odd number, i.e. k=1,3,5 ... during N-1, M ₁(k ₁)=R (k)

Wherein, k ₁=(k+1)/2;

When k is an even number, i.e. k=2,4,6 ... during N, M ₂(k ₂)=R (k)

Wherein, k ₂=k/2.

After above-mentioned odd even was described conversion more, entropy coder 720,730 was respectively to many descriptions signal M ₁(k ₁) and M ₂(k ₂) encode, and the data of coding gained and the formation of masking threshold signal combination separately description bit stream are sent.

Referring to Fig. 8, Fig. 8 is multiple description encoding device structure and the schematic diagram that matches with multiple description coded device among Fig. 7.Among Fig. 8, the bit stream of 810 pairs of descriptions of receiving 1 of entropy decoder carries out decoding processing, and will finish the description 1 that is generated after the decoding processing and send to odd even synthesis module 830, odd even synthesis module 840; The bit stream of 820 pairs of descriptions of receiving 2 of entropy decoder carries out decoding processing, and will finish the description 2 that is generated after the decoding processing and send to odd even synthesis module 850, odd even synthesis module 840.

The position of the switch of the synthetic module output signal of the odd even shown in Fig. 8 can be selected automatically according to the situation that receives description, as: when receiving only a description, with regard to receiving that the odd even synthesis module of this description decodes to this description, and be communicated with the residual signal of finishing reconstruct that the output of this odd even synthesis module is generated after odd even is synthetic with output; When two descriptions all receive, with regard to receiving that the odd even synthesis module 840 of these two descriptions decodes to these two descriptions, and be communicated with the residual signal of finishing reconstruct that the output of odd even synthesis module 840 is generated after odd even is synthetic with output.

Particularly,, just respectively two descriptions are decoded, obtain describing signal two: describe 1 (M more if two descriptions have all received ₁) and describe 2 (M ₂), and these two descriptions are synthesized residual signal, concrete composition algorithm is as follows:

When k is an odd number, i.e. k=1,3,5 ... during N-1, make R (k)=M ₁(k ₁), k wherein ₁=(k+1)/2;

When k is an even number, i.e. k=2,4,6 ... during N, make R (k)=M ₂(k ₂), k wherein ₂=k/2.

If receive only a description, just this description is decoded, obtaining describing one signal more.

Signal is described for describing 1 (M when this ₁) time:

When k is an odd number, i.e. k=1,3,5 ... during N-1, make R (k)=M ₁(k ₁), k wherein ₁=(k+1)/2,

When k is an even number, i.e. k=2,4,6 ... during N, make R (k)=0, wherein k ₂=k/2.

When this describes signal for describing 2 (M ₂) time:

When k is an odd number, i.e. k=1,3,5 ... during N-1, make R (k)=0, wherein k ₁=(k+1)/2,

When k is an even number, i.e. k=2,4,6 ... during N, make R (k)=M ₂(k ₂), k wherein ₂=k/2.

So far, just successfully reconstructed residual signal.

At the situation that is realized multi-description coding-decoding by many descriptions dualistic transformation algorithm, its operating principle is: by transformation matrix T two incoherent variables A and B are transformed into two variable C and D with certain correlation.The size of correlation is determined by transformation matrix T between variable C and the D.Particularly, described transformation matrix is divided into orthogonal matrix and nonopiate matrix usually, and the scope of the correlation that these two kinds of transformation matrixs are introduced is inequality, and the implementation method of corresponding dualistic transformation algorithm is also different fully.

Below, be that example is described the dualistic transformation algorithms of describing more only with nonopiate matrix.

Each transformation parameter is defined as in the described dualistic transformation algorithm:

Input matrix: $\left[\begin{array}{c}A\\ B\end{array}\right]$

The dualistic transformation matrix: $T=\left[\begin{array}{cc}a& b\\ c& d\end{array}\right]$

The dualistic transformation output matrix: $\left[\begin{array}{c}C\\ D\end{array}\right]=T\left[\begin{array}{c}A\\ B\end{array}\right]$

Wherein, a, b, c, d is the coefficient of transformation matrix, is determining the redundancy of the introducing between two descriptions, and ad-bc=1; A, B are input signal; C, D are output signal.

The specific algorithm of integer transform is as follows: at coding side, encoder carries out the antithesis direct transform to signal A and B:

$\stackrel{\‾}{A}=\left[\frac{A}{Q}\right],$ $\stackrel{\‾}{B}=\left[\frac{B}{Q}\right]$

$W=\stackrel{\‾}{B}+\left[\frac{1+c}{d}\stackrel{\‾}{A}\right]$

$\stackrel{\‾}{D}=\left[\mathrm{dW}\right]-\stackrel{\‾}{A}$

$\stackrel{\‾}{C}=W-\left[\frac{1-b}{d}\stackrel{\‾}{D}\right]$

Wherein, Q is a quantization step, and W is an intermediate variable, and symbol ' [] ' expression rounds operation.

With

Be respectively the integer variable after A and B quantize,

With

Be respectively two integer output variables,

With

To be encoded respectively and be described bit stream to form two.

In decoding end, there are following three kinds of situations in concrete decode procedure because of the difference of the description that receives:

(1)

With

When two descriptions are all correctly received, by the decoder of receiving these two descriptions to signal

With Carry out following antithesis inverse transformation:

$W=\stackrel{\‾}{C}+\left[\frac{1-b}{d}\stackrel{\‾}{D}\right]$

$\stackrel{\‾}{A}=\left[\mathrm{dW}\right]-\stackrel{\‾}{D}$

$\stackrel{\‾}{B}=W-\left[\frac{1+c}{d}\stackrel{\‾}{A}\right]$

$\hat{A}=\stackrel{\‾}{A}Q,$ $\hat{B}=\stackrel{\‾}{B}Q$

Wherein, Q is a quantization step; W is an intermediate variable; With

Be respectively

With

Value behind the re-quantization; ' [] ' symbolic representation rounds operation.

(2) has only a description

When correctly being received, signal at first to losing

Predict:

$\tilde{C}=\stackrel{\‾}{C}Q$

$\hat{D}=\frac{{\mathrm{\&sigma;}}_c*{\mathrm{\&sigma;}}_d}{{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="H061A3571320060801E000051.png,H061A3571320060801E000053.png"\; state="\{\{state\}\}">c</figure-callout>}}^2+{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="H061A3571320060801E000051.png,H061A3571320060801E000053.png"\; state="\{\{state\}\}">q</figure-callout>}}^2}*\tilde{C}*\cos \mathrm{\&phi;}$

Right again

With

Carry out inverse transformation:

$\left[\begin{array}{c}\hat{A}\\ B\end{array}\right]=T^{-1}\left[\begin{array}{c}\tilde{C}\\ \hat{D}\end{array}\right]$

Wherein, σ _c, σ _dAnd σ _qBe respectively the standard deviation of variable C, D and quantization error;

For

Through the value after the decoder, inverse quantification;

For decoder right

The value that prediction recovers out; Matrix T ^-1Inverse matrix for matrix T; Be the decoder reconstruct value of coming out; Cos φ is the coefficient correlation between variable C, the D.(3) has only a description

When correctly being received, signal at first to losing

Predict,

$\tilde{D}=\stackrel{\&OverBar;}{D}Q$

$\hat{C}=\frac{{\mathrm{\&sigma;}}_c*{\mathrm{\&sigma;}}_d}{{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="H061A3571320060801E000051.png,H061A3571320060801E000053.png"\; state="\{\{state\}\}">c</figure-callout>}}^2+{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="q"\; filenames="H061A3571320060801E000051.png,H061A3571320060801E000053.png"\; state="\{\{state\}\}">q</figure-callout>}}^2}*\tilde{D}*\cos \mathrm{\&phi;}$

Again to

Carry out inverse transformation:

$\left[\begin{array}{c}\hat{A}\\ \hat{B}\end{array}\right]=T^{-1}\left[\begin{array}{c}\hat{C}\\ \tilde{D}\end{array}\right]$

Wherein, σ _c, σ _dAnd σ _qBe respectively the standard deviation of variable C, D and quantization error; For Through the value after the decoder, inverse quantification; For decoder through right

The value that prediction recovers out; Matrix T ^-1Inverse matrix for matrix T; Be the decoder reconstruct value of coming out; Cos φ is the coefficient correlation between variable C, the D.

Need to prove: C, D by input signal A, B directly and matrix T multiply each other and obtain:

And

Be that A, B after quantizing are obtained through integer transform.

Referring to Fig. 9, Fig. 9 is multiple description coded device structure of residual signal of the present invention and schematic diagram.Among Fig. 9, residual signal divides 910 pairs of residual signals of receiving of module is carried out the signal branch to handling, and will finish the two-way residual signal that processing generates and send to dualistic transformation module 920; The two-way residual signal that 920 pairs of dualistic transformation modules are received carries out the dualistic transformation processing respectively and generates the two-way residual signal describing M ₁, M ₂, again with M ₁, M ₂Send to entropy coder 930, entropy coder 940 respectively; 930 couples of M that receive of entropy coder ₁Carry out encoding process, and will finish the M that is generated after the encoding process ₁Bit stream send; 940 couples of M that receive of entropy coder ₂Carry out encoding process, and will finish the M that is generated after the encoding process ₂Bit stream send.

Particularly, residual signal is represented with R (k): R (k), and k=1,2,3 ... N

Wherein, N is the number (N is generally even number) of residual signal, also is simultaneously half of audio analysis window length.

The algorithm signal of describing two of residual signal is described M with two respectively more ₁(k ₁) and M ₂(k ₂) expression:

1:M is described ₁(k ₁), k ₁=1,2,3 ... N/2

2:M is described ₂(k ₂), k ₂=1,2,3 ... N/2

Principle shown in Figure 9 can be expressed as following steps:

The parameter of step 1, loop initialization variable k=1 and other dualistic transformation.

Step 2, residual signal divide 910 pairs of residual signals of receiving of module are divided handling, and generate variables A and B and send to dualistic transformation module 920; Wherein, A=R (k), B=R (k+1).

Step 3,920 couples of A of dualistic transformation module and B carry out the dualistic transformations of describing more, obtain with to describe 1 corresponding

And it is corresponding with description 2

$M_1\left(k_1\right)=\stackrel{\&OverBar;}{C},$ K wherein ₁=(k+1)/2,

$M_2\left(k_2\right)=\stackrel{\&OverBar;}{D},$ K wherein ₂=(k+1)/2;

Certainly, also will with

Send to entropy coder 930, will

Send to entropy coder 940.

Step 4, make k=k+2, if k＜N-1 forwards step 2 to; Otherwise, forward step 5 to.

Step 5, entropy coder 930,940 be the many descriptions signal M to receiving respectively ₁(k ₁), M ₂(k ₂) encode, and the data of coding gained and the bit stream that masking threshold signal separately forms description are together sent.

Need to prove that describe in the dualistic transformation, transformation matrix plays an important role more, it is controlling the redundancy of being introduced between two descriptions.

Referring to Figure 10, Figure 10 is multiple description encoding device structure and the schematic diagram that matches with multiple description coded device among Fig. 9.Among Figure 10, the bit stream of 1001 pairs of descriptions of receiving 1 of entropy decoder carries out decoding processing, and will finish the description 1 that is generated after the decoding processing and send to antithesis contravariant change sides decoder 1003, antithesis inverse transformation center decoder 1004; The bit stream of 1002 pairs of descriptions of receiving 2 of entropy decoder carries out decoding processing, and will finish the description 2 that is generated after the decoding processing and send to antithesis contravariant change sides decoder 1005, antithesis inverse transformation center decoder 1004.Have, the antithesis contravariant residual signal that decoder, antithesis inverse transformation center decoder all handle to describe self gained of changing sides sends to the residual signal synthesis module again; By the residual signal synthesis module residual signal of receiving is synthesized processing, generate synthetic residual signal.

The position of the switch of the residual signal synthesis module output signal shown in Figure 10 can be selected automatically according to the situation that receives description, as: when receiving only a description, with regard to the decoder of changing sides with the antithesis contravariant of receiving this description this description is decoded, and be communicated with the change sides output of the residual signal synthesis module that decoder connected of this antithesis contravariant, the residual signal of finishing reconstruct that after synthesizing, is generated with output; When two descriptions all receive, with regard to receiving that the antithesis inverse transformation center decoder 1004 of these two descriptions decodes to these two descriptions, and be communicated with the change sides output of the residual signal synthesis module that decoder 1004 connected of antithesis contravariant, the residual signal of finishing reconstruct that after synthesizing, is generated with output.

Concrete decoding algorithm can be expressed as following steps:

The parameter of step 1, loop initialization variable k=1 and other dualistic transformation.

The number of the description that step 2, basis receive can be used different decoding algorithms the description that receives is decoded.Particularly, carry out following processing respectively according to the situation that receives description:

(1) if entropy decoder 1001,1002 has received two description: M altogether ₁And M ₂, so, the entropy decoder that receives foregoing description determines that respectively the description of self receiving is pairing

$\stackrel{\&OverBar;}{C}=M_1\left(k_1\right),$ K wherein ₁=(k+1)/2,

$\stackrel{\&OverBar;}{D}=M_2\left(k_2\right),$ K wherein ₂=(k+1)/2.

Then, will

With

Send to antithesis inverse transformation center decoder 1004, by antithesis inverse transformation center decoder 1004 according to describe that the dualistic transformation algorithms are obtained and

And send to residual signal synthesis module 1007; Afterwards, that receive and synthesize processings by 1007 pairs of residual signal synthesis modules, and will finish and synthesize the residual signal R (k) that finishes reconstruct that processing generated and send.

Wherein, $R\left(k\right)=\hat{A},$ $R(k+1)=\hat{B}.$

(2) if received only M ₁, so, receive M ₁Entropy decoder determine M ₁Pairing

$\stackrel{\&OverBar;}{C}=M_1\left(k_1\right),$ K wherein ₁=(k+1)/2,

Described entropy decoder will

Send to the antithesis contravariant decoder of changing sides, the antithesis contravariant decoder of changing sides goes out according to the dualistic transformation algorithm predicts of describing more And it is right With

Carry out the antithesis inverse transformation obtain and

Send it to the residual signal synthesis module again, by this residual signal synthesis module to receive and synthesize processings, and will finish and synthesize the residual signal R (k) that finishes reconstruct that processing generated and send.

Wherein, $R\left(k\right)=\hat{A},$ $R(k+1)=\hat{B}.$

(3) if received only M ₂, so, receive M ₂Entropy decoder determine M ₂Pairing

$\stackrel{\&OverBar;}{D}=M_2\left(k_2\right),$ K wherein ₂=(k+1)/2.

Described entropy decoder will

Send to the antithesis contravariant decoder of changing sides; The antithesis contravariant is changed sides decoder according to how description dualistic transformation algorithm predicts go out

And it is right

With

Carry out the antithesis inverse transformation obtain and

Send it to the residual signal synthesis module again, by this residual signal synthesis module to receive and synthesize processings, and will finish and synthesize the residual signal R (k) that finishes reconstruct that processing generated and send.

Wherein, $R\left(k\right)=\hat{A},$ $R(k+1)=\hat{B}.$

Step 3 makes k=k+2, if k＜N-1 forwards step 2 to; Otherwise, forward step 4 to.

Step 4 finishes.

So far, residual signal has just been come out by successful reconstruct.

At the multi-description coding-decoding of being realized by many descriptions scalar quantization algorithm, its operating principle is: the scalar quantizer with the many outputs of single input is carried out quantization encoding to information source, and will quantize the many descriptions of a plurality of output results of gained as this information source; Scalar quantization decoder with the single output of input more than is encoded to many description inputs, and the single output result of the gained of will decoding is as the reconstruction signal of this information source.

Be described as example with two below and set forth principle and the methods of describing scalar quantization more.

Referring to Figure 11, Figure 11 is the Audio Signal Processing schematic diagram of a preferred embodiment of the present invention.Among Figure 11, x is an original audio signal; L is through resulting index behind the quantization encoding; Adaptation a () 1103 be used for index l be matched to index to (i, j); I and j that limit decoder 1104,1106 is respectively applied for receiving decode, and produce to finish decoded signal With

Center decoder 1105 is used for i that receives and j are decoded, and produces to finish decoded signal

In the process of whole many description scalar quantizations, most important is exactly the realization of adaptation a () 1103, just how l is matched to (i, j).Illustrate below:

X is through after the encoding process of encoder 1102, the scope of the index l that is generated as shown in figure 12, for: 1～10, the quantized interval that promptly quantizes x always has ten.

At this moment, the function of adaptation a () 1103 be can use table 1 realize, promptly l i and j be matched to:

The table 1 scalar quantization tables of describing more

The essence of the matching process of being realized by table 1 is that the corresponding relation between the quantized interval of these three quantizers as shown in figure 13 more with the quantizer (the index value l behind corresponding the quantification) of the quantizer of describing two (index value i and j behind corresponding respectively the quantification) when replacing single the description.In actual applications, can retouch quantified precision more and be controlled at the redundancy of being introduced between describe more, can realize this adjustment by the mode that designs different forms usually by adjusting.Generally speaking, to describe quantified precision high more more, and the redundancy of being introduced is just high more.

In decoder one side, decoding process can have three kinds of concrete forms:

(1) when receiving i and j, center decoder 1105 can find unique l value according to table 1, and what can can't harm solves signal l.

Particularly, center decoder 1105 at first finds the corresponding respectively row and column of i and j in table 1, and then looks for this row and this to be listed in the value of intersection in the table 1, and will be worth the value l before scalar quantization is described in conduct more.

(2) when receiving only i, receive that the decoder of i can go out the value of l according to the data estimation in the table 1.The method of estimation has a variety of, as: with the mean value of the pairing each row of data of i in the table 1 value as l; Also can be with the maximum of the pairing each row of data of i in the table 1 or minimum value value as l.

(3) when receiving only j, the decode procedure of concrete decode procedure when receiving only i is identical.

Need to prove, when the design decoding algorithm, can specifically determine optimum evaluation method according to the characteristics of information source.

Encoding and decoding principle described in Figure 11 can be applied among Figure 14,15; In fact, Figure 14,15 described encoding-decoding process are just specifically used one of the encoding and decoding principle among Figure 11.

Referring to Figure 14, Figure 14 is multiple description coded device structure and the schematic diagram based on Figure 11.Among Figure 14,1401 pairs of residual signals of receiving of two description scalar quantizer carry out scalar quantization to be handled, and will finish description 1 that processing generates, describes 2 and send to entropy coder 1402, entropy coder 1403 respectively; Encoding process is carried out in 1402 pairs of descriptions of receiving of entropy coder 1, and the bit stream that will finish the description 1 that is generated after the encoding process sends; Encoding process is carried out in 1403 pairs of descriptions of receiving of entropy coder 2, and the bit stream that will finish the description 2 that is generated after the encoding process sends.

Particularly, residual signal is represented with R (k): R (k), and k=1,2,3 ... N

Wherein, N is the number of residual signal, also is simultaneously half of audio analysis window length.

The algorithm signal of describing two of residual signal is described M with two respectively more ₁(k ₁) and M ₂(k ₂) expression:

M ₁(k ₁)，k ₁＝1，2，3，......N；

M ₂(k ₂)，k ₂＝1，2，3，......N。

In actual applications, principle shown in Figure 14 can be expressed as following steps:

Step 1, loop initialization variable k:k=1.

Step 2, with residual signal R (k) as index value, search the index of coupling to M according to table 2 ₁(k ₁), M ₂(k ₂); K wherein ₁=k ₂=k.M ₁(k ₁), M ₂(k ₂) index describes signals to needing to carry out two of encoding process just, and this index to be equivalent among Figure 11 index to (i, j).

The table 2 scalar quantization tables of describing more

Step 3, make k=k+1, if k＜N forwards step 2 to; Otherwise, forward step 4 to.

Step 4, by receiving M ₁(k ₁), M ₂(k ₂) entropy coder respectively to M ₁(k ₁), M ₂(k ₂) carry out encoding process, and the bit stream that will the encode data of gained and masking threshold signal separately form description together sends.

In the process of many description scalar quantizations, the design of table 2 plays a part crucial.Difference between each data in the table 2 on the diagonal is more little, and the quantified precision of description is just high more, and the code rate of identical tonequality is just high more.

With the corresponding decode procedure of the cataloged procedure of describing among Figure 14 as shown in figure 15.Referring to Figure 15, Figure 15 is multiple description encoding device structure and the schematic diagram based on Figure 11.Among Figure 15, the bit stream of 1501 pairs of descriptions of receiving 1 of entropy decoder carries out decoding processing, and will finish the description 1 that is generated after the decoding processing and send to and describe scalar quantizations limit decoder 1503 more, describe scalar quantization center decoder 1504 more; The bit stream of 1502 pairs of descriptions of receiving 2 of entropy decoder carries out decoding processing, and will finish the description 2 that is generated after the decoding processing and send to and describe scalar quantizations limit decoder 1505 more, describe scalar quantization center decoder 1504 more.

The many descriptions scalar quantization limit decoder shown in Figure 15 and the position of the switch of describing scalar quantization center decoder output signal can be selected according to the situation that receives description automatically, as: when receiving only a description, just the scalar quantization decoding is carried out in this description, and be communicated with the residual signal of finishing reconstruct that these outputs of describing scalar quantization limit decoder are generated after decoding with output more with many descriptions scalar quantization limit decoder of receiving this description; When two descriptions all receive, just two description scalar quantization decodings are carried out in these two descriptions, and be communicated with the outputs of scalar quantization center decoder 1504 of describing more to export the residual signal of finishing reconstruct that after decoding, is generated with many descriptions scalar quantization center decoder 1504 of receiving these two descriptions.

Particularly, principle shown in Figure 15 can be expressed as following steps:

Step 1, k=1 is set.

Step 2, basis receive the situation of describing bit stream and handle respectively:

If received two description: M ₁(k ₁) and M ₂(k ₂), can uniquely find the value of residual signal R (k) according to table 2;

If received only description 1:M ₁(k ₁), find according to table 2 and to describe 1 pairing row, and with the value of absolute value minimum in this row value as residual signal R (k);

If received only description 2:M ₂(k ₂), find according to table 2 and to describe 2 pairing row, and in will being listed as the value of absolute value minimum as the value of residual signal R (k);

K wherein ₁=k ₂=k.

Step 3, make k=k+1, if k＜N forwards step 2 to; Otherwise, forward step 4 to.

Step 4, end.

So far, just successfully reconstructed residual signal.

The above is scalar quantization principle and method at two descriptions; In actual applications, when the channel packet loss greater than 25% even when higher, if still adopt the scalar quantization mode of two descriptions, describe and be interrupted even when losing simultaneously continuously, the necessary audio frequency continuity of voice communication will be had a strong impact on when two so.In this case, if can increase the number of description, problem such as can effectively avoid the tonequality brought than high packet loss to descend and audio frequency is discontinuous; As seen, comprising three or three algorithm frameworks described above also is very necessary in actual applications usually.

For this reason, be described as example with three below and set forth principle and the methods of describing scalar quantization more; Algorithm as for other type and more descriptions can design as reference.And in these arthmetic statements, the object of multi-description coding-decoding mainly is a residual signal; Certainly, in actual applications, the object of multi-description coding-decoding also can be masking threshold signal or other audio signal components.

Referring to Figure 16, Figure 16 is the Audio Signal Processing schematic diagram of another preferred embodiment of the present invention.Among Figure 16, x is an original audio signal, and l is through resulting index behind the quantization encoding; Adaptation a () 1603 be used for index l be matched to index set (h, i, j); Single limit decoder of describing is used for h, the i or the j that receive are decoded, and produces and finishes decoded signal

Or

Two descriptions limit decoder is used for the h that receives and i, h and j, i and two such descriptions of j are decoded, and produces and finishes decoded signal

The center decoder is used for whole three description signal h, the i and the j that receive are decoded, and produces and finishes decoded signal

Many descriptions scalar quantization process shown in Figure 16 is identical with many descriptions scalar quantization Principle of Process shown in Figure 11, does not repeat them here.In fact, no matter be that two descriptions, three are described or the encoding-decoding process of more descriptions, its operating principle all is identical:

At cataloged procedure, quantity according to the description that will generate, be respectively that the residual signal identical with described description quantity described and the masking threshold signal description with residual signal and masking threshold signal encoding, respectively each residual signal is described again and wherein one road masking threshold signal description close the road; The a plurality of descriptions of final generation, and all include residual signal and masking threshold signal in each description.

At decode procedure, residual signal in each description that receives is described and the shunt of masking threshold signal description, all residual signals that the shunt back is generated are described and are decoded as one road residual signal again, also all masking threshold signal descriptions that generated after the shunt are decoded as one road masking threshold signal.

Encoding and decoding principle described in Figure 16 can be applied among Figure 17,18; In fact, Figure 17,18 described encoding-decoding process are just specifically used one of the encoding and decoding principle among Figure 16.

Referring to Figure 17, Figure 17 is multiple description coded device structure and the schematic diagram based on Figure 16.Among Figure 17, three describe 1701 pairs of residual signals of receiving of scalar quantizer carries out three and describes scalar quantization and handle, and will finish description 1 that processing generates, describes 2, describes 3 and send to entropy coder 1702, entropy coder 1703, entropy coder 1704 respectively; Entropy coder 1702,1703,1704 carries out encoding process to the description of receiving respectively, and will finish the description bit stream that is generated after the encoding process and send.

Particularly, residual signal is represented with R (k): R (k), and k=1,2,3 ... N;

Wherein, N is the number of residual signal, and also is audio analysis window length half.

The algorithm signal of describing three of residual signal is described M with two respectively more ₁(k ₁), M ₂(k ₂) and M ₃(k ₃) expression:

M ₁(k ₁)，k ₁＝1，2，3，......N；

M ₂(k ₂)，k ₂＝1，2，3，......N；

M ₃(k ₃)，k ₃＝1，2，3，......N。

In actual applications, principle shown in Figure 17 can be expressed as following steps:

Step 1, loop initialization variable k:k=1.

Step 2, with residual signal R (k) as index value, and search the three-dimensional coordinate point M of coupling ₁(k ₁), M ₂(k ₂) and M ₃(k ₃); K wherein ₁=k ₂=k ₃=k.Three-dimensional coordinate point M ₁(k ₁), M ₂(k ₂) and M ₃(k ₃) need to carry out three of encoding process just and describe signals; And three-dimensional coordinate point M ₁(k ₁), M ₂(k ₂) and M ₃(k ₃) be equivalent among Figure 16 three describe signals (h, i, j).

Step 3, make k=k+1, if k＜N forwards step 2 to; Otherwise, forward step 4 to.

Step 4, by receiving M ₁(k ₁), M ₂(k ₂), M ₃(k ₃) entropy coder respectively to M ₁(k ₁), M ₂(k ₂), M ₃(k ₃) carry out encoding process, and the bit stream that will the encode data of gained and masking threshold signal separately form description together sends.

With the corresponding decode procedure of the cataloged procedure of describing among Figure 17 as shown in figure 18.Referring to Figure 18, Figure 18 is multiple description encoding device structure and the schematic diagram based on Figure 16.Among Figure 18, the bit stream of 1801 pairs of descriptions of receiving 1 of entropy decoder carries out decoding processing, and will finish the description 1 that is generated after the decoding processing and send to and more describe scalar quantizations limit decoder 1804,1806,1809 and describe scalar quantization center decoder 1807 more; The bit stream of 1802 pairs of descriptions of receiving 2 of entropy decoder carries out decoding processing, and will finish the description 2 that is generated after the decoding processing and send to and more describe scalar quantizations limit decoder 1805,1806,1808 and describe scalar quantization center decoder 1807 more; The bit stream of 1803 pairs of descriptions of receiving 3 of entropy decoder carries out decoding processing, and will finish the description 3 that is generated after the decoding processing and send to and more describe scalar quantizations limit decoder 1808,1809,1810 and describe scalar quantization center decoder 1807 more.

The many descriptions scalar quantization limit decoder shown in Figure 18 and the position of the switch of describing scalar quantization center decoder output signal can be selected according to the situation that receives description automatically, as: when receiving only one or two when describing, with regard to receiving that many descriptions scalar quantization limit decoder of described description decodes to this description, and be communicated with the residual signal of finishing reconstruct that these outputs of describing scalar quantizations limit decoder are generated after decoding with output more; When three descriptions all receive, with regard to receiving that many descriptions scalar quantization center decoder 1807 of these three descriptions decodes to these three descriptions, and be communicated with the residual signal of finishing reconstruct that the outputs of describing scalar quantizations center decoder 1807 are generated after decoding with output more.

Particularly, principle shown in Figure 180 can be expressed as following steps:

Step 1, k=1 is set.

Step 2, basis receive the situation of describing bit stream and handle respectively:

If received three description: M ₁(k ₁), M ₂(k ₂) and M ₃(k ₃), can uniquely find the value of residual signal R (k) according to the description of receiving;

If receive only M ₁(k ₁), M ₂(k ₂) and M ₃(k ₃) in two descriptions, find corresponding axle according to the description of receiving, and this gone up the value of the value of absolute value minimum as residual signal R (k);

If received only description a: M ₁(k ₁), M ₂(k ₂) or M ₃(k ₃), find corresponding plane according to the description of receiving, and the value of the value of absolute value minimum on this plane as residual signal R (k);

K wherein ₁=k ₂=k ₃=k.

Step 3, make k=k+1, if k＜N forwards step 2 to; Otherwise, forward step 4 to.

Step 4, end.

So far, just successfully reconstructed residual signal.

Need to prove: describe in the process of scalar quantization three, design shown in Figure 19 plays a part crucial when searching the value of residual signal according to description.Referring to Figure 19, Figure 19 is the three-dimensional charts of many descriptions scalar quantizer encoding and decoding of a preferred embodiment of the present invention.Among Figure 19, the reference axis of three quadratures is being represented three description: M respectively ₁(k ₁), M ₂(k ₂), M ₃(k ₃).The design principle of Figure 19 and table 2 the same substantially, difference only is to have become the straight line that becomes 60 degree angles respectively with three reference axis by the diagonal in the table 2, difference between each data on this straight line is more little, illustrate that the data that distribute around it are few more, the number of the value of the residual signal of the value correspondence of each description is just few more, the quantified precision of describing is just high more, and the code rate of identical tonequality is also just high more.

In actual applications, above-described all multi-description coding-decoding algorithms all can guarantee: do not have in ideal and utilize two descriptions that receive normally audio signal to be decoded under the situation of packet loss, also can in certain packet loss scope, utilize a description that receives to go the description of estimating that another one is lost, and then utilize the correlation between describing to recover and reconstructed audio signal.Yet when packet loss continued to increase, two descriptions may be lost simultaneously, and at this moment the multiple description encoding device also is difficult to carry out preferably audio decoder and audio reconstruction.Therefore, in order when packet loss is serious, to improve tonequality, can further improve and perfect the multi-description coding-decoding algorithm; For this reason, proposition is several based on multiple description coded bag-losing hide Processing Algorithm below:

1. separate in the multiple description coded algorithm in aforesaid odd even owing to strange describe that to describe with idol be fully independently two descriptions, between them without any correlation; Therefore, when a description is therein lost, can't utilize another description that receives to estimate the description of losing.For this reason, can adopt following bag-losing hide Processing Algorithm:

The description that the description that normally receives with former frame replaces present frame to lose; Perhaps, multiply by a decay factor (can be set to: between 0.5 to 0.9) with the normal description that receives of former frame, with take advantage of the value description that replaces present frame to lose; Perhaps, the description of losing with the linear interpolation replacement present frame of the normal description that receives of present frame and 0.

2. when seriously causing two descriptions to be lost simultaneously because of packet loss, adopt the normal description that receives of former frame to multiply by a decay factor (can be set to: between 0.5 to 0.9) respectively, with take advantage of the value description that replaces present frame to lose respectively, and estimate present frame audio frequency parameter or audio signal with this.

3. when seriously causing certain description to be lost continuously because of packet loss, adopt the algorithm that successively decreases frame by frame, utilize the normal description that receives of last frame to multiply by a decay factor (can be set to: between 0.5 to 0.9) frame by frame, with take advantage of the value description that replaces present frame to lose respectively.

More than among each figure, as the entropy coding/decoding device appears, in fact can only use an entropy coding/decoding device and coding/decoding is carried out in description handle, rather than use two even a plurality of entropy coding/decoding device in the image pattern like that; And, also the entropy coding/decoding device can be replaced with the coder/decoder of other type; Have again, before described residual signal is handled, can be further to its round, quantification and/or encoding process.

By the above as can be seen, acoustic signal processing method provided by the present invention, system and audio signal transmitting/receiving device all can effectively improve the audio quality in the communication process, improve user satisfaction.

Claims (38)

1. acoustic signal processing method is characterized in that:

At the audio signal transmitting terminal, to describe and multichannel masking threshold signal description for the multichannel residual signal at the residual signal and the masking threshold signal encoding of Audio Signal Processing gained respectively, respectively with road He Lu in each road residual signal description and the multichannel masking threshold signal description, generate the description that multichannel all includes residual signal description and masking threshold signal description again;

At the audio signal receiving terminal, all that receive are included all that residual signal is described and the description of masking threshold signal description in each residual signal that is comprised in describing describe and the shunt of masking threshold signal description, all residual signals that the shunt back is generated are described and are decoded as one road residual signal again, also all masking threshold signal descriptions that the shunt back is generated are decoded as one road masking threshold signal, residual signal and the masking threshold signal that obtains after the decoding carried out parameter reconstruct and the synthetic processing of time-frequency, generate the audio signal of reconstruct.

2. the method for claim 1 is characterized in that, the method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the odd even separating treatment, and to finishing multichannel residual signal description the carrying out entropy coding processing that this processing generates;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the odd even separating treatment, and carry out entropy coding and handle finishing multichannel masking threshold signal description that this processing generates.

3. method as claimed in claim 2 is characterized in that, all residual signals is described the method that is decoded as one road residual signal be:

All residual signals are described carry out the entropy decoding processing, and describe and carry out that odd even is synthetic to be handled finishing all residual signals after the decoding processing, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and carrying out to finishing all masking threshold signal descriptions after the decoding processing that odd even is synthetic to be handled, all the masking threshold signal descriptions after will carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

4. the method for claim 1 is characterized in that, the method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the signal branch to handling, and describe, again the multichannel residual signal that generates is described and carried out entropy coding and handle finishing the multichannel residual signal that multichannel residual signal that processing generates carries out dualistic transformation respectively and generate corresponding way;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the signal branch to handling, and, again the multichannel masking threshold signal description that generates is carried out entropy coding and handle finishing the multichannel masking threshold signal description that multichannel masking threshold signal that processing generates carries out dualistic transformation respectively and generates corresponding way.

5. method as claimed in claim 4 is characterized in that, all residual signals is described the method that is decoded as one road residual signal be:

All residual signals are described carry out the entropy decoding processing, and describe and carry out the antithesis inverse transformation and handle finishing all residual signals after the decoding processing, all residual signals after will finishing the antithesis inverse transformation again and handling are described and are synthesized one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and carry out the antithesis inverse transformation and handle finishing all masking threshold signal descriptions after the decoding processing, all the masking threshold signal descriptions that will finish again after the antithesis inverse transformation is handled synthesize one road masking threshold signal.

6. as each described method of claim 2 to 5, it is characterized in that described multichannel is described as two-way and describes.

7. the method for claim 1 is characterized in that, the method that residual signal is encoded to the description of multichannel residual signal is:

Described residual signal is carried out the scalar quantization encoding process of describing more, and to finishing multichannel residual signal description the carrying out entropy coding processing that processing generates;

With the masking threshold signal encoding is that the method for multichannel masking threshold signal description is:

Described masking threshold signal is carried out the scalar quantization encoding process of describing more, and carry out entropy coding and handle finishing multichannel masking threshold signal description that processing generates.

8. method as claimed in claim 7 is characterized in that, all residual signals is described the method that is decoded as one road residual signal be:

The entropy decoding processing is carried out in all residual signals descriptions, and describe and carry out the scalar quantization decoding processing of describing more finishing all residual signals after the decoding processing, will carry out all residual signals of describing after the scalar quantization decoding processing more and describe and be decoded as one road residual signal;

The method that all masking threshold signal descriptions is decoded as one road masking threshold signal is:

All masking threshold signal descriptions are carried out the entropy decoding processing, and all masking threshold signal descriptions of finishing after the decoding processing are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

9. as claim 7 or 8 described methods, it is characterized in that described multichannel is described as the above description of two-way.

10. the method for claim 1 is characterized in that, described residual signal is to carry out after time frequency analysis, the residual signal analysis resulting to original audio signal;

11. method as claimed in claim 10 is characterized in that:

Described time frequency analysis method is: described original audio signal is comprised the processing of revising discrete cosine transform MDCT, obtain the time-frequency conversion parameter;

The method of described residual signal analysis is: remove irrelevant information of the sense of hearing or irrelevant degree in the described time-frequency conversion parameter.

12. the method for claim 1 is characterized in that, described masking threshold signal is to carry out after the psychoacoustic model analysis resulting to original audio signal.

13. the method for claim 1 is characterized in that, further description of multichannel residual signal and the multichannel masking threshold signal description that transmitting terminal generated is carried out undistorted coding and audio pack processing.

14. method as claimed in claim 13 is characterized in that, before receiving terminal carries out shunt, further carries out audio pack and unpacks and undistorted decoding processing.

15. an audio signal processing is characterized in that, this system comprises the multiple description coded device that is positioned at the audio signal transmitting terminal, is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal; Also comprise the multiple description encoding device that is positioned at the audio signal receiving terminal, form by the residual signal multiple description encoding device that all links to each other, masking threshold signal multiple description encoding device with splitter;

Wherein, residual signal/multiple description coded the device of masking threshold signal, the residual signal that is used for receiving is encoded to the multichannel residual signal and describes, and is multichannel masking threshold signal description with the masking threshold signal encoding that receives, and each road description that the back of will encoding is generated sends to mixer;

Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the description that multichannel all includes residual signal description and masking threshold signal description;

Splitter, the residual signal that all that are used for receiving include all that residual signal is described and each of the description of masking threshold signal description comprised in describing describe and the masking threshold signal description along separate routes, and all residual signals descriptions and the masking threshold signal description that are generated after will be along separate routes send to residual signal/masking threshold signal multiple description encoding device;

Residual signal/masking threshold signal multiple description encoding device, all residual signals descriptions that are used for receiving are decoded as one road residual signal, and all masking threshold signal descriptions that receive are decoded as one road masking threshold signal;

Described multiple description encoding device further links to each other with the parameter reconstruct module that is connected with the time-frequency synthesis module;

Wherein, the parameter reconstruct module is used to receive residual signal and the masking threshold signal that the multiple description encoding device is generated, and the signal of receiving is carried out parameter reconstruct handle, and will finish the time-frequency conversion parameter that is generated after handling again and send to the time-frequency synthesis module;

The time-frequency synthesis module is used for the time-frequency conversion parameter of receiving is carried out the synthetic processing of time-frequency, generates the audio signal of reconstruct.

16. system as claimed in claim 15 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal comprises continuous odd even separation module, entropy coder;

Wherein, the odd even separation module, be used for described residual signal is carried out the odd even separating treatment, and will finish multichannel residual signal that this processing generates and describe and send to described entropy coder, described masking threshold signal is carried out the odd even separating treatment, and will finish the multichannel masking threshold signal description that this processing generates and send to described entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

17. system as claimed in claim 16 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, odd even synthesis module;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals descriptions that will finish after the decoding processing send to the odd even synthesis module, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the odd even synthesis module;

The odd even synthesis module, be used for the synthetic processing of odd even is carried out in all residual signals descriptions of receiving, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal, all masking threshold signal descriptions of receiving are carried out odd even is synthetic to be handled, all the masking threshold signal descriptions after will carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

18. system as claimed in claim 15 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal comprises that the residual signal/masking threshold signal branch that links to each other successively is to module, dualistic transformation module and entropy coder;

Wherein, residual signal/masking threshold signal divides module, be used for described residual signal is carried out the signal branch to handling, and will finish the multichannel residual signal that processing generates and send to the dualistic transformation modules of describing more, described masking threshold signal is carried out the signal branch to handling, and will finish the multichannel masking threshold signal that processing generates and send to the dualistic transformation modules of describing more;

The dualistic transformation modules of describing more, be used for the multichannel residual signal that the multichannel residual signal of receiving carries out dualistic transformation respectively and generates corresponding way is described, again the multichannel residual signal that is generated is described and sent to entropy coder, the multichannel masking threshold signal of receiving is carried out dualistic transformation respectively and generates the multichannel masking threshold signal description of corresponding way, again the multichannel masking threshold signal description that is generated is sent to entropy coder;

Entropy coder is used for receiving that the multichannel residual signal is described and multichannel masking threshold signal description carries out the entropy coding processing.

19. system as claimed in claim 18 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises entropy decoder, antithesis inverse transformation decoder and the residual signal/masking threshold signal synthesizing module that links to each other successively;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and will finish all residual signals that generated after the decoding processing and describe to send to describe antithesis inverse transformation decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the antithesis inverse transformation decoders of describing more;

The antithesis inverse transformation decoders of describing more, being used for that the antithesis inverse transformation is carried out in all residual signals descriptions of receiving handles, and all residual signals descriptions that will finish after the processing of antithesis inverse transformation send to residual signal/masking threshold signal synthesizing module, all masking threshold signal descriptions of receiving are carried out the antithesis inverse transformation handle, and all the masking threshold signal descriptions that will finish after the antithesis inverse transformation is handled send to residual signal/masking threshold signal synthesizing module;

Residual signal/masking threshold signal synthesizing module, all residual signals descriptions that are used for receiving synthesize one road residual signal, and all masking threshold signal descriptions that receive are synthesized one road masking threshold signal.

20. system as claimed in claim 15 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal comprises continuous many descriptions scalar quantizer, entropy coder;

Wherein, the scalar quantizer of describing more, be used for described residual signal is carried out the scalar quantization encoding process of describing more, and will finish multichannel residual signal that processing generates and describe and send to entropy coder, described masking threshold signal is carried out describing the scalar quantization encoding process more, and will finish the multichannel masking threshold signal description that processing generates and send to entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

21. system as claimed in claim 20 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, describes the scalar quantization decoder more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals that will finish after the decoding processing describe to send to describe the scalar quantization decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the scalar quantization decoders of describing more;

The scalar quantization decoders of describing more, be used for all residual signals of receiving are described to carry out describing the scalar quantization decoding processing more, all residual signals after the scalar quantization of description more than will the carrying out decoding processing are described and are decoded as one road residual signal, all masking threshold signal descriptions of receiving are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

22. system as claimed in claim 20 is characterized in that:

The described scalar quantizer of describing is to handle two scalar quantizer of describing described above more more, and the described scalar quantization decoders of describing are to handle two scalar quantization decoders of describing described above more more.

23. system as claimed in claim 15 is characterized in that, described multiple description coded device further links to each other with the residual signal analysis module that is connected with the time frequency analysis module, also links to each other with psycho-acoustic module;

Wherein, the time frequency analysis module is used for original audio signal is carried out time frequency analysis, and will finish and analyze the time-frequency conversion parameter that the back generated and send to the residual signal analysis module;

The residual signal analysis module is used for time-frequency conversion parameter and the masking threshold signal received are carried out the residual signal analysis, and the residual signal that will be generated after will analyzing sends to described multiple description coded device;

Psycho-acoustic module is used for original audio signal is carried out the psychoacoustic model analysis, and the masking threshold signal that is generated after analyzing is sent to described multiple description coded device and residual signal analysis module.

24., it is characterized in that described multiple description coded device further links to each other with undistorted coding and audio pack processing module as each described system of claim 15 to 23;

Undistorted coding and audio pack processing module are used to receive the description that multiple description coded device generates, and undistorted coding and audio pack processing are carried out in the description of receiving.

25. system as claimed in claim 24 is characterized in that, described multiple description encoding device further unpacks with audio pack and undistorted decoder module links to each other;

Audio pack unpacks and undistorted decoder module, is used for that audio pack is carried out in the multichannel description of receiving and unpacks and undistorted decoding processing.

26. an audio signal dispensing device is characterized in that, this device is made up of the multiple description coded device of the residual signal that all links to each other with mixer, the multiple description coded device of masking threshold signal;

Residual signal/multiple description coded the device of masking threshold signal, the residual signal that is used for receiving is encoded to the multichannel residual signal and describes, with the masking threshold signal encoding that receives is multichannel masking threshold signal description, and each road description that the back of will encoding is generated sends to mixer;

Mixer, each the road residual signal that is used for will receiving is respectively described road He Lu with all masking threshold signal descriptions, generates the description that multichannel all includes residual signal description and masking threshold signal description.

27. device as claimed in claim 26 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal comprises continuous odd even separation module, entropy coder;

Wherein, the odd even separation module, be used for described residual signal is carried out the odd even separating treatment, and will finish multichannel residual signal that this processing generates and describe and send to described entropy coder, described masking threshold signal is carried out the odd even separating treatment, and will finish the multichannel masking threshold signal description that this processing generates and send to described entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

28. device as claimed in claim 26 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal, comprise successively residual signal/masking threshold signal branch of linking to each other to module, dualistic transformation module and entropy coder are described more;

Wherein, residual signal/masking threshold signal divides module, be used for described residual signal is carried out the signal branch to handling, and will finish the multichannel residual signal that processing generates and send to the dualistic transformation modules of describing more, described masking threshold signal is carried out the signal branch to handling, and will finish the multichannel masking threshold signal that processing generates and send to the dualistic transformation modules of describing more;

The dualistic transformation modules of describing more, be used for the multichannel residual signal that the multichannel residual signal of receiving carries out dualistic transformation respectively and generates corresponding way is described, again the multichannel residual signal that is generated is described and sent to entropy coder, the multichannel masking threshold signal of receiving is carried out dualistic transformation respectively and generates the multichannel masking threshold signal description of corresponding way, again the multichannel masking threshold signal description that is generated is sent to entropy coder;

Entropy coder is used for receiving that the multichannel residual signal is described and multichannel masking threshold signal description carries out the entropy coding processing.

29. device as claimed in claim 26 is characterized in that, the described residual signal/multiple description coded device of masking threshold signal comprises continuous many descriptions scalar quantizer, entropy coder;

Wherein, the scalar quantizer of describing more, be used for described residual signal is carried out the scalar quantization encoding process of describing more, and will finish multichannel residual signal that processing generates and describe and send to entropy coder, described masking threshold signal is carried out describing the scalar quantization encoding process more, and will finish the multichannel masking threshold signal description that processing generates and send to entropy coder;

Entropy coder is used for description of multichannel residual signal and the multichannel masking threshold signal description received are carried out the entropy coding processing.

30. device as claimed in claim 29 is characterized in that, the described scalar quantizer of describing are to handle two scalar quantizer of describing described above more more.

31. device as claimed in claim 26 is characterized in that, described multiple description coded device further links to each other with the residual signal analysis module that is connected with the time frequency analysis module, also links to each other with psycho-acoustic module;

Wherein, the time frequency analysis module is used for original audio signal is carried out time frequency analysis, and will finish and analyze the time-frequency conversion parameter that the back generated and send to the residual signal analysis module;

The residual signal analysis module is used for time-frequency conversion parameter and the masking threshold signal received are carried out the residual signal analysis, and the residual signal that will be generated after will analyzing sends to described multiple description coded device;

Psycho-acoustic module is used for original audio signal is carried out the psychoacoustic model analysis, and the masking threshold signal that is generated after analyzing is sent to described multiple description coded device and residual signal analysis module.

32., it is characterized in that described multiple description coded device further links to each other with undistorted coding and audio pack processing module as each described device of claim 26 to 31;

Undistorted coding and audio pack processing module are used to receive the multichannel description that multiple description coded device generates, and undistorted coding and audio pack processing are carried out in the multichannel description of receiving.

33. an audio signal receiving system is characterized in that, this device is made up of the residual signal multiple description encoding device that all links to each other with splitter, masking threshold signal multiple description encoding device;

Wherein, splitter, the residual signal that all that are used for receiving include all that residual signal is described and each of the description of masking threshold signal description comprised in describing describe and the masking threshold signal description along separate routes, and all residual signals descriptions and the masking threshold signal description that are generated after will be along separate routes send to residual signal/masking threshold signal multiple description encoding device;

Residual signal/masking threshold signal multiple description encoding device, all residual signals descriptions that are used for receiving are decoded as one road residual signal, and all masking threshold signal descriptions that receive are decoded as one road masking threshold signal;

Described multiple description encoding device further links to each other with the parameter reconstruct module that is connected with the time-frequency synthesis module;

Wherein, the parameter reconstruct module is used to receive residual signal and the masking threshold signal that the multiple description encoding device is generated, and the signal of receiving is carried out parameter reconstruct handle, and will finish the time-frequency conversion parameter that is generated after handling again and send to the time-frequency synthesis module;

The time-frequency synthesis module is used for the time-frequency conversion parameter of receiving is carried out the synthetic processing of time-frequency, generates the audio signal of reconstruct.

34. device as claimed in claim 33 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, odd even synthesis module;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals descriptions that will finish after the decoding processing send to the odd even synthesis module, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the odd even synthesis module;

The odd even synthesis module, be used for the synthetic processing of odd even is carried out in all residual signals descriptions of receiving, all residual signals after will carrying out that odd even is synthetic and handling are described and are synthesized one road residual signal, all masking threshold signal descriptions of receiving are carried out odd even is synthetic to be handled, all masking threshold signal descriptions of respecting after carrying out that odd even is synthetic and handling synthesize one road masking threshold signal.

35. device as claimed in claim 33 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises continuous successively entropy decoder, describes antithesis inverse transformation decoder and residual signal/masking threshold signal synthesizing module more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and will finish all residual signals that generated after the decoding processing and describe to send to describe antithesis inverse transformation decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the antithesis inverse transformation decoders of describing more;

The antithesis inverse transformation decoders of describing more, being used for that the antithesis inverse transformation is carried out in all residual signals descriptions of receiving handles, and all residual signals descriptions that will finish after the processing of antithesis inverse transformation send to residual signal/masking threshold signal synthesizing module, all masking threshold signal descriptions of receiving are carried out the antithesis inverse transformation handle, and all the masking threshold signal descriptions that will finish after the antithesis inverse transformation is handled send to residual signal/masking threshold signal synthesizing module;

Residual signal/masking threshold signal synthesizing module, all residual signals descriptions that are used for receiving synthesize one road residual signal, and all masking threshold signal descriptions that receive are synthesized one road masking threshold signal.

36. device as claimed in claim 33 is characterized in that, described residual signal/masking threshold signal multiple description encoding device comprises continuous entropy decoder, describes the scalar quantization decoder more;

Wherein, entropy decoder, be used for the entropy decoding processing is carried out in all residual signal descriptions, and all residual signals that will finish after the decoding processing describe to send to describe the scalar quantization decoders more, all masking threshold signal descriptions are carried out the entropy decoding processing, and all the masking threshold signal descriptions that will finish after the decoding processing send to the scalar quantization decoders of describing more;

The scalar quantization decoders of describing more, be used for all residual signals of receiving are described to carry out describing the scalar quantization decoding processing more, all residual signals after the scalar quantization of description more than will the carrying out decoding processing are described and are decoded as one road residual signal, all masking threshold signal descriptions of receiving are carried out the scalar quantization decoding processing of describing more, will carry out all masking threshold signal descriptions of describing after the scalar quantization decoding processing more and be decoded as one road masking threshold signal.

37. device as claimed in claim 36 is characterized in that, the described scalar quantization decoders of describing are to handle two scalar quantization decoders of describing described above more more.

38. device as claimed in claim 33 is characterized in that, described multiple description encoding device further unpacks with audio pack and undistorted decoder module links to each other;

Audio pack unpacks and undistorted decoder module, is used for that audio pack is carried out in the description of receiving and unpacks and undistorted decoding processing.

Images