CN102760441B - Background noise coding/decoding device and method as well as communication equipment - Google Patents

Abstract

Embodiments of the present invention provide a background noise coding/decoding device, method and communication equipment, so as to flexibly adjust the SID frame coding method and the discontinuous transmission mechanism according to the background noise of the communication environment. The technical scheme of the present invention can flexibly adopt different SID frame coding methods and discontinuous transmission mechanisms according to different types of background noise, and adapt to the coding accuracy and time resolution required by different types of background noise under the condition of rational use of actual resources of the communication system To achieve the purpose of providing various types of background noise at the decoding end with high quality under the premise of rational use of channel resources.

Description

A kind of background noise encoding/decoding device, method and communication facilities

Technical field

The present invention relates to voice communication technology, particularly a kind of background noise encoding/decoding device, method and communication facilities.

Background technology

In voice communication course, the voice signal being transmitted can be divided into audible signal and without acoustical signal two classes, wherein: audible signal is the voice signal sending when efficient voice being detected, mainly comprises communication party's voice and the information such as ground unrest at that time; The ambient noise signal sending in the gap that efficient voice do not detected without acoustical signal, mainly comprises diversity of settings noise, such as white noise, background brouhaha and quiet etc.Concerning communicating pair, audible signal is useful signal, is the carrier of the main communication information, and is garbage signal without acoustical signal, does not comprise the main communication information.

In voice communication course, what mainly pay close attention to due to communicating pair is useful audible signal, therefore considers the pressure of bandwidth, does not wish to transmit useless without acoustical signal, to reduce transmission bandwidth, increases power system capacity.If but only pass audible signal, can cause the discontinuous of ground unrest, make a side who listens to feel very uncomfortable, especially more obvious in the situation that ground unrest is stronger, even can make voice indigestion, so can only reduce as far as possible transmit useless without acoustical signal, on the basis that guarantees user's susceptibility, reduce bandwidth as far as possible, CNG (Comfort Noise generation has arisen at the historic moment for this reason, comfort noise generates) technology, in CNG technology, transmitting terminal adopts silence compression coding techniques coding background noise frames in noiseless signal phase, and adopt DTX (Discontinuous Transmission, discontinuous transmission) mode is transmitted, receiving end adopts the mode of interpolation to produce continuous ambient noise signal, to increase the comfort of listener.

As shown in Figure 1, for adopting the communication system principle schematic diagram of silence compression coding techniques, wherein, transmitting terminal comprises:

VAD (Voice Activity Detector, speech detector), original signal is inputted VAD after certain pre-service, VAD is conventionally according to the eigenwert of each frame signal, signal to noise ratio (S/N ratio) for example, in detection signal, whether comprise efficient voice, and according to testing result, signal is sent into different scramblers and encode;

Speech coder, for the audible signal that comprises efficient voice being encoded to speech frame output, speech frame is transferred to receiving end by communication channel;

Non-voice scrambler, adopting silence compression technology is background noise frames by the noiseless Signal coding that does not comprise efficient voice;

Background noise frames transmitter, the DTX mode of adopt setting by background noise frames by traffic channel to receiving end, for example every several frames, send a background noise frames, or determine whether to send a background noise frames according to the characteristic parameter situation of change of ground unrest;

Accordingly, receiving end comprises:

Voice decoder, for the speech frame of decoding and receiving from communication channel, and exports after being reconstructed into audible signal;

Non-voice demoder, for the background noise frames of decoding and receiving from communication channel;

Reconstructor, for adopting the mode of interpolation that decoded data is reconstructed into continuous ambient noise signal, to increase the comfort of listener.

The non-voice scrambler of transmitting terminal and background noise frames transmitter form DTX unit, and the non-voice demoder of receiving end and reconstructor form CNG unit, and silence compression technology is by being positioned at VAD and the DTX of transmitting terminal, and the CNG realization that is positioned at receiving end.

With CELP (Code-Excited Linear Prediction, the Code Excited Linear Prediction) speech coding technology commonly used, be example, CELP is a kind of high efficiency speech coding technology.Owing to having adopted perceptual weighting, analyze the technology such as synthetic, vector quantization and rear filtering, CELP can provide high-quality synthetic speech in middle low rate.In CELP speech coding technology, at transmitting terminal, carry out speech signal analysis to obtain the characteristic parameter of voice, the characteristic parameter of voice is encoded in code stream and is sent; Receiving end carries out phonetic synthesis, reconstructed speech signal according to the characteristic parameter in code stream.Voice signal s (n) is divided into the short time period, is called frame, and the speech analysis of transmitting terminal and the phonetic synthesis of receiving end are all carried out frame by frame.

At transmitting terminal, original signal s (n) can regard the output that a pumping signal e (n) encourages a wave filter v (n) to produce as, be s (n)=e (n) * v (n), receiving end is used identical wave filter to synthesize voice, is called composite filter.Therefore the characteristic parameter transmitting in the encoding code stream of voice signal comprises composite filter parameter and excitation parameters, wherein: composite filter parameter is mainly line spectral frequencies LSF (Line Spectrum Frequency) quantization parameter, and pumping signal parameter comprises: adaptive codebook parameter (comprising pitch delay parameter and fundamental tone gain parameter) and fixed codebook parameters (position and symbol and the fixed codebook gain parameter that comprise non-zero pulses in fixed codebook).

Different from speech coding parameters, what ground unrest coding parameter was described is ground unrest characteristic, because the pumping signal of ground unrest can be thought simple noise random series, and these sequences all can simply produce with random noise generation device at encoding and decoding end, then with energy parameter, control the amplitude of these sequences, just can produce final pumping signal, therefore pumping signal characteristic parameter can simply represent with energy parameter, and do not need to further describe with other some characteristic parameters, so different from speech frame in ground unrest encoding code stream, be, its excitation parameters is the energy parameter of current background noise frame, and identical with speech frame in ground unrest encoding code stream, the composite filter parameter in encoding code stream is also line spectral frequencies LSF quantization parameter, is concrete quantization method difference to some extent.By above analysis, can think that to the coded system of ground unrest be exactly a kind of simple voice coding modes in essence.

Through DTX compressed encoding and by the background noise frames of discontinuous transmission, be commonly referred to as SID (Silence Descriptor, silence description frames) frame.In current various speech coding systems, each intrasystem ambient noise signal is all used SID frame format and discontinuous transmission mechanism.

Wherein, G.729B silence compression scheme is a kind of representative silence compression technology, G.729B based on CELP algorithm model, so its ground unrest parameter of transmitting also obtains based on CELP model, comprises for describing composite filter parameter and the excitation parameters of ground unrest.At coding side, for example the court verdict of VAD is 0 o'clock, represent that current signal is noise, through certain pretreated original signal, be admitted to ground unrest code device, at ground unrest code device, ground unrest parameter is extracted and quantized, form SID frame, then ground unrest code device determines whether sending a frame SID frame according to the parameter situation of change in SID frame, for example, when the filter parameter in SID frame and energy parameter change greatly, sends a frame SID frame.G.729B, in, the interval of two adjacent S ID frames is at least 2 frames.In decoding end, if enter the ground unrest decoding and reconstituting stage, in ground unrest decoding device, according to the SID frame receiving, synthesize the comfort noise of description encoding end ground unrest characteristic.G.729B the bit number that in SID frame is each parametric distribution is as shown in table 1 below:

Table 1.

Bit distributes (from a high position to low level)	Parametric description
		S1	Quantitative prediction device index
S2-S6	One-level LSF quantisation vector index
		S7-S10	Secondary LSF quantisation vector index
S11-S15	Gain (energy)

AMR (Adaptive Multi-Rate speech codec, self-adaptation multi code Rate of Chinese character speech coder) and AMR-WB (Adaptive Multi-Rate Wide Band speech codec, wideband adaptive multi code Rate of Chinese character speech coder) DTX in silence compression scheme adopts fixing transmit mechanism, no matter how the characteristic of ground unrest changes, and every 8 frames send a SID frame.The bit number that in the SID frame of AMR is each parametric distribution is as shown in table 2 below, is that the bit number of each parametric distribution is as shown in table 3 below in the SID frame of AMR-WB.

Table 2.

Bit distributes (from a high position to low level)	Parametric description
		S1-S3	Quantitative prediction device index
S4-S11	One-level LSF quantisation vector index
		S12-S20	Secondary LSF quantisation vector index
S21-S29	Three grades of LSF quantisation vector index
		S30-S35	Logarithm frame energy index

Table 3.

Bit distributes (from a high position to low level)	Parametric description
		S1-S6	One-level LSF quantisation vector index
S7-S12	Secondary LSF quantisation vector index

S13-S18	Three grades of LSF quantisation vector index
		S19-S23	Level Four LSF quantisation vector index
S24-S28	Pyatyi LSF quantisation vector index
		S29-S34	Logarithm frame energy index
S35	Shake sign

Existing ground unrest coded system can be carried out simple code to ground unrest, and synthesizes simple comfortable ground unrest in decoding end, can better meet to a certain extent the requirement of answering quality.Yet along with the development of the communication technology, voice coding quality is more and more higher, corresponding, the coding quality of the ground unrest in sound stage is also more and more higher, in this case, the ground unrest when making voice communication is more continuous, also needs to improve the coding quality of silent period ground unrest.Existing comparatively single SID frame coding mode and discontinuous transmission mechanism obviously cannot meet the specific requirement of multiple communication environment.

Summary of the invention

The embodiment of the present invention provides a kind of background noise encoding/decoding device, method and communication facilities, to realize according to the ground unrest situation of communication environment, adjusts flexibly SID frame coding mode and discontinuous transmission mechanism.

A ground unrest code device, comprising:

Background noise analysis unit, at least one signal characteristic parameter value of background extraction noise signal;

Ground unrest taxon, for determining ground unrest type according to described at least one signal characteristic parameter value;

Coding switch unit and at least two discontinuous transmission unit, described coding switch unit is for inputting corresponding discontinuous transmission unit according to described ground unrest type by ambient noise signal, each discontinuous transmission unit is for being encoded to by ambient noise signal silence description frames the discontinuous transmission that comprises discontinuous transmission unit indication information, and wherein the silence description frames coded system of each discontinuous transmission unit and/or silence description frames discontinuous transmission mechanism are not identical.

A ground unrest decoding device, comprising:

The first recognition unit, for identifying the discontinuous transmission unit indication information of silence description frames;

Decoding switch unit and at least two comfort noise generation units, described decoding switch unit is used for according to the recognition result of described recognition unit, described silence description frames is inputted to corresponding comfort noise generation unit, each comfort noise generation unit is according to the corresponding decoding of the coded system of the silence description frames receiving silence description frames, and reconstruct ambient noise signal.

A communication facilities, comprising:

The first speech detector, for detection of whether comprising efficient voice classification output in the signal gathering;

Speech coder, for being speech frame continuous wave output by the Signal coding that comprises efficient voice;

Also comprise ground unrest code device, this ground unrest code device specifically comprises:

Background noise analysis unit, for obtaining at least one signal characteristic parameter value of the ambient noise signal that does not comprise efficient voice;

Ground unrest taxon, for determining ground unrest type according to described at least one signal characteristic parameter value;

Coding switch unit and at least two discontinuous transmission unit, described coding switch unit is for inputting corresponding discontinuous transmission unit according to described ground unrest type by ambient noise signal, each discontinuous transmission unit is for being encoded to by ambient noise signal silence description frames the discontinuous transmission that comprises discontinuous transmission unit indication information, and wherein the silence description frames coded system of each discontinuous transmission unit and/or silence description frames discontinuous transmission mechanism are not identical.

A communication facilities, comprises ground unrest decoding device, and described ground unrest decoding device comprises:

The first recognition unit, for identifying the discontinuous transmission unit indication information of silence description frames;

Decoding switch unit and at least two comfort noise generation units, described decoding switch unit is used for according to the recognition result of described recognition unit, described silence description frames is inputted to corresponding comfort noise generation unit, each comfort noise generation unit is according to the corresponding decoding of the coded system of the silence description frames receiving silence description frames, and reconstruct ambient noise signal.

A ground unrest coding method, comprising:

At least one signal characteristic parameter value of background extraction noise signal;

According to described at least one signal characteristic parameter value, determine ground unrest type;

According to a kind of discontinuous transmission mechanism of described ground unrest type selecting, ambient noise signal is encoded to silence description frames the discontinuous transmission that comprises discontinuous transmission mechanism indication information, wherein the silence description frames coded system in each discontinuous transmission mechanism and/or the discontinuous send mode of silence description frames are not identical.

A coding/decoding method, comprising:

Discontinuous transmission mechanism indication information in identification silence description frames, wherein the silence description frames coded system in each discontinuous transmission mechanism is not identical;

According to described discontinuous transmission mechanism indication information, adopt corresponding decoding process decoding silence description frames, and be reconstructed into ambient noise signal.

Application technical solution of the present invention, can be according to the different SID frame coding method of the dissimilar flexible employing of ground unrest and discontinuous transmission mechanism, in the situation that rationally utilizing communication system real resource, be adaptive to the desired encoding precision of dissimilar ground unrest and time sense, reach under the prerequisite of rationally utilizing channel resource in the high-quality object that all kinds ground unrest is provided of decoding end.

Accompanying drawing explanation

Fig. 1 is the communication system principle schematic diagram of existing employing silence compression coding techniques;

Fig. 2, Fig. 3, Fig. 5 and Fig. 6 are respectively a kind of ground unrest code device primary structure schematic diagram that the embodiment of the present invention provides;

The switch mode principle schematic that Fig. 4 provides for the embodiment of the present invention;

Fig. 7, Fig. 8 and Fig. 9 are respectively a kind of communication facilities primary structure schematic diagram that the embodiment of the present invention provides.

Embodiment

Further analyze existing ground unrest coding, in each coding/decoding system, all adopt single ground unrest frame coding mode, cannot describe targetedly the concrete ambient noise signal of the communication environment of various characteristics, thereby meet the multifarious demand of communication environment; Single discontinuous transmission mechanism also cannot be transmitted background noise frames targetedly to the background noise frames of different qualities, so that cannot produce targetedly good comfort noise in decoding end; And the unicity of discontinuous transmission mechanism also cannot be adaptive to the actual state of communication system, for example communication bandwidth, can not rationally be applied down to the resource of communication system.

Therefore, a kind of encoding and decoding technique scheme that the embodiment of the present invention provides, can according to ground unrest characteristic, carry out respective handling targetedly, at coding side, ground unrest is analyzed, according to the characteristic of ground unrest, adopt different DTX scheme codings to send SID frame, and adopt corresponding CNG scheme reconstructed background noise signal at receiving end.

As shown in Figure 2, be a kind of ground unrest code device 20 that the embodiment of the present invention provides, this ground unrest code device 20 is for silence compression navamander, and primary structure comprises:

At least two DTX unit 21, comprise 211,DTX unit, DTX unit 2 12......DTX unit 21n, each DTX unit 21 is not continuous wave output for ambient noise signal being encoded to SID frame according to different DTX mechanism, wherein, difference between different DTX mechanism comprises: SID frame coding mode is not identical, SID frame discontinuous transmission mechanism is not identical, or the two is not identical;

Background noise analysis unit 22, for calculating at least one signal characteristic parameter value according to ambient noise signal;

Ground unrest taxon 23, for determining ground unrest type according to described at least one signal characteristic parameter value;

Coding switch unit 24, for inputting corresponding DTX unit 21 according to described ground unrest type by the ambient noise signal of reception.

The function of background noise analysis unit 22 is to analyze the ambient noise signal of input to obtain the characteristic parameter of background signal, analytical approach is to calculate one or more signal characteristic parameter values that can characterize ambient noise signal, and these signal characteristic parameters can be one of following parameter or combination in any: when long, SNR (signal to Noise Ratio), ground unrest are estimated level, ground unrest energy hunting estimation, ground unrest spectrum fluctuation estimation, ground unrest vibration frequency etc.Other can characterization signal background the parameter of size, kind (stable state/unstable state, changes size, changes speed), SNR (when long, in short-term, peak value) also can be calculated for analyzing.Except ambient noise signal being inputted background noise analysis unit 22, VAD can also be adjudicated to current demand signal is the indicator signal input background noise analysis unit 22 of background signal, and background noise analysis unit 22 triggers and calculates signal characteristic parameter value according to ambient noise signal indication information.

The function of ground unrest taxon 23 is the signal characteristic parameter values that calculate according to background noise analysis unit 22, divides the type of current background noise signal.Ground unrest type at least defines two kinds, can define according to whether stablizing of ground unrest, for example: ground unrest is divided into stationary background noise signal and unstable state ambient noise signal two classes; Whether stablizing and the size definition of SNR when long according to ground unrest and for example: ground unrest is divided into the large SNR ambient noise signal of stable state, the little SNR ambient noise signal of stable state, the large SNR ambient noise signal of unstable state, little SNR ambient noise signal four classes of unstable state.Can also be according to the size definition ground unrest type of ground unrest institute inclusion information meaning, such as ambient noise signal can be divided into melodious ambient noise signal (as music, flowing water, bird's twitters etc.), environmental background noise signal is (as cocktail party, curbside, airport etc.) or other ambient noise signal three classes.In a word, the classified types of ground unrest can be set as required, does not limit the enforcement of technical solution of the present invention.

Coding is the pre-stored applicable DTX unit 21 of each ground unrest type in switch unit 24, the ground unrest type definite according to ground unrest taxon 23, between at least two DTX unit 21, switch, select the DTX unit 21 corresponding with current background noise type to process.

Generally each ground unrest type is distinguished corresponding a kind of different DTX scheme, but consider demand or the communication channel resources situation of change of concrete communication scenes, can also set different switch modes, the switching controls scheme that each switch mode is corresponding different, can the corresponding concrete switching controls scheme of each switch mode is pre-stored in coding switch unit 24, and by the required switch mode of switch mode steering order indication coding switch unit 24, switch mode steering order can be limited instruction set of the current channel resource situation of reflection, the corresponding a kind of switch mode of each specific switch mode steering order, it is the distribution method of a kind of DTX unit.Also can, when switch mode steering order each time, concrete switching controls scheme corresponding to required switch mode be sent to coding switch unit 24 simultaneously.

Coding switch unit 24 is determined corresponding switch mode according to the indication information of concrete reason.To those skilled in the art, concrete switch mode switching control method is a lot, will not enumerate here.

Therefore as shown in Figure 3, in the code device of ground unrest shown in Fig. 2 20, can also increase command reception unit 25 is set, receive switch mode steering order and be transmitted to coding switch unit 24, switch mode steering order can be operating personnel from outside input, can be also transmitting terminal according to the communication channel resource conditions generation detecting etc.

Coding switch unit 24 can be selected appropriate switch mode according to switch mode steering order, has at least DTX unit 21 corresponding to ground unrest type not identical between any two switch modes.The ground unrest code device 20 of take comprises that four DTX unit 21 are example, the setting example of switch mode as shown in Figure 4:

When command reception unit 25 is not set, coding switch unit 24 is only that ground unrest is selected appropriate DTX unit according to ground unrest type, and for example switch mode 1;

When being provided with command reception unit 25, coding switch unit 24 is determined the pattern of selecting DTX unit according to switch mode steering order.For example:

While receiving switch mode steering order 1, according to switch mode 1, switch;

While receiving switch mode steering order 2, according to switch mode 2, switch;

While receiving switch mode steering order 3, according to switch mode 3, switch.

Switch mode steering order can be sent according to current communication channel resource conditions, for example, when current channel resource is nervous, wherein the ground unrest of part ground unrest type may be used samely compared with the DTX unit of low rate, and even all ground unrest types are all used minimum speed limit DTX unit.Again for example, when there is no switch mode steering order, only have wherein part DTX unit to be used, but when switch mode steering order represents that current channel resource is abundant or nervous, another part DTX unit is used.

Example is as shown in Figure 4 visible:

Instruction 1 indication enters switch mode 1, each ground unrest type corresponding a kind of different DTX unit all now, and instruction 1 can mean that current channel resource is normal;

Instruction 2 indications enter switch mode 2, the DTX unit that now part ground unrest type can be corresponding identical, as ground unrest Class1 and the equal corresponding DTX of ground unrest type 2 unit 212, ground unrest type 3 and the equal corresponding DTX of ground unrest type 4 unit 214, instruction 2 can represent current channel resource, and abundant (212 compare DTX unit, DTX unit 211 code rates are high, take resource many; 214 compare DTX unit, DTX unit 213 code rates are high, take resource many), or can represent that more nervous (212 compare DTX unit, DTX unit 211 code rates low, takies resource few to current channel resource; 214 compare DTX unit, DTX unit 213 code rates are low, take resource few);

Instruction 3 indications enter switch mode 3, the equal corresponding DTX unit 214 only of all ground unrest types now, instruction 3 can represent current channel resource, and very abundant (DTX unit 214 code rates are the highest, take resource maximum), or can represent current channel resource very nervous (DTX unit 214 code rates are minimum, take resource minimum).

As shown in Figure 5, when very sufficient or ground unrest is melodious especially at current channel resource, can also speech coder be set in ground unrest code device 20 again, utilize speech coder to carry out full-rate codes to specific background noise.Satisfied the imposing a condition of signal characteristic parameter value of specific background noise, whether ground unrest taxon 23 meets and imposes a condition to judge specific background noise type according at least one signal characteristic parameter value, and indicate coding switch unit 24 when specific background noise, to be switched to speech coder, speech coder is encoded to speech frame continuous wave output by this specific background noise signal.Certainly, also can utilize the existing speech coder of transmitting terminal to bear this function.

Consult shown in Fig. 1, each DTX unit 21 comprises two parts, and one is non-voice scrambler, and for ambient noise signal being encoded to SID frame, it two is background noise frames transmitter, for sending background noise frames according to the discontinuous mode of setting.Difference between DTX unit can be the difference of SID frame coding mode, can be also the difference of discontinuous transmission mechanism, or the two is neither same.Wherein: the difference of SID frame coding mode can be embodied in the following aspects:

1, the difference of coding parameter

Different SID frames can adopt the parameter of different types of parameter and varying number;

2, the difference of identical parameters quantified precision

For the different SID frames of same parameters, can adopt different quantified precisions;

Therefore, the structure of the SID frame that different SID frame coding modes produce may be different with length, and for example: the coding parameter of stationary background noise signal noise can be relatively simple, and the quantified precision of parameter can be relatively low, the length of SID frame can be relatively short.Corresponding, the coding parameter of unstable state ambient noise signal can be relatively many, a kind of method is to increase the excitation parameters in speech coding parameters in existing SID frame, as adaptive codebook parameter (pitch delay and gain) and fixed codebook parameters (position of non-zero pulses and symbol and gain) etc., the quantified precision of parameter can be relatively high, as increasing quantified precision of existing SID frame median filter parameter and excitation parameters etc., the length of SID frame also can correspondingly increase.And for example, the SID frame of melodious ambient noise signal should be used the most complicated coded system (can be full-rate codes), the SID frame of environmental background noise signal is used time complicated coded system, and the SID frame of other ambient noise signal is used better simply coded system.

Different ground unrest types can also be used different discontinuous transmission mechanism, and the difference of the SID frame transmit mechanism of different background noise type is mainly reflected on the transmission intercal of adjacent two SID frames.For example, for stationary background noise signal, because the change of background is less, the Quality Down susceptibility after DTX/CNG is not high to such ground unrest for human auditory system, only needs with relatively long discontinuous transmission interval.For unstable state ambient noise signal, because background change is in time larger, in order to guarantee that such ground unrest can have a higher subjective quality after DTX/CNG, need to use relatively short discontinuous transmission interval or even continuous transmission.And for example, for melodious background, due to the non-Noise Characteristic of itself, the requirement of subjective quality is quite high, so DTX system just need to be carried out continuous transmission.For environmental background noise, due to itself, carried certain environmental information, in order to ensure not losing these information, DTX system also need to be upgraded SID frame with higher speed.For other ground unrest, due to true appearance the non-staple object of reduction background, so the renewal of carrying out SID frame that DTX system just can be relatively slow.

The transmission intercal of SID frame is divided into two kinds, constant transmissions interval and Adaptive Transmission interval.Under the mechanism at constant transmissions interval, the interval of adjacent two SID interframe is fixed, and the size at interval just depends on the type of ground unrest.Under the mechanism at Adaptive Transmission interval, although the transmission of SID frame is to be adaptive to the speed that noise changes, also there is the restriction of minimum time interval.In the embodiment of the present invention, according to different background noise type, can determine to adopt different minimum intervals.

As shown in Figure 6, the primary structure schematic diagram for the first communication facilities provided by the invention, comprising: VAD10, ground unrest code device 20, speech coder 30 and communication interface 40, wherein:

VAD10, for detection of efficient voice, sends audible signal into speech coder 30 and is encoded to speech frame, and communication interface sends speech frame by communication channel; Ambient noise signal is sent into ground unrest code device 20, ground unrest code device 20 output SID frames, communication interface sends SID frame by communication channel.

In ground unrest code device 20, coding switch unit 24 switches applicable DTX unit 21 according to the type of ambient noise signal.

As shown in Figure 7, the second communication facilities primary structure schematic diagram providing for the embodiment of the present invention, wherein: the coding switch unit 24 in ground unrest code device 20 is determined switch mode according to the instruction of command reception unit 25, and switch applicable DTX unit 21 according to the switch mode of the type of ambient noise signal and current use.

Shown in Figure 5, in the ground unrest code device 20 in the communication facilities shown in Fig. 6 and Fig. 7, speech coder can also be set, speech coder completes the full-rate codes of ground unrest.

As shown in Figure 8, if utilize existing speech coder 30 to complete the full-rate codes of ground unrest, the switch unit 24 of encoding is connected on speech coder 30, ambient noise signal is sent into speech frame the continuous wave output that is encoded to full rate in speech coder 30.

The technical scheme that adopts the embodiment of the present invention to provide, the coding send mode of corresponding each the SID frame of needs arranges corresponding DTX unit indication information in SID frame, and DTX unit indication information is determined corresponding decoding and reconstituting scheme for decoding end.For speech frame, by Voice decoder, decoded and continuous wave output.

Due to 22 pairs of background noise analysis unit, calculating the calculating of background characteristics parameter value calculates identical with the part of each DTX unit 21 when encoding SID frame, therefore background noise analysis unit 22 can also connect each DTX unit 21, analysis result is exported to each DTX unit 21, thereby part reduces the calculated amount of DTX unit 21.

As shown in Figure 9, a kind of communication apparatus construction schematic diagram providing for the embodiment of the present invention, the unit that completes receiving end function mainly comprises: ground unrest decoding device 50, recognition unit 60, decoding switch unit 70, Voice decoder 80 and communication interface, wherein:

This ground unrest decoding device 50 comprises at least two CNG unit (51,52...5n), and each CNG unit is according to a kind of decoding and reconstituting scheme decoding SID frame of setting and be reconstructed into ground unrest;

Recognition unit 60, for identifying speech frame and the SID frame receiving by communication interface, and further identifies the DTX unit indication information in SID frame;

Decoding switch unit 70, for switching according to the recognition result of recognition unit 60, sends speech frame into voice decipher 80, SID frame is sent into corresponding CNG unit and carry out decoding and reconstituting.

Wherein, consult shown in Fig. 1, each CNG unit comprises non-voice demoder, for the background noise frames of decoding and receiving from communication channel; Reconstructor, for adopting the mode of interpolation that decoded data is reconstructed into continuous ambient noise signal, to increase the comfort of listener.

If first recognition unit 60 identifies speech frame or SID frame and SID frame that the coded frame receiving is full-rate codes, further identify the DTX unit indication information in this SID frame., should have 1 or several for carrying the bit of DTX unit indication information in the code stream of SID frame here, bit bit position both can also can be at end in the beginning of SID frame, or for example, at other assigned address of SID frame, shown in following table 3 or table 4:

Table 3.

Bit distributes (from a high position to low level)	Parametric description
		S1	DTX unit indication information
S2	Quantitative prediction device index
		S3-S7	One-level LSF quantisation vector index
S8-S11	Secondary LSF quantisation vector index
		S12-S16	Gain (energy)

Table 4.

Bit distributes (from a high position to low level)	Parametric description
		S1	Quantitative prediction device index
S2-S6	One-level LSF quantisation vector index
		S7-S10	Secondary LSF quantisation vector index
S11-S15	Gain (energy)
		S16-S17	DTX unit indication information

In recognition unit 60, can comprise two recognin unit, a recognin unit is for identifying speech frame or the SID frame that the coded frame receiving is full-rate codes, if identify speech frame, notify decoding switch unit 70, decoding switch unit is sent speech frame into Voice decoder 80 decoding continuous wave outputs; If identifying is SID frame, SID frame is sent into another recognin unit and further identified the DTX unit indication information in SID frame.

Because different ground unrest types has different SID frame format or update mechanism, according to DTX unit indication information, select suitable DTX decoding scheme decoding SID frame, noise interpolating method in different decoding schemes is different, the comfort noise being finally restored out.

Communication facilities can possess transmitting terminal function, also concrete receiving end function, structure shown in Fig. 9 conventionally and the merging of structure shown in Fig. 7 or Fig. 8 be arranged in a communication facilities.

With specific embodiment, be elaborated again below.

Embodiment mono-

At coding side, ambient noise signal is inputted background noise analysis unit 22 frame by frame, background noise analysis unit 22 calculates the subband level level[k of each current demand signal frame, i], wherein k and i represent to be respectively the level of i frame k subband and to be kept in internal memory.The calculating of subband can be realized also and can be realized by the method for conversion by bank of filters.

When the enough somes of signal frame accumulation, background noise analysis unit 22 starts according to being kept at the subband level in internal memory, calculates ground unrest spectrum fluctuation parameters spec_var[i], and after upgrade result of calculation while receiving each frame background noise frames.

Ground unrest spectrum fluctuation parameters spec_var[i] computing method be:

$\mathrm{spec}\_\mathrm{var}\left[i\right]=\underset{m=i-L+1}{\overset{i}{\mathrm{\Σ}}}\left(\underset{n=i-L+1,n\≠m}{\overset{i}{\mathrm{\Σ}}}\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{(\left(\mathrm{level}\right[k,m]-\mathrm{level}[k,n\left]\right)/\left(\frac{1}{L\·N}\underset{m=i-L+1}{\overset{i}{\mathrm{\Σ}}}\underset{k=1}{\overset{N}{\mathrm{\Σ}}}\mathrm{level}[k,i]\right))}^2\right)\right)$

The quantity that wherein N is subband, L is chosen as to do ambient noise signal number of frames average when long.Ground unrest spectrum fluctuation parameters spec_var[i] calculating can be also based on LSF coefficient, that is:

$\mathrm{spec}\_\mathrm{var}\left[i\right]=\underset{m=i-L+1}{\overset{i}{\mathrm{\Σ}}}\left(\underset{n=i-L+1,n\≠m}{\overset{i}{\mathrm{\Σ}}}\left(\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{(\left(\mathrm{lsf}\right[k,m]-\mathrm{lsf}[k,n\left]\right)/\left(\frac{1}{L\·N}\underset{m=i-L+1}{\overset{i}{\mathrm{\Σ}}}\underset{k=1}{\overset{N}{\mathrm{\Σ}}}\mathrm{lsf}[k,i]\right))}^2\right)\right)$

Lsf[k wherein, i] represent k lsf coefficient of i frame, N is lsf number of coefficients, L is chosen as to do ambient noise signal number of frames average when long.

Background noise analysis unit 22 exports the ground unrest spectrum fluctuation parameters calculating to ground unrest taxon 23.

Coding switch unit 24 is that current background noise frame is selected corresponding DTX unit according to the ground unrest type information of ground unrest taxon 23 outputs.

In the present embodiment, two kinds of background type have been defined: stationary background noise signal and unstable state ambient noise signal.Ground unrest taxon 23 compares and incorporates current background noise into in stable state or unstable state a kind of by ground unrest being composed to fluctuation parameters and thresholding spec_var_thr.If spec_var[i] < spec_var_thr, current background noise signal is classified as stationary background noise signal, otherwise is classified as unstable state ambient noise signal.

In the present embodiment, DTX system has two kinds of different programmes of work:

Scheme one adopts every 8 frames to send the discontinuous transmission mechanism of a SID frame, only comprises wave filter and energy parameter in SID frame;

Scheme two adopts every 3 frames to send the discontinuous transmission mechanism of a SID frame, and in SID frame, except comprising wave filter, energy parameter has also increased fixed codebook parameters (belonging to excitation parameters) outward.

When current background noise belongs to stable state class, coding switch unit 24 is selected programme of work one.When current background noise belongs to unstable state class, coding switch unit 24 is selected programme of work two.

During every kind of SID frame assembling, a DTX unit indication information is set in frame to state which kind of DTX unit this SID frame belongs to.In the present embodiment, DTX unit indication information can represent with 1 bit, and 0 and 1 represents respectively Yi HeDTX unit, DTX unit two.

In decoding end, first recognition unit 60 judges the DTX unit under this SID frame according to the DTX unit indication information in SID frame, and decoding switch unit 70 is inputted corresponding CNG unit 50 by SID frame and carried out decoding and reconstituting.Wherein:

If DTX unit indication information is 0, corresponding CNG unit obtains filter parameter and energy parameter and 8 ambient noise signal frames is thereafter carried out to interpolation according to the SID frame format decoding in above-mentioned DTX programme of work one;

If DTX unit indication information is 1, corresponding CNG unit obtains filter parameter, energy parameter and fixed codebook parameters and 3 ambient noise signal frames is thereafter carried out to interpolation according to the SID frame format decoding in above-mentioned DTX programme of work two.

Embodiment bis-

At coding side, ambient noise signal is inputed to background noise analysis unit 22 frame by frame, and background noise analysis unit 22 calculates the subband level level[k of current background noise signal frame, i], frame energy pow[i] and signal to noise ratio (S/N ratio) snr[i in short-term], and be kept in internal memory.Wherein:

Subband level level[k, i] calculating can realize also and can realize by the method for conversion by bank of filters;

The account form of frame energy is:

wherein N represents total sub band number;

The account form of signal to noise ratio (S/N ratio) is in short-term:

bckr_noise_pow[i wherein] ground unrest energy is estimated when long.

When the enough somes of ambient noise signal frame accumulation, background noise analysis unit 22 starts according to being kept at short-time characteristic parameter (the subband level in internal memory, frame energy, signal to noise ratio (S/N ratio) in short-term) calculate ground unrest energy hunting pow_var[i], signal to noise ratio (S/N ratio) snr_long[i when long], ground unrest energy is estimated bckr_noise_pow[i when long], and when receiving each ambient noise signal frame, upgrade result of calculation.

When long signal to noise ratio (S/N ratio), when the renewal of all the other calculation of parameter results occurs over just current demand signal frame and is ambient noise signal frame, when when long, the renewal of signal to noise ratio (S/N ratio) occurs over just current demand signal frame and is non-ambient noise signal frame.

Signal to noise ratio (S/N ratio) snr_long[i when long] computing method be:

wherein L is chosen as to do non-ambient noise signal number of frames average when long.

Ground unrest energy hunting pow_var[i] computing method be:

$\mathrm{pow}\_\mathrm{var}\left[i\right]=\frac{1}{L}*\underset{m=i-L+1}{\overset{i}{\mathrm{\&Sigma;}}}{(\mathrm{pow}\left[m\right]-\frac{1}{L}*\underset{m=i-L+1}{\overset{i}{\mathrm{\&Sigma;}}}\mathrm{pow}\left[m\right])}^2,$ Wherein L is chosen as to do ambient noise signal number of frames average when long.

When long, ground unrest energy is estimated bckr_noise_pow[i] computing method be:

Bckr_noise_pow[i]=(1-α) * bckr_noise_pow[i-1]+α * pow[i], wherein α is the scale factor of 0 to 1, in the present embodiment, value is 5%.

Background noise analysis unit 22 is by the ground unrest energy hunting pow_var[i calculating], signal to noise ratio (S/N ratio) snr_long[i when long] export ground unrest taxon 23 to.

In the present embodiment, four kinds of background type have been defined: the large SNR ambient noise signal of stable state, the little SNR ambient noise signal of stable state, the large SNR ambient noise signal of unstable state and the little SNR ambient noise signal of unstable state.

Ground unrest taxon 23 is by by pow_var[i] and snr_long[i] compare with an energy hunting thresholding pow_var_thr and a signal-noise ratio threshold snr_thr respectively, it is a class of 4 kinds of ground unrest type kinds that current background noise is incorporated into, wherein:

As pow_var[i] when > pow_var_thr and snr_long > snr_thr, current background is the large SNR ambient noise signal of unstable state;

As pow_var[i] when > pow_var_thr and snr_long < snr_thr, current background is the little SNR ambient noise signal of unstable state;

As pow_var[i] when < pow_var_thr and snr_long > snr_thr, current background is the large SNR ambient noise signal of stable state;

As pow_var[i] when < pow_var_thr and snr_long < snr_thr, current background is the little SNR ambient noise signal of stable state.

In the present embodiment, one has four kinds of different DTX unit, wherein:

DTX unit one adopts every 8 frames to send the discontinuous transmission mechanism of a SID frame, only comprises filter parameter and energy parameter in SID frame;

DTX unit two adopts every 8 frames to send the discontinuous transmission mechanism of a SID frame, also comprises fixed codebook parameters in SID frame except comprising filter parameter and energy parameter;

DTX unit three adopts every 3 frames to send the discontinuous transmission mechanism of a SID frame, also comprises fixed codebook parameters in SID frame except comprising filter parameter and energy parameter;

DTX unit four adopts the continuous transmission mechanism of full-rate codes.

Coding switch unit 24, except receiving the ground unrest type information of ground unrest taxon 23, is also connected with a switch mode steering order interface.Switch mode steering order has 5 kinds of indications, represent respectively that bandwidth is normal, bandwidth is more, bandwidth is less, bandwidth is a lot of and bandwidth seldom.

The switch mode relation of switch mode steering order and DTX system is as follows:

When switch mode steering order represents that bandwidth is normal, the switch mode of DTX system is: the large SNR class of stable state background is adopted to DTX unit one; The little SNR class of stable state background is adopted to DTX unit two; The large SNR class of unstable state background is adopted to DTX unit three; The little SNR class of unstable state background is adopted to DTX unit four.

When switch mode steering order represents that bandwidth is more, the switch mode of DTX system is: stable state large SNR class background and the little SNR class of stable state background are all adopted to DTX unit two; Unstable state large SNR class background and the little SNR class of unstable state background are all adopted to DTX unit four;

When switch mode steering order represents that bandwidth is less, the switch mode of DTX system is: stable state large SNR class background and the little SNR class of stable state background are all adopted to DTX unit one; Unstable state large SNR class background and the little SNR class of unstable state background are all adopted to DTX unit three;

When switch mode steering order represents that bandwidth is a lot, the switch mode of DTX system is: four kinds of ground unrest types are all adopted to DTX unit four;

When switch mode steering order represents that bandwidth seldom, the switch mode of DTX system is: four kinds of ground unrest types are all adopted to DTX unit one.

During every kind of SID frame assembling, a DTX unit indication information is set in frame to state which kind of DTX unit this SID frame belongs to.DTX unit indication information in the present embodiment can represent with 2 bits, wherein: 00,01 and 10 represent respectively Er HeDTX unit, Yi,DTX unit, DTX unit three, so scheme four due to be full-rate codes without sign out.

In decoding end, first recognition unit 60 judges the DTX unit under this SID frame according to the DTX unit indication information in SID frame, and indicates corresponding CNG unit decodes reconstruct, wherein:

If DTX unit indication information is 00, according to the SID frame format decoding in above-mentioned DTX programme of work one, obtains filter parameter and energy parameter and 8 ambient noise signal frames are thereafter carried out to interpolation;

If DTX unit indication information is 01, according to the SID frame format decoding in above-mentioned DTX programme of work two, obtain excitation parameters, filter parameter and energy parameter also carry out interpolation to 8 ambient noise signal frames thereafter.

If DTX unit indication information is 10, according to the SID frame format decoding in above-mentioned DTX programme of work three, obtains filter parameter, energy parameter and fixed codebook parameters and 3 ambient noise signal frames are thereafter carried out to interpolation.

Those skilled in the art can be flexibly according to signal characteristic parameter dividing background noise type, and sets switch mode, will not enumerate here.

In sum, the embodiment of the present invention provides a kind of coding method, comprising:

Obtain at least one signal characteristic parameter value of ambient noise signal to be sent;

According to described at least one signal characteristic parameter value, determine ground unrest type;

According to a kind of DTX mechanism of described ground unrest type selecting, ambient noise signal is encoded to SID frame the discontinuous transmission that comprises DTX mechanism indication information, wherein the SID frame coding mode in each DTX mechanism and/or the discontinuous send mode of SID frame are not identical.

Further, described coding method also comprises: set different DTX mechanism preference pattern, i.e. different switch mode, has at least DTX mechanism corresponding to ground unrest type not identical between any two DTX mechanism preference pattern; And according to selected DTX mechanism preference pattern, for each ground unrest type is determined corresponding DTX mechanism.

A nearlyer step, described method also comprises: when setting value that described at least one signal characteristic parameter value is, ambient noise signal is encoded to the speech frame of full rate and sends continuously.

Wherein, signal characteristic parameter value comprises one of following or combination in any: when long, SNR, ground unrest estimate that level, ground unrest energy hunting estimation, the fluctuation of ground unrest spectrum estimate and ground unrest vibration frequency.

The embodiment of the present invention also provides a kind of method that corresponding SID frame is decoded, and comprising:

DTX mechanism indication information in identification SID frame, wherein the SID frame coding mode in each DTX mechanism is not identical;

According to described DTX mechanism indication information, adopt corresponding decoding process decoding SID frame, and be reconstructed into ambient noise signal.

The technical scheme that the application embodiment of the present invention proposes, can be according to the actual state of the dissimilar and communication system of ground unrest, adopt different SID frame coding methods and discontinuous transmission mechanism, in the situation that rationally utilizing communication system real resource, be adaptive to the desired encoding precision of dissimilar ground unrest and time sense, reach under the prerequisite of rationally utilizing channel resource in the high-quality object that all kinds ground unrest is provided of decoding end.

Obviously, those skilled in the art can carry out various changes and modification and not depart from the spirit and scope of the present invention the present invention.Like this, if within of the present invention these are revised and modification belongs to the scope of the claims in the present invention and equivalent technologies thereof, the present invention is also intended to comprise these changes and modification interior.

Claims (31)

1. A background noise encoding device, characterized in that, comprising: A background noise analysis unit, configured to obtain at least one signal characteristic parameter value of a background noise signal; the background noise signal includes background noise, and the background noise includes white noise, background noise and silence; A background noise classification unit, configured to determine the type of background noise according to the at least one signal characteristic parameter value; A coding switching unit and at least two discontinuous transmission units, the coding switching unit is used to input the background noise signal into the corresponding discontinuous transmission unit according to the background noise type, and each discontinuous transmission unit is used to code the background noise signal The silence description frame containing the indication information of the discontinuous transmission unit is not transmitted continuously, and the encoding method and/or discontinuous transmission mechanism of the silence description frame of each discontinuous transmission unit is different, and the silence description frame of each discontinuous transmission unit The encoding methods are different, specifically: the encoding parameters of different silence description frames are different, or the quantization accuracy of different silence description frames with the same encoding parameter is different, and the discontinuous transmission mechanisms of the silence description frames of each discontinuous transmission unit are different. Specifically: send The transmission intervals of two adjacent silence description frames are different. 2. The background noise encoding device according to claim 1, further comprising: an instruction receiving unit, configured to receive a switching mode control instruction and forward it to the encoding switching unit, and the switching mode control instruction is used to instruct encoding switching The unit uses a switching pattern, and at least one background noise type between any two switching patterns corresponds to a different discontinuous transmission unit. 3. The background noise encoding device according to claim 1, further comprising: a speech encoder, wherein the encoding switching unit converts the background noise signal according to one of the specified background noise types recognized by the background noise classification unit The input speech encoder encodes speech frames and outputs them continuously. 4. The background noise encoding device according to claim 1, 2 or 3, wherein the discontinuous transmission unit specifically comprises: A non-speech encoder for encoding background noise signals into silence description frames; The silence description frame sender is used for discontinuously sending the silence description frame. 5. A background noise decoding device for decoding the silence description frame output by the background noise coding device according to claim 1, characterized in that it comprises: The first identification unit is used to identify the DTU indication information in the silence description frame; A decoding switching unit and at least two comfort noise generating units, the decoding switching unit is used to input the silence description frame into the corresponding comfort noise generating unit according to the identification result of the identifying unit, and each comfort noise generating unit receives The encoding method of the silence description frame corresponds to decoding the silence description frame and reconstructing the background noise signal. 6. The background noise decoding device according to claim 5, further comprising: A second identification unit, the second identification unit is used to identify the speech frame and the silence description frame, and send the silence description frame to the first identification unit, or notify the decoding switching unit when the speech frame is recognized; Speech decoder, the decoding switching unit sends the speech frame recognized by the second recognition unit to the speech decoder for decoding and continuous output. 7. The background noise decoding device according to claim 6, wherein the first identification unit and the second identification unit are combined. 8. A communication device comprising: The first voice detector is used to detect whether valid voice is included in the collected signal and to classify and output; The first speech coder is used for encoding the signal comprising effective speech into speech frames and outputting them continuously; It is characterized in that it also includes a background noise coding device, and the background noise coding device specifically includes: A background noise analysis unit, configured to obtain at least one signal characteristic parameter value of a background noise signal that does not contain effective speech; the background noise signal includes background noise, and the background noise includes white noise, background noise and silence; A background noise classification unit, configured to determine the type of background noise according to the at least one signal characteristic parameter value; A coding switching unit and at least two discontinuous transmission units, the coding switching unit is used to input the background noise signal into the corresponding discontinuous transmission unit according to the background noise type, and each discontinuous transmission unit is used to code the background noise signal The silence description frame containing the indication information of the discontinuous transmission unit is not transmitted continuously, and the encoding method and/or discontinuous transmission mechanism of the silence description frame of each discontinuous transmission unit is different, and the silence description frame of each discontinuous transmission unit The encoding methods are different, specifically: the encoding parameters of different silence description frames are different, or the quantization accuracy of different silence description frames with the same encoding parameter is different, and the discontinuous transmission mechanisms of the silence description frames of each discontinuous transmission unit are different. Specifically: send The transmission intervals of two adjacent silence description frames are different. 9. The communication device according to claim 8, wherein the background noise encoding device further comprises: an instruction receiving unit, configured to receive the switching mode control instruction and forward it to the encoding switching unit, and the switching mode control instruction uses In order to instruct the coding switching unit to use a switching mode, at least one background noise type between any two switching modes corresponds to different discontinuous transmission units. 10. The communication device according to claim 8, wherein the coding switching unit is further connected to the first speech coder, and the coding switching unit recognizes one of the specified background noises according to the background noise classification unit Type, the background noise signal is input into the first speech coder to be encoded into speech frames and output continuously. 11. The communication device according to claim 8, wherein the background noise coding device further comprises: a second speech coder, and the coding switching unit recognizes one of the specified background noises according to the background noise classification unit type, the background noise signal is input into the second speech coder to be encoded into speech frames and output continuously. 12. The communication device according to claim 8, 9, 10 or 11, wherein the discontinuous transmission unit specifically comprises: A non-speech coder, used to encode the background noise signal into a silence description frame according to the set silence description frame encoding method; The silence description frame sender is configured to send the silence description frame according to the set discontinuous transmission mechanism. 13. The communication device of claim 8, further comprising: The first identification unit is used to identify the DTU indication information in the silence description frame; A decoding switching unit and at least two comfort noise generating units, the decoding switching unit is used to input the silence description frame into the corresponding comfort noise generating unit according to the identification result of the identifying unit, and each comfort noise generating unit receives The encoding method of the silence description frame corresponds to decoding the silence description frame and reconstructing the background noise signal. 14. The communication device of claim 13, further comprising: A second identification unit, the second identification unit is used to identify the speech frame and the silence description frame, and send the silence description frame to the first identification unit, or notify the decoding switching unit when the speech frame is recognized; Speech decoder, the decoding switching unit sends the speech frame recognized by the second recognition unit to the speech decoder for decoding and continuous output. 15. The communication device according to claim 14, wherein the first identification unit and the second identification unit are combined. 16. A communication device, comprising a background noise decoding device, characterized in that the background noise decoding device comprises: The first identification unit is used to identify the DTU indication information in the silence description frame; A decoding switching unit and at least two comfort noise generating units, the decoding switching unit is used to input the silence description frame into the corresponding comfort noise generating unit according to the identification result of the identifying unit, and each comfort noise generating unit receives The encoding method of the silence description frame corresponds to decoding the silence description frame, and reconstructing the background noise signal; the background noise signal includes background noise, and the background noise includes white noise, background noise and silence. 17. The communication device of claim 16, further comprising: A second identification unit, the second identification unit is used to identify the speech frame and the silence description frame, and send the silence description frame to the first identification unit, or notify the decoding switching unit when the silence description frame is recognized; A speech decoder, wherein the decoding switching unit sends the speech frame recognized by the second recognition unit to the speech decoder for decoding and outputting. 18. The communication device according to claim 17, wherein the first identification unit and the second identification unit are combined. 19. A background noise encoding method, comprising: Obtaining at least one signal characteristic parameter value of a background noise signal; the background noise signal includes background noise, and the background noise includes white noise, background noise and silence; determining a background noise type according to the at least one signal characteristic parameter value; Select a discontinuous transmission mechanism according to the background noise type to encode the background noise signal into a silence description frame containing the discontinuous transmission mechanism indication information and send it non-continuously, wherein the silence description frame encoding method in each discontinuous transmission mechanism and /or the non-continuous transmission modes of the silence description frames are different, and the coding methods of the silence description frames of each discontinuous transmission unit are different, specifically: the encoding parameters of different silence description frames are different, or the quantization accuracy of the silence description frames with the same encoding parameters is different Similarly, the discontinuous transmission mechanisms of the silence description frames of each discontinuous transmission unit are different, specifically: the transmission intervals of two adjacent silence description frames to be sent are different. 20. The encoding method according to claim 19, wherein said selecting a discontinuous transmission mechanism according to the background noise type is: selecting the background noise type and the discontinuous transmission mechanism in the mode according to the set discontinuous transmission mechanism If the corresponding relationship is selected, at least one discontinuous transmission mechanism corresponding to the background noise type is different between any two switching modes. 21. The encoding method according to claim 19, further comprising: when the at least one signal characteristic parameter value satisfies a set condition, encoding the background noise signal into a full-rate speech frame and sending it continuously. 22. The encoding method according to claim 19, 20 or 21, wherein the signal characteristic parameters include one or any combination of the following: long-term SNR, background noise estimation level, background noise energy fluctuation estimation, background noise Noise spectral fluctuation estimation and background noise fluctuation frequency. 23. The encoding method according to claim 22, characterized in that, according to the value of each signal characteristic parameter, the background noise type includes a steady-state background noise signal and a non-stationary background noise signal. 24. The encoding method according to claim 23, wherein when the signal characteristic parameter comprises a long-term SNR: The steady-state background noise signal further includes: a steady-state large SNR background noise signal and a steady-state small SNR background noise signal; The unsteady background noise signal further includes: an unsteady large SNR background noise signal and an unsteady small SNR background noise signal. 25. The encoding method according to claim 19, wherein the differences between different silence description frame encoding methods include: at least one silence description frame encoding parameter is different, or the quantization precision of the same silence description frame encoding parameter is different . 26. The encoding method according to claim 25, characterized in that, said silence description frame encoding parameters include synthesis filter parameters and excitation parameters. 27. The encoding method according to claim 26, wherein the excitation parameters include adaptive codebook parameters and/or fixed codebook parameters. 28. The encoding method according to claim 19, wherein the difference between discontinuous transmission modes of different silence description frames comprises: the transmission interval between two adjacent silence description frames sent is different. 29. The encoding method according to claim 28, wherein the transmission interval is adaptively determined according to the change of the background signal, and the minimum intervals between the discontinuous transmission modes of different silence description frames are different. 30. The encoding method according to claim 28, wherein the transmission intervals of the discontinuous transmission modes of each silence description frame are different fixed values. 31. A method for decoding the silence description frame generated by the method according to claim 19, comprising: Identify the discontinuous transmission mechanism indication information in the silence description frame, where the encoding methods of the silence description frame in each discontinuous transmission mechanism are different; According to the discontinuous transmission mechanism indication information, the silence description frame is decoded in a corresponding decoding manner, and reconstructed into a background noise signal.

Images