WO2005096508A1

WO2005096508A1 - Enhanced audio encoding and decoding equipment, method thereof

Info

Publication number: WO2005096508A1
Application number: PCT/CN2004/001034
Authority: WO
Inventors: Xingde Pan; Dietz Martin; Andreas Ehret; Holger HÖRICH; Xiaoming Zhu; Michael Schug; Weimin Ren; Lei Wang; Hao Deng; Fredrik Henn
Original assignee: Beijing Media Works Co., Ltd; Beijing E-World Technology Co., Ltd.; Coding Technologies Ab
Priority date: 2004-04-01
Filing date: 2004-09-09
Publication date: 2005-10-13

Abstract

The present invention relates to enhanced audio encoding equipment, including band extending module , re-sampling module, psychological acoustic analyzing module , time-frequency mapping module , quantizing module , entropy coding module and bit stream diplexing module ; the band extending module analyzes the original inputted audio signals in whole band-width , extracts the spectral envelope of the high-frequency part and the parameter characterized of the dependency between the lower and higher parts of the frequency spectral , and then outputs them to the bit stream diplexing module ; the re-sampling module re-samples the inputted audio signals , changes the sampling rate , and outputs them to the psychological acoustic analyzing module and the time-frequency mapping module ; The present invention is applicable to hi-fi compress encoding of audio signals which have configurations about all sorts of sampling rates and sound channels , it can support audio signals whose sampling rate has a range of 8kHz - 192kHz ; and it can support all the sound channel configurations which are available ; it also can support audio encode/decode which has wide range object code rate.

Description

Enhanced audio codec device and method

The present invention relates to the field of audio codec technology, and in particular to an enhanced audio codec apparatus and method based on a perceptual model.

Background technique

In order to obtain a high-fidelity digital audio signal, the digital audio signal is audio encoded or audio compressed for storage and transmission. The purpose of encoding an audio signal is to achieve a transparent representation of the audio signal with as few odds as possible, for example, there is little difference between the originally input audio signal and the encoded output audio signal.

In the early 1980s, the emergence of CDs represented the many advantages of using digital representations of audio speech, such as high fidelity, large dynamic range, and robustness. However, these advantages are at the expense of a very high data rate. For example, the sampling rate of the CD-quality stereo signal requires a sampling rate of 44.1 kHz, and each sample value is equally quantized with 15 bits, so that the uncompressed data rate reaches 1.41 Mb/s. Such a high data rate brings great inconvenience to the transmission and storage of data, especially in the case of multimedia applications and wireless transmission applications, and is limited by bandwidth and cost. In order to maintain high quality audio signals, new network and wireless multimedia digital audio systems are required to reduce the rate of data without compromising the quality of the audio. In response to the above problems, various audio compression techniques have been proposed which can obtain high compression ratio and high fidelity audio signals. Typical MPEG-1/-2/-4 technology of the International Organization for Standardization ISO/IEC , Dolby's AC-2/AC-3 technology, Sony's ATRAC/MiniDi sc/SDDS extraction, and Lucent's PAC/EPAC/MPAC technology. The following is a detailed description of MPEG-2 AAC technology and Dolby's AC-3 technology.

MPEG-1 technology and MPEG-2 BC technology are high-quality audio coding techniques mainly used for mono and stereo audio signals, with the need for multi-channel audio coding that achieves higher coding quality at lower bit rates. Increasingly, MPEG-2 BC encoding technology emphasizes backward compatibility with MPEG-1 technology, because five-channel high-quality encoding cannot be achieved at a code rate lower than 540 kbps. In response to this shortcoming, MPE&-2 AAC technology was proposed, which can achieve higher quality coding for five-channel signals at a rate of 320 kbps.

Figure 1 shows a block diagram of an MPEG-2 AAC encoder comprising a booster 101, a filter bank 102, a time domain noise shaping module 103, an intensity/coupling module 104, a psychoacoustic model, a second order backward self. The adaptive predictor 105, and the difference stereo module 106, the bit allocation and quantization coding mode 107, and the bitstream multiplexing module 108, wherein the bit allocation and quantization coding module 107 further includes compression! ^匕/distortion processing controller, scale factor module, non-uniform quantizer, and entropy encoding module.

The filter bank 102 employs a modified discrete cosine transform (MDCT) whose resolution is signal adaptive, that is, a 2048-point MDCT transform is used for the steady-state signal, and a 256-point MDCT transform is used for the transient signal; thus, for 48 kHz The sampled signal has a maximum frequency resolution of 23 Hz and a maximum inter-resolution of 2. 6 ras. At the same time, a sine window and a Ka i ser-Bes sel window can be used in the filter bank 102, and a sine window is used when the harmonic interval of the input signal is less than 140 Hz, and Ka i ser is used when a strong component of the input signal is greater than 220 Hz. -Bes sel window.

After the audio signal passes through the gain controller 101, it enters the filter bank 102, performs filtering according to different signals, and then processes the spectral coefficients output by the filter bank 102 through the time domain noise shaping module 103. The time domain noise shaping technique is in the frequency domain. Perform linear prediction analysis on the spectral coefficients, and then control the quantization noise according to the analysis. The shape in the time domain is used to achieve the purpose of controlling the pre-echo.

The intensity/coupling module 104 is used for stereo encoding of signal strength, since for a high frequency band (greater than 2 kHz) the sense of direction of the hearing is related to the change in signal strength (signal envelope), regardless of the waveform of the signal, That is, the constant envelope signal has no influence on the sense of direction of the hearing, so this feature and the related information between multiple channels can be used to combine several channels into one common channel for encoding, which forms an intensity/coupling technique.

The second-order backward adaptive predictor 105 is used to eliminate the redundancy of the steady-state signal and improve the coding efficiency. The difference and stereo (M/S) module 106 operates for a pair of channels, which are two channels such as a channel or a left and right surround channel in a channel signal or a multi-channel signal. The M/S module 106 utilizes the correlation between the two channels of the channel pair to achieve the effect of reducing the code rate and improving the coding efficiency. The bit allocation and quantization coding module 107 is implemented by a nested loop process in which the non-homogeneous quantizer performs lossy coding and the entropy coding module performs lossless coding, which removes redundancy and reduces correlation. The nested loop includes an inner loop and an outer loop, wherein the inner loop adjusts the step size of the non-uniform quantizer until the supplied bits are used up, and the outer loop uses the ratio of the quantization noise to the masking threshold to estimate the encoding quality of the signal. . The last encoded signal is passed through a bitstream multiplexing module 108 to form an encoded audio stream output.

In the case where the sampling rate is scalable, the input signal simultaneously generates four equal-band bands in the quad-band polyphase filter bank (PQF), and each band uses MDCT to generate 256 spectral coefficients, for a total of 1024. A gain controller 101 is used in each frequency band. In the decoder, the high frequency PQF band can be ignored to obtain a low sampling rate signal.

Figure 2 shows a block diagram of the corresponding MPEG-2 AAC decoder. The decoder includes a bitstream demultiplexing module 201, a lossless decoding module 202, an inverse quantizer 203, a scale factor module 204, and a difference stereo (M/S) module 205, a prediction module 206, an intensity/coupling module 207, a time domain A noise shaping module 208, a filter bank 209, and a gain control module 210. The encoded audio stream is demultiplexed by the bitstream demultiplexing module 201 to obtain a corresponding data stream and control stream. After the above signal is decoded by the lossless decoding module 202, an integer representation of the scale factor and a quantized value of the signal spectrum are obtained. The inverse quantizer 203 is a set of non-homogeneous quantizers implemented by a companding function for converting integer quantized values into reconstructed spectra. Since the scale factor module in the encoder differentiates the current scale factor from the previous scale factor and then uses the Huffman code for the difference value, the scale factor module 204 in the decoder performs Huffman decoding to obtain the corresponding difference value, and then recovers. A true scale factor. The M/S module 205 converts the difference channel to the left and right channels under the control of the side information. Since the second-order backward adaptive predictor 105 is used in the encoder to eliminate the redundancy of the steady-state signal and improve the coding efficiency, predictive decoding is performed by the prediction module 206 in the decoder. The intensity/coupling module 207 performs intensity/coupling decoding under the control of the side information, and then outputs it to the time domain noise shaping module 208 for time domain noise shaping decoding, and finally performs comprehensive filtering by the filter bank 209, and the filter bank 209 adopts the reverse direction. Improved Discrete Cosine Transform (IMDCT) technique.

For the case where the sampling frequency is scalable, the high frequency PQF band can be ignored by the gain control module 210 to obtain a low sampling rate signal.

MPEG-2 AAC codec technology is suitable for medium and high bit rate audio signals, but the encoding quality of low bit rate or low bit rate audio signals is poor. At the same time, the codec technology involves more codec modules. High complexity is not conducive to real-time implementation.

FIG. 3 is a schematic structural diagram of an encoder using Dolby AC-3 technology, including a transient signal detection module 301, an improved discrete cosine transform filter MDCT 302, a spectral envelope/exponential encoding module 303, a mantissa encoding module 304, Forward-backward adaptive sensing model 305, parameter bit allocation module 306, and bitstream multiplexing module 307. The audio signal is judged to be a steady state signal or a transient signal by the transient signal detecting module 301, and the time domain data is mapped to the frequency domain data by the signal adaptive MDCT filter bank 302, wherein a long window of 512 points is applied to the steady state signal. , a pair of short windows applied to the transient signal.

The spectral envelope/index encoding module 303 encodes the exponential portions of the signals in three modes according to the requirements of the code rate and frequency resolution, namely the D15, D25, and D45 encoding modes. The AC-3 technique differentially encodes the spectral envelope in frequency because at most ± ² increments are required, each increment represents a 6 dB level change, and the first DC term is encoded in absolute value, and the remaining indices are used. Differential encoding. In the D15 spectrum envelope index coding, each 4th number needs about 2.33 bits, 3 difference packets are encoded in a 7-bit word length, and the D15 coding mode provides a fine frequency by 4 times time resolution. Resolution. Since fine frequency resolution is only required for relatively stationary signals, and the spectrum of such signals remains relatively constant over many blocks, for steady state signals, D15 is occasionally transmitted, usually every 6 sound blocks (one The spectral envelope of the data frame is transmitted once. When the signal spectrum is unstable, it is necessary to update the estimated value of the spectrum estimate with a smaller frequency resolution coding, usually using the D25 and D45 coding modes. The D25 coding mode provides a suitable frequency resolution and time resolution, and is differentially coded every other frequency coefficient, so that each index requires approximately 1.15 bits. When the spectrum is stable on 2 to 3 blocks and then suddenly changes, the D25 encoding mode can be used. The D45 coding mode is differentially coded every three frequency coefficients, so that each index requires approximately 0.58 bits. The D45 encoding mode provides high temporal resolution and low frequency resolution, so it is generally used in the encoding of transient signals.

The forward-backward adaptive sensing model 305 is used to estimate the masking threshold for each frame of the signal. The forward adaptive part is only applied to the encoder end. Under the limitation of the code rate, an optimal set of perceptual model parameters is estimated by iterative loop, and then these parameters are passed to the backward adaptive part to estimate each frame. Mask the threshold. The backward adaptive part is applied to both the encoder side and the decoder side.

The parameter bit allocation module 306 analyzes the spectral envelope of the audio signal based on the masking criteria to determine the number of bits allocated to each mantissa. The module 306 utilizes a bit pool for global bit allocation for all channels. When encoding is performed in the mantissa encoding module 304, bits are cyclically extracted from the bit pool and allocated to all channels, and the quantization of the mantissa is adjusted according to the number of bits that can be obtained. For the purpose of compression coding, the AC-3 encoder also uses high frequency coupling technology to divide the high frequency part of the coupled signal into 18 sub-bands according to the critical bandwidth of the human ear, and then select some channels starting from a certain sub-band. Coupling. Finally, an AC-3 audio stream output is formed by the bit stream multiplexing module 307.

Figure 4 shows the flow diagram for decoding with Dolby AC-3. First, the bit 3⁄4⁄4 encoded by the AC-3 encoder is input, and data frame synchronization and error detection are performed on the bit stream. If a data error is detected, error concealment or muting processing is performed. The bit stream is then unpacked to obtain the main information and the side information, and then exponentially decoded. When performing exponential decoding, two side information is needed: one is the number of indexed packets; one is the indexing strategy used, such as D15, D25 or D45 mode. The decoded index and bit allocation side information are then bit-allocated, indicating the number of bits used for each packed mantissa, resulting in a set of bit allocation pointers, each bit allocation pointer corresponding to an encoded mantissa. The bit allocation pointer indicates the quantizer used for the mantissa and the number of bits occupied by each mantissa in the code stream. The single coded mantissa value is dequantized, converted to a dequantized value, the mantissa occupying zero bits is restored to zero, or replaced by a random jitter value under the control of the jitter flag. The decoupling operation is then performed, and the decoupling is to recover the high frequency portion of the coupled channel from the common coupling channel and the coupling factor, including the exponent and the mantissa. If the 2/0 mode encoding is used at the encoding end, matrix processing is applied to a certain sub-band, then the solution is; the 5th end needs to convert the sub-band and the difference channel value into left and right channel values through matrix recovery. The dynamic range control value of each audio block is included in the code stream, and the value is subjected to dynamic range compression to change the amplitude of the coefficient, including Index and mantissa. The frequency domain coefficients are inversely transformed into a time domain sample, and then the time domain samples are windowed, and adjacent blocks are overlapped and added to reconstruct a PCM audio signal. When the number of channels outputted by the decoding is smaller than the number of channels in the encoded bit stream, it is also necessary to downmix the audio signal and finally output the PCM stream.

Dolby AC-3 encoding technology is mainly for high bit rate multi-channel surround sound signals, but when the 5.1 bit encoding bit rate is 4 氐 384 kbps, its encoding effect is poor; and for mono and dual Channel stereo coding efficiency is also awkward.

In summary, the existing codec technology cannot comprehensively solve the codec quality from very low code rate, low bit rate to high bit rate audio signal and mono and two channel signals, and the implementation is complicated.

Summary of the invention

The technical problem to be solved by the present invention is to provide an apparatus and method for enhancing audio codec to solve the problem of low coding efficiency and poor quality of the lower rate audio signal in the prior art.

The enhanced audio coding apparatus of the present invention includes a frequency band extension module, a resampling module, a psychoacoustic analysis module, a time-frequency missing module, a quantization and entropy coding module, and a bit stream multiplexing module; The original input audio signal is analyzed over the entire frequency band, and the spectral envelope of the high frequency portion and its characteristics related to the low frequency portion are extracted and output to the bit stream multiplexing module; the resampling module is used for inputting The audio signal is resampled, the sampling rate of the audio signal is changed, and the audio signal of the changed sampling rate is output to the psychoacoustic analysis module and the time-frequency mapping module; the psychoacoustic analysis module is configured to calculate the input audio signal a masking threshold and a mask ratio, which are output to the quantization and entropy encoding module; the time-frequency mapping module is configured to convert a time domain audio signal into a frequency domain coefficient; and the quantization and entropy encoding module is used in the psychoacoustic analysis The frequency domain coefficients are quantized and entropy encoded under the control of the mask ratio output by the module, and output to the bits Multiplexing module; the bit stream multiplexing module for the received data are multiplexed, encoded audio stream is formed.

The enhanced audio decoding device of the present invention comprises a bit stream demultiplexing module, an entropy decoding module, an inverse quantizer group, a frequency-EI inter-mapping module and a band extension module; and the bit stream demultiplexing module is used for compression The audio data stream is demultiplexed, and outputs corresponding data signals and control signals to the entropy decoding module and the band extension module; the entropy decoding module is configured to decode the foregoing signals, and recover the quantized values of the spectra, Outputting to the inverse quantizer; the inverse quantizer group is configured to reconstruct an inverse quantization spectrum and output to the frequency-time mapping module; the frequency-time mapping module is configured to perform frequency-time on the spectral coefficients Mapping, obtaining a time domain audio signal of a low frequency band; the frequency band expansion module, configured to receive frequency band extension control information output by the bit stream demultiplexing module and the frequency-time mapping module, and a time domain audio signal of a low frequency band , reconstructing the high frequency signal portion and outputting the wideband audio signal.

The invention is applicable to high-fidelity compression coding of audio signals of various sampling rates and channel configurations, and can support audio signals with sampling rates between 8 kHz and 192 kHz; all possible channel configurations can be supported; and a wide range of support is supported. Target; code rate audio encoding/decoding.

DRAWINGS

Figure 1 is a block diagram of an MPEG-2 AAC encoder;

Figure 2 is a block diagram of an MPEG-2 AAC decoder;

Figure 3 is a schematic structural view of an encoder using Dolby AC-3 technology;

Figure 4 is a schematic diagram of a decoding process using Dolby AC-3 technology;

Figure 5 is a schematic structural view of an audio encoding device of the present invention;

6 is a schematic structural diagram of an audio decoding device of the present invention;

Figure 7 is a schematic structural view of Embodiment 1 of the coding apparatus of the present invention; FIG. 8 is a schematic structural diagram of Embodiment 1 of a decoding apparatus according to the present invention; FIG.

9 is a schematic structural diagram of Embodiment 2 of an encoding apparatus according to the present invention;

Figure 10 is a schematic structural diagram of Embodiment 2 of the decoding device of the present invention;

Figure 11 is a schematic structural view of Embodiment 3 of the invention encoding device;

Figure 12 is a schematic structural diagram of Embodiment 3 of the decoding device of the present invention;

Figure 13 is a schematic structural view of Embodiment 4 of the invention encoding device;

14 is a schematic diagram of a filtering structure using a Harr wavelet-based wavelet transform;

Figure 15 is a schematic diagram of time-frequency division obtained by using Harr wavelet-based wavelet transform;

16 is a schematic structural diagram of Embodiment 4 of the invention decoding apparatus;

Figure 17 is a schematic structural view of Embodiment 5 of the coding apparatus of the present invention;

18 is a schematic structural diagram of Embodiment 5 of the decoding device of the present invention;

Figure 19 is a schematic structural view of Embodiment 6 of the invention encoding device;

20 is a schematic structural diagram of Embodiment 6 of the decoding device of the present invention;

Figure 21 is a schematic structural view of Embodiment 7 of the invention encoding device;

Figure 22 is a schematic structural diagram of Embodiment 7 of the decoding device of the present invention;

Figure 23 is a schematic structural view of Embodiment 8 of the inventive encoding device;

Figure 24 is a schematic structural view of Embodiment 9 of the coding apparatus of the present invention;

Figure 25 is a schematic structural view of Embodiment 10 of the encoding apparatus of the present invention;

Figure 26 is a schematic structural view of Embodiment 11 of the encoding apparatus of the present invention;

Figure 27 is a schematic structural view of Embodiment 12 of the encoding apparatus of the present invention;

Figure 28 is a schematic structural view of Embodiment 13 of the encoding apparatus of the present invention;

Figure 29 is a schematic structural view of Embodiment 14 of the encoding apparatus of the present invention;

Figure 30 is a schematic structural diagram of Embodiment 8 of the decoding apparatus of the present invention;

Figure 31 is a schematic structural view of Embodiment 15 of the inventive encoding device;

Figure 32 is a schematic structural view of Embodiment 16 of the encoding apparatus of the present invention;

Figure 33 is a schematic structural view of Embodiment 17 of the inventive encoding device;

Figure 34 is a schematic structural view of Embodiment 18 of the encoding apparatus of the present invention;

Figure 35 is a schematic structural view of Embodiment 19 of the encoding apparatus of the present invention;

Figure 36 is a schematic structural diagram of Embodiment 9 of the decoding apparatus of the present invention;

Figure 37 is a schematic structural diagram of Embodiment 10 of the decoding apparatus of the present invention;

Figure 38 is a schematic structural diagram of Embodiment 11 of the decoding device of the present invention;

39 is a schematic structural diagram of Embodiment 12 of a decoding apparatus according to the present invention;

Figure 40 is a block diagram showing the structure of a thirteenth embodiment of the decoding apparatus of the present invention.

Detailed ways

FIG. 1 to FIG. 4 are schematic structural diagrams of several encoders of the prior art, which have been introduced in the background art, and are not described herein again.

It should be noted that, for convenience and clarity of the present invention, the specific embodiment of the following codec apparatus is described in a corresponding manner, but does not indicate that the encoding apparatus and the decoding apparatus must be corresponding.

As shown in FIG. 5, the audio encoding apparatus provided by the present invention includes a resampling module 50, a psychoacoustic analysis module 51, a time-frequency mapping module 52, a quantization and entropy encoding module 53, a band extension module 54, and a bit stream multiplexing module 55; The resampling module 50 is configured to resample the input audio signal; the psychoacoustic analysis module 51 The masking threshold and the signal mask ratio of the audio signal after resampling are calculated, and the signal type is analyzed; the time-frequency mapping module 52 is configured to convert the time domain audio signal into frequency domain coefficients; and the quantization and entropy encoding module 53 is used in the psychoacoustic analysis module 51. The frequency domain coefficients are quantized and entropy encoded under the control of the output mask ratio, and output to the bit stream multiplexing module 55. The band expansion module 54 is configured to analyze the input audio signal over the entire frequency band to extract the high frequency portion. The spectral envelope and its characteristics related to the frequency portion are output to the bit stream multiplexing module 55; the bit stream multiplexing module 55 is used to expand the resampling module 50, the quantization and entropy encoding module 53 and the frequency band The data output by module 54 is multiplexed to form an audio coded stream. '

The digital audio signal is resampled in the resampling module 50, the sampling rate is changed, and the resampled signal is input to the psychoacoustic analysis module 51 and the time-frequency mapping module 52, respectively, and the frame audio is calculated in the psychoacoustic analysis module 51. The masked threshold and the signal mask ratio of the signal are then passed to the quantization and entropy encoding module 53 as a control signal; on the other hand, the audio signal in the time domain is converted to frequency domain coefficients by the time-frequency mapping module 52. The above-described frequency domain coefficients are quantized and entropy encoded in the quantization and entropy coding module 53 under the control of the mask ratio output by the psychoacoustic analysis module 51. On the other hand, the original digital audio signal is subjected to analysis by the band expansion module 54, and the spectral envelope and spectral characteristic parameters of the high frequency portion are obtained and output to the bit stream multiplexing module 55. The encoded data and control signals are multiplexed in a bit:; jfc multiplexing module 55 to form an enhanced audio coded code stream.

The respective constituent modules of the above audio encoding device will be described in detail below.

The resampling module 50 is used to resample the input audio signal, and the resampling includes upsampling and downsampling. The following sampling is used as an example to illustrate resampling. In this embodiment, the resampling module 50 includes a low pass filter and a down sampler, wherein the pass filter is used to limit the frequency band of the audio signal, eliminating aliasing that may be caused by downsampling. The input audio signal is low-pass filtered and downsampled. Assuming that the input audio signal is s (n) and the output filtered by the low-pass filter with the impulse response h (n) is V (η), then v(n) = h(k)s(n - k ); pair (η)

k=-∞

The sequence after downsampling is η, then x(m) = v(Mm) = 3⁄4 h(k)s{Mm - k). Thus, the sample rate of the resampled audio signal fe is reduced by a factor of M compared to the original input audio signal <?^>.

The psychoacoustic analysis module 51 is mainly used for calculating the masking value, the mask ratio and the perceptual entropy of the input audio signal, and analyzing the signal type. The perceptual entropy calculated by the psychoacoustic analysis module 51 can dynamically analyze the number of bits required for the current frame signal to be transparently encoded, thereby adjusting the bit allocation between frames. The psychoacoustic analysis module 51 outputs the mask ratio of each sub-band to the quantization and entropy coding module 53, which controls it.

The time-frequency mapping module 52 is configured to implement the transformation of the audio signal from the time domain signal to the frequency domain coefficient, and is composed of a filter bank, and specifically may be a discrete Fourier transform (DFT) filter bank, a discrete cosine transform (DCT) filter bank, Modified discrete cosine transform (MDCT) filter bank, cosine modulated filter bank, wavelet transform filter bank, etc.

The frequency domain coefficients obtained by the time-frequency mapping are output to the quantization and entropy coding module 53 for quantization and coding processing.

The quantization and entropy encoding module 53 further includes a non-linear quantizer group and an encoder, where the quantizer can be a scalar quantizer or a vector quantizer. The vector quantizer is further divided into two categories: memoryless vector quantizer and memory vector quantizer. For a memoryless vector quantizer, each input vector is independently quantized, independent of the previous vectors; the i-resonant vector quantizer considers the previous vector when quantifying a vector, ie, uses correlation between vectors Sex. The main memoryless vector quantizers include a full search vector quantizer, a tree search vector quantizer, a multilevel vector quantizer, a gain/waveform vector quantizer, and a split mean vector quantizer; the main memory vector quantizer includes predictive vector quantization And finite state vector quantizer. If a scalar quantizer is employed, the non-linear quantizer group further includes M sub-band quantizers. In each sub-band quantizer, the scale factor is mainly used for quantization, specifically: performing nonlinear suppression on all frequency domain coefficients in the M scale factor bands, and then using the scale factor to determine the frequency domain coefficient of the sub-band. Quantization is performed, and the quantized spectrum obtained by the integer is output to the encoder, and the first scale factor in each frame signal is output as a common scale factor to the bit stream multiplexing module 55, and other scale factors are differentially processed with the previous scale factor. Output to the encoder.

The scale factor in the above steps is a constantly changing value, which is adjusted according to the bit allocation strategy. The present invention provides a bit allocation strategy with minimal global perceptual distortion, as follows:

First, each sub-band quantizer is initialized, and the quantized values of the spectral coefficients in all sub-bands are all zero. At this time, the quantization noise of each sub-band is equal to the energy value of each sub-band, and the noise masking ratio R of each sub-band is equal to its signal-to-mask ratio SMR, and the ratio of the number of remaining pixels is 0, and the number of remaining bits is equal to the target. Number of bits

Secondly, finding the subband with the largest noise masking ratio R, if the maximum noise masking ratio R is less than or equal to 1, the scale factor is unchanged, the output is divided into the result, and the bit allocation process ends; otherwise, the corresponding subband quantizer is The scale factor is reduced by one unit, and then the number of bits required to increase the subband Δ5, . (β) is calculated. If the remaining bits of the subband

), confirm the modification of the scale factor, and subtract the remaining bit number A by Δ5 _; (3⁄4), recalculate the noise mask of the subband, and then continue to find the subband with the noise masking ratio NMR, repeat Perform the next steps. If the remaining number of bits of the subband is < Δ5 _; (β), the modification is canceled, the previous scale factor and the remaining number of bits are retained, and finally the allocation result is output, and the bit allocation t process ends.

If a vector quantizer is used, the frequency domain coefficients are composed into a plurality of dimensional vectors and input into the nonlinear quantizer group. For each dimensional vector, the spectral flattening is performed according to the flattening factor, that is, the dynamic range of the spectrum is reduced, and then the vector quantizer is used. The subjective perceptual distance criterion finds the codeword with the smallest distance from the vector to be quantized in the codebook, and passes the corresponding codeword index to the encoder. The flattening factor is adjusted according to the bit quantization strategy of the vector quantization, and the bit allocation of the vector quantization is controlled according to the perceived importance between different subbands.

After the above quantization process, the quantized coefficients and the statistical redundancy of the side information are further removed by the entropy coding technique. Entropy coding is a kind of source coding technology. The basic idea is to give shorter-length codewords to symbols with higher probability of occurrence and longer codewords to symbols with lower probability of occurrence, so that the length of average codewords The shortest. According to Shannon's noiseless coding theorem, if the symbols of the transmitted N source messages are independent, then use

H(x) 1

Appropriate variable length coding, the average length of the code will satisfy +■ where H(x) lg ₂ (D)

Represents the entropy of the source, representing the symbol variable. Since the entropy r>> is the shortest limit of the average codeword length, the above formula indicates that the average length of the codeword is very close to its lower bound entropy HUd, so this variable length coding technique becomes "entropy coding". Huffman entropy encoding major coding, arithmetic coding, or run-length coding method, the entropy encoding may be employed in the present invention, the above-described coding; method of any one of ^five horses.

After the scalar quantizer quantizes the quantized 谙 and the differentially processed scale factor is entropy encoded in the encoder, the codebook serial number, the scale factor coded value and the lossless coded quantized spectrum are obtained, and then the codebook sequence number is entropy encoded to obtain The code book serial number encodes the value, and then outputs the scale factor code value, the code book sequence code value, and the lossless code quantization spectrum to the bit stream multiplexing module 55.

The codeword index obtained by the vector quantizer quantization is subjected to one-dimensional or multi-dimensional entropy coding in the encoder to obtain The coded value of the codeword index is then output to the bitstream multiplexing module 55.

After the original input audio signal is input to the j-band extension module 54, the analysis is performed on the entire frequency band, and the spectral envelope of the high-frequency portion and its characteristics related to the low-frequency portion are extracted, and output as the band extension control information to the bit stream multiplexing module. 55.

The basic principle of band extension is: For most audio signals, the characteristics of the high-frequency part have a strong correlation with the characteristics of the low-frequency part, because the high-frequency part of the jt匕 audio signal can be effectively reconstructed through its low-frequency part. Thus, the high frequency component of the audio signal may not be transmitted. To ensure proper reconstruction of the high frequency portion, only a small number of band extension control signals are transmitted in the compressed audio stream.

The band expansion module 54 includes a parameter extraction module and a spectral envelope extraction module, and the input signal enters the parameter extraction module to extract parameters in different time-frequency regions i or representing spectral characteristics of the input signal, and then in the spectral envelope extraction module, The time-frequency resolution estimates the spectral envelope of the high frequency portion of the signal. In order to ensure that the time-frequency resolution is best suited to the characteristics of the current input signal, the spectral resolution of the spectral envelope can be chosen freely. The parameters of the input signal spectrum characteristics and the spectral envelope of the high frequency portion are sent to the bit stream multiplexing module 55 for multiplexing as the output of the band extension.

The bitstream multiplexing module 55 receives the code stream including the common scale factor, the scale factor coded value, the codebook sequence number coded value, and the lossless coded quantized spectrum output by the quantization and entropy coding module 53 or the coded value of the codeword index and the band extension. After the information output by the module 54 is multiplexed, a compressed audio data stream is obtained.

The encoding method based on the above encoder specifically includes: analyzing an input audio signal over the entire frequency band, extracting a high spectral envelope and a signal spectral characteristic parameter as a frequency band extension control signal; resampling the input audio signal; calculating the resampling Signal-to-mask ratio of the signal; time-frequency mapping of the resampled signal to obtain frequency domain coefficients of the audio signal; quantization and entropy coding of the frequency domain coefficients; multiplexing of the band extension control signal and the encoded audio code stream , get the compressed audio stream.

The resampling process consists of two steps: limiting the frequency band of the audio signal; and multiplying the audio signal of the limited band.

There are many methods for time-frequency transform of time-domain audio signals, such as discrete Fourier transform (DFT), discrete cosine transform (DCT), modified discrete cosine transform (MDCT), chord-modulated filter bank, wavelet transform, and so on. The following is a description of the process of time-frequency mapping with modified discrete cosine transform MDCT and cosine modulation filtering as examples.

For the case of time-frequency transform using the modified discrete cosine transform (MDCT), first select the time domain signals of the previous frame and the current frame samples, and then window the time domain signals of the two frames of the two frames, and then The MDCT transform is performed on the windowed signal to obtain a frequency domain coefficient.

The impulse response of the MDCT analysis filter is:

Then the MDCT transform is: X(k) = function; x (n) MDCT

Transformed input time domain signal; J > is the output frequency domain signal of the MDCT transform.

In order to satisfy the condition of complete signal reconstruction, the window function of MDCT transformation must satisfy the following two conditions: w(2M - l - n) = w(n) and w ² (n) + w ² (n + M) = l .

In practice, the S ine window can be used as a window function. Of course, the above limitation on the window function can also be modified by using a bi-orthogonal transform with a specific analysis filter and synthesis filter. For the case of time-frequency transform using cosine modulation filtering, first select the time domain signal of the previous frame sample and the current frame sample, and then perform windowing operation on the time domain signals of two samples of the two frames, and then The windowed signal is subjected to cosine modulation conversion to obtain a frequency domain coefficient.

The impulse response of the traditional cosine modulation filtering technique is

n=0X---,N,- 1

Where 0 ≤ <M-1, 0 ≤ n < 2KM-\, is an integer greater than zero, 1 ^. Assume that the analysis window of the M sub-band cosine-modulated filter bank (analytical prototype filter) has an impulse response length of N. , integrated window (integrated prototype filter)

The impact response length is N _s . When the analysis window and the synthesis window are equal, that is, _α θ) = _Α θ), and N _a = N, the cosine-modulated filter bank represented by the above two equations is an orthogonal filter bank, at which time the matrix 7 and F ( [H] _nk =h _k (n), [F] _llk = f _k {n) ) is an orthogonal transformation matrix. To obtain a linear phase filter bank, a symmetric window ' is further defined'; foot p _a (2KM-\-n) = p _a (n). In order to ensure the complete reconfiguration of orthogonal and biorthogonal systems, the window function needs to be full of certain conditions, see the literature "Multirate Systems and Filter Banks", PP Vaidynathan, Prentice Hall, Englewood Cliffs, NJ, 1993.

Calculating the masking threshold and the mask ratio of the resampled signal includes the following steps:

The first step is to map the signal to the time domain to frequency i or. The fast Fourier transform and the hanning window technique can be used to convert the time domain data into frequency domain coefficients X[k , expressed as amplitude and phase ί3⁄4 ]; :] = Γ[ : ^Μ , then the energy of each subband Is the sum of the energy of all the spectral lines in the subband, ie e[b]=, where sum and ^ respectively represent the upper and lower boundaries of the subband 6.

The second step is to determine the tonal and non-sounding components in the signal. The tonality of the signal is estimated by inter-predicting each spectral line. The Euclidean distance between the predicted and true values of each spectral line is mapped to an unpredictable measure, and the highly predictive spectral components are considered to be tones. Very strong, and low-predictive spectral components are considered to be noise-like.

The magnitude r _pred and the phase φ _{ρ ά of the} predicted value can be expressed by the following formula: r _pred [k] = r _M [k] + (r _f _! [k] - r _t _ ₂ [^] ) , indicates the coefficient of the current frame; -1 indicates the coefficient of the previous frame; - 2 indicates the coefficient of the first two frames.

Then, the formula for calculating the unpredictability is: C —

Among them, the Euclidean distance ifet(jq), jr _prerf [i:]) is calculated by the following formula: dist(X[k], X _pred [k]) = - X _pred [k]

= i ^r ik] cos(( ]) - r _pred [k] cos [A:])) ² +- (r[k] sin(^[^])- r _pred [k] sin( [ ])) ² † ² Therefore, the unpredictability of each sub-band is the kk _{h of the} unpredictability of the energy of all the lines in the sub-band

Weighted sum, ie c[b]= 2 :]Γ ² [:]. The subband energy eO] and the unpredictability are respectively convoluted with the extension function to obtain the subband energy extension and the preband unpredictability extension^[6], and the extension function of the mask pair subband 6 is expressed as a function to eliminate the extension function. The effect on the energy transformation needs to be normalized to the sub-band unpredictability extension 6], and the normalized result is expressed as [b] = ^3⁄4 by S [b]. Similarly, to eliminate the effect of the extension function on the energy of the subband, define the normalized energy extension? | ] is [6] = ^, where normal n[b]

6- max

The factor is: "[b]= is the number of subbands divided by the frame signal.

;=1 According to the normalized unpredictability extension jb], the pitch of the subband can be calculated: t[b] = -0.299 - 0.43 log _e (c _s [b]) , and 0 ≤ t[6] ≤ l . At that time, it indicates that the sub-band signal is a pure tone; when 6] = 0, it indicates that the sub-band signal is white noise.

The third step is to calculate the signal-to-noise ratio (SNR) of each sub-band. Set the value of Noise-Masking-Tone (NMT) of all sub-bands to 5dB, and the value of Tone-Making-Noise (TMN) to 18dB, so that noise is not perceived. , then the signal-to-noise ratio SNR [b] required for each sub-band is SA3⁄4[6] = 1 St[b] + 6(1 - t[b]).

The fourth step is to calculate the masking value of each sub-band and the perceptual entropy of the signal, and perform signal type analysis. According to the normalized signal amount of each sub-band obtained by the foregoing steps and the required signal-to-noise ratio SNR, the noise energy threshold [6] of each sub-band is calculated as / ] = [&] 10 - ^6]/10 .

In order to avoid the effects of pre-echo, the noise energy threshold of the current frame? [6] and the noise energy threshold of the previous frame.

" _prev [b] compares and obtains the masked threshold of the signal as "[0] = minO ], 2 _rev [b]), which ensures that the masking threshold is not generated by high-energy shock at the near end of the analysis window. And there is a deviation.

Further, considering the influence of the static masking threshold ^i ?r[6], the masking threshold for selecting the final signal is static. The value of the masking threshold and the masked threshold calculated above is greater, ie

. Then the perceptual entropy calculated using the following formula, i.e. _{pe ^ - ifbwidthb X log lQ (} n [b] / (e [b] + l))), where represents the number of spectral lines cbwidth _b included in each sub-band.

Determine whether the perceptual entropy of a certain frame signal exceeds the specified threshold PE-SWITCH. If it exceeds, the frame signal is fast-changing type, otherwise it is a slow-change type.

Step 5: Calculate the mask ratio (S i gna l - to- Ma sk Ra tio , SMR for short) of each sub-band signal. each

After the mask ratio of the subband signal is obtained, the frequency domain coefficients are quantized and entropy encoded according to the mask ratio, wherein the quantization may be scalar quantization or vector quantization.

The scalar quantization includes the following steps: nonlinearly compressing the frequency domain coefficients in all scale factor bands; and then quantizing the frequency domain coefficients of the subbands by using the scale factor of each subband to obtain a quantized spectrum represented by an integer; selecting each frame The first scale factor in the signal is used as a common scale factor; other scale factors are differentially processed from their previous scale factor.

The vector quantization comprises the following steps: constituting a plurality of multi-dimensional vector signals by frequency domain coefficients; performing spectral flattening according to a flattening factor for each dimension vector; finding a code having the smallest distance from the vector to be quantized in the codebook according to the main 覌 perceptual distance measure criterion Word, get its codeword index.

The entropy coding step comprises: entropy coding the quantized spectrum and the differentially processed scale factor to obtain a codebook serial number, a scale factor coded value, and a lossless coded quantized spectrum; entropy coding the codebook sequence number to obtain a codebook serial number coded value.

Or: A one-dimensional or multi-dimensional entropy encoding of the codeword index, which will be the encoded value of the codeword index.

The above entropy coding method may use any of the existing Huffman codec, arithmetic coding or run length coding.

After the quantization and entropy coding processing, the encoded audio code stream is obtained, and the code stream is multiplexed together with the common scale factor and the band extension control signal to obtain a compressed audio code stream.

Figure 6 is a block diagram showing the structure of an audio decoding device of the present invention. The tone decoding apparatus includes a bit stream demultiplexing module 601, an entropy decoding module 602, an inverse quantizer group 603, a frequency-time mapping module 604, and a band extension module 605. After the compressed audio code stream is demultiplexed by the bit stream demultiplexing module 601, corresponding data signals and control signals are obtained, which are output to the entropy decoding module 602 and the band extension module 605. The teaching signal and the control signal are subjected to decoding processing in the entropy decoding module 602 to recover the quantized value of the spectrum. The upper quantized value is reconstructed in the inverse quantizer group 603, and the inverse quantized spectrum is obtained. The inverse quantized spectrum is output to the frequency-time mapping module 604, and the time domain audio signal is obtained through frequency-time mapping, and is in the band extension module. In the middle 605, the high frequency signal portion is reconstructed to obtain a wide-band time domain audio signal.

The bit stream demultiplexing module 601 decomposes the compressed audio code stream to obtain corresponding data signals and control signals, and provides corresponding decoding information for other modules. After the compressed audio data stream is demultiplexed, the signals output to the entropy decoding module 602 include a common scale factor, a scale factor encoding straight, a codebook serial number encoded value, and a lossless encoded quantized spectrum, or an encoded value of a codeword index; The control information is extended to the band extension module 605.

In the encoding apparatus, if the scalar quantizer is used in the quantization and entropy coding module 53, in the decoding apparatus, the 嫡 decoding module 602 receives the common scale factor, the scale factor coded value output by the bit stream demultiplexing module 601, The code book serial number coded value and the lossless coded quantized spectrum are then decoded by the code book number, the spectral coefficient is decoded, and The scale factor is decoded, the quantized spectrum is reconstructed, and the integer representation of the scale factor and the quantized value of the spectrum are output to the inverse quantizer group 603. The decoding method adopted by the entropy decoding module 602 corresponds to an entropy-encoded encoding method in the encoding device, such as Huffman decoding, #术 decoding, or run-length decoding.

After receiving the quantized value of the spectrum and the gas expression of the scale factor, the inverse quantizer group 603 inversely quantizes the transmitted quantized value into a non-scaled reconstructed spectrum (inverse quantized spectrum), and outputs inverse quantization to the frequency-time mapping module 604. Spectrum. The inverse quantizer group 603 may be a uniform quantizer group or a non-uniform quantizer group realized by a companding number. In the encoding apparatus, the quantizer group employs a scalar quantizer, and the inverse quantizer group 603 also employs a scalar inverse quantizer in the numerator apparatus. In the scalar inverse quantizer, the quantized values of the spectra are first nonlinearly expanded, and then each spectral factor is used to obtain all the spectral coefficients (inverse t-spectrum) in the corresponding scale factor bands.

If the vector quantizer is used in the quantization and domain coding module 53, in the decoding apparatus, the entropy decoding module 602 receives the coded value of the codeword index output by the bitstream demultiplexing module 601, and the coded value of the ^1 codeword index is used. The entropy decoding method corresponding to the entropy encoding method at the time of encoding is decoded to obtain a corresponding codeword index.

The codeword index is output to the inverse quantizer group 603, and the quantized value (inverse quantization 侮) is obtained by querying the codebook, and output to the frequency-time mapping module 604. The inverse quantizer group 603 employs an inverse vector quantizer.

The inverse quantization spectrum is processed by the mapping of the frequency-time mapping module 604 to obtain a time domain audio signal of a low frequency band. The frequency-time mapping block 604 may be an inverse discrete cosine transform (IDCT) filter bank, an inverse discrete Fourier transform (IDFT) filter bank, an inverse modified discrete cosine transform (IMDCT) filter bank, an inverse wavelet transform filter bank, and Cosine modulating wave group, etc.

The band extension module 605 receives the band extension information output by the bit stream demultiplexing module 601 and the time domain audio signal of the low frequency band of the frequency-time mapping module 604, and reconstructs the high frequency signal part by spectrum shifting and high frequency adjustment. Output a wideband audio signal.

The decoding method based on the above solution includes: demultiplexing the compressed audio code stream to obtain data information and control information; performing entropy decoding on the upper information to obtain a quantized value; performing inverse quantization processing on the quantized value of the spectrum to obtain an inverse Quantization spectrum; 4 inverse quantization spectrum for frequency-time mapping, to obtain a low-frequency time domain audio signal; according to the band extension control signal, reconstruct the high-frequency portion of the time domain audio signal to obtain a wide-band audio signal.

If the demultiplexed information includes a codebook serial number code value, a common scale factor, a scale factor coded value, and a lossless coded quantized spectrum, it indicates that the spectral coefficients are quantized using a scalar quantization technique in the encoding device, and the entropy decoding step is performed. The method comprises: decoding the code number of the code book to obtain the code book number of all the scale factor bands; decoding the quantized coefficients of all the scale factor bands according to the code book corresponding to the code book serial number; decoding the scale factors of all the scale factor bands, reconstructing the quantization i Lu. The entropy decoding method adopted in the above process corresponds to an entropy coding method in the coding method, such as a run length decoding method, a Huffman decoding method, an arithmetic decoding method, and the like.

In the following, the process of entropy decoding is illustrated by using the run length decoding method to decode the code book number, using the Huffman decoding method to decode the quantized coefficients, and using the Huffman decoding method to decode the scale factor.

First, the codebook number of all scale factor bands is obtained by the run-length decoding method. The decoded codebook sequence number is an integer of a certain interval, and the interval is [0, 11], then only the valid range is set, that is, 0. The codebook number between 11 and 11 corresponds to the corresponding spectral coefficient Huffman codebook. For the all-zero sub-band, you can select the corresponding code number of the code book, and the optional 0 number of the type.

After decoding the codebook number of each scale factor band, the quantized coefficients of all scale factor bands are decoded using the spectral coefficient Huffman codebook corresponding to the codebook number. If the codebook number of a scale factor is within the valid range, and the embodiment is between 1 and 11, then the codebook number corresponds to a spectral coefficient codebook, and the codebook is decoded from the quantized spectrum to obtain the scale factor band. The codeword index of the quantized coefficients, and then unpacked from the codeword index To the quantized coefficient. If the codebook number of the scale factor band is not between 1 and 11, the codebook number does not correspond to any spectral coefficient codebook, and the quantized coefficient of the scale factor band is not decoded, and the quantized coefficients of the subband are all set to zero.

The scale factor is used to reconstruct the spectral value based on the inverse quantized spectral coefficient. If the codebook number of the scale factor is in the effective range, each codebook number corresponds to a scale factor. When decoding the above scale factor, first reading the code stream occupied by the first scale factor, and then performing Huffman decoding on other scale factors, and sequentially obtaining the difference between each scale factor and the previous scale factor, The difference is added to the previous scale factor value to obtain each scale factor. If the quantized coefficients of the current subband are all zero, the scale factor of the subband does not need to be decoded.

After the above entropy decoding process, an integer representation of the spectral quantized value and the scale factor is obtained, and then the quantized value of the spectrum is inversely quantized to obtain an inverse quantized spectrum. The inverse quantization process includes: nonlinearly expanding the quantized values of the spectra; and obtaining all the spectral coefficients (inverse quantization) in the corresponding scale factor bands according to each scale factor.

If the demultiplexed information includes the coded value of the codeword index, it indicates that the coding device uses the vector quantization technique to quantize the spectral coefficients, and the entropy decoding step includes: adopting an entropy corresponding to the entropy coding method in the encoding device. The decoding method decodes the encoded value of the codeword index to obtain a codeword index. The codeword index is then inversely quantized to obtain an inverse quantized spectrum.

The method of performing frequency-time mapping processing on the inverse quantization spectrum corresponds to the time-frequency mapping processing method in the encoding method, and may be an inverse discrete cosine transform (IDCT), an inverse discrete Fourier transform (IDFT), or an inverse modified discrete cosine transform ( IMDCT), inverse wavelet transform and other methods are completed.

The inverse-corrected discrete cosine transform IMDCT is taken as an example to illustrate the frequency-time mapping process. The frequency-time mapping process consists of three steps: IMDCT transformation, time domain windowing, and time domain superposition.

First, the IMDCT transform is performed on the pre-predicted spectrum or the inverse quantized spectrum to obtain the transformed time domain signal x, ., „. IMDCT

,

The transformed expression is: where, represents the sample number,

And 0 ≤ w < N , indicating the number of time domain samples, the value is 2048, 2 + 1) / 2; / indicates the frame number; r indicates the spectrum number.

Secondly, the time domain signal obtained by the IMDCT transform is windowed in the time domain. To satisfy the full reconstruction condition, the window function & must satisfy the following two conditions: w(2M-l- ") = w(") and w ² (") + w ² (w + M) = l.

Typical window functions are Sine windows, Kaiser-Bessel windows, and the like. The present invention employs a fixed window function whose window function is: w( -£) = cos(pl/2*( +Q.5)/N-0.94*sin (2*/? / * (1+0.5) )) / (2*pi))) , where k = 0... Nl; & represents the kth coefficient of the window function, with w(k) = w (2*^-1-1) ; The number of samples, which is 1024. Alternatively, the scalar orthogonal transform can be used to modify the above-mentioned restrictions on the window function using a specific analysis filter and synthesis filter.

Finally, the windowed time domain signal is superimposed to obtain a time domain audio signal. Specifically, the first m samples of the signal obtained after the windowing operation and the subsequent samples of the previous frame signal are overlapped and added to obtain m output time domain audio samples, timeSam _ln = preSam _in + preSam^wn, Which represents the frame number, Indicates the sample number, with 0 ≤ " ≤ and N is 2048.

2

After obtaining the time domain audio signal, the high frequency portion of the audio signal is reproduced based on the frequency extension spread control information and the time domain audio signal to obtain a wideband audio signal.

Figure 7 is a schematic illustration of a first embodiment of an encoding device of the present invention. This embodiment adds a difference and difference stereo (M/S) encoding module 56 on the basis of FIG. 5, which is located between the output of the time-frequency mapping module 52 and the input of the quantization and entropy encoding module 53, the psychoacoustic analysis module. 51 to the output signal type analysis results. For multi-channel signals, the psychoacoustic analysis module 51 calculates the masking threshold for the sum and difference channels in addition to the masking threshold for the audio signal mono, and outputs to the quantization and entropy code module 53. The and difference stereo encoding module 56 can also be located between the quantizer group and the encoder in the quantization and entropy encoding module 53.

And the difference stereo encoding module 56 converts the frequency domain coefficients/residual sequences of the left and right channels into the difference channel frequency domain coefficients/residual sequences by using the correlation between the two channels of the channel pair to achieve The purpose of improving coding efficiency and stereo panning is therefore only applicable to channel-pair signals with consistent word types. If it is a mono signal or a channel pair signal with inconsistent signal type, .] does not perform and differential stereo encoding.

The encoding method based on the encoding apparatus shown in FIG. 7 is basically the same as the encoding method based on the encoding apparatus shown in FIG. 5, except that the following steps are added: Before the quantization and entropy encoding processing of the frequency domain coefficients, it is judged whether the audio signal is Multi-channel signal, if it is a multi-channel signal, it is judged whether the signal types of the left and right channel signals are consistent. If the signal types are the same, it is judged whether the scale factor bands corresponding to the two channels satisfy the difference and the stereo coding conditions. If it is satisfied, it is stereo encoded, and the frequency of the difference channel is obtained: 3⁄4 coefficient; if not, no and difference stereo coding is performed; if it is a mono signal or a signal type inconsistent multi-channel signal , the frequency domain coefficients are not processed.

And the difference stereo coding can be applied before the quantization process, before the entropy coding, that is, after the quantization of the frequency domain coefficients, whether the audio signal is a multi-channel signal, such as a multi-channel signal. , determining whether the signal classes of the left and right channel signals are consistent. If the signal types are the same, it is determined whether the scale factor bands corresponding to the two channels satisfy the difference and the sound coding conditions, and if they are satisfied, the difference is performed. Stereo encoding; if not, no and difference stereo encoding processing is performed; if it is a mono signal or a multi-channel signal with inconsistent signal types, the frequency domain teaching is not performed and the differential stereo encoding processing is performed.

There are many methods for judging whether the scale factor band can be performed and differential stereo coding. The judgment method adopted by the present invention is: by K-L transform. The specific judgment process such as "f:

If the spectral coefficient of the left channel scale factor band is 7, the spectral factor of the corresponding scale factor band of the right channel is r (k) , and its correlation matrix

C _rr =丄^; r(A} * r{k); is the number of spectral lines of the scale factor

N k=0

Perform K-L transformation on the correlation matrix C to get J

r

RCRi = A =

2C, „

The rotation angle a satisfies tan(2) = when β = ±, it is the sum and difference encoding mode.

c _ll - o _r Therefore, when the absolute value of the rotation angle a deviates; when r/4 is small (for example, 3 /16 < |^| < 5 /16 ), the two-channel tooth can be subjected to differential and stereo coding between the scale factor bands.

If the sum and difference stereo coding are applied before the quantization process, the frequency domain coefficients in the scale factor band corresponding to the left and right channels are replaced by the frequency domain coefficients of the linear transform and the difference channel:

Wherein, M represents the sum frequency domain coefficient; S represents the difference channel frequency domain coefficient; L represents the left channel frequency domain coefficient; and R represents the right channel frequency domain coefficient.

If the sum and difference stereo coding are applied after quantization, the quantized frequency domain coefficients of the left and right channels in the scale factor band are replaced by the frequency domain coefficients of the linear transform and the difference channel:

Where: A represents the quantized sum channel frequency domain coefficient; represents the quantized difference channel frequency domain coefficient; £ quantized left channel frequency domain coefficient; represents the quantized right channel frequency domain coefficient.

After the quantization and the difference stereo coding are placed, the phase of the left and right channels can be effectively removed, and since the quantization is performed, lossless coding can be achieved.

Fig. 8 is a schematic diagram of the first embodiment of the decoding apparatus. The decoding apparatus adds a difference and stereo decoding module 606, based on the decoding apparatus shown in FIG. 6, between the output of the inverse quantizer group 603 and the input of the frequency-time generating module 604, and receives the bit stream demultiplexing. The signal type analysis result and the cover stereo control signal output by the module 601 are used to convert the inverse quantized spectrum of the difference channel and the ϋ quantized spectrum of the left and right channels according to the above control information.

In the sum and difference stereo control signals, there is a flag to indicate whether the current channel pair needs to be decoded by the cover stereo. If necessary, there is also a flag on each scale factor band indicating whether the corresponding scale factor is required or not. The stereo decoding, and difference stereo decoding module 606 determines whether the quantized values of the inverse quantized spectrum/spectrum in some scale factor bands need to be subjected to differential stereo decoding based on the flag bits of the scale factor band. If the sum and difference stereo coding are performed in the encoding device, the inverse quantized spectrum must be subjected to the sum and difference stereo I code in the decoding device.

The sum and difference stereo decoding module 606 can also be located between the output of the entropy decoding module 602 and the input of the inverse quantizer group 603, and receive the difference stereo control signal and signal type output from the bit stream demultiplexing module 601.

The decoding method based on the decoding apparatus shown in FIG. 8 is basically the same as the decoding method based on the decoding apparatus shown in FIG. 6, except that the following steps are added: After the inverse quantization spectrum is obtained, if the signal type analysis result indicates that the signal types are the same, Determining whether inverse quantization spectrum and differential stereo decoding are required according to the difference stereo control signal; if necessary, determining whether the scale factor band requires and difference stereo decoding according to the flag bits on each scale factor band, if necessary, The inverse quantized spectrum of the difference channel in the scale factor band is converted into the inverse quantized spectrum of the left and right channels, and then subjected to subsequent processing; if the signal type is inconsistent or does not need to be performed with the difference stereo win code, the inverse quantization spectrum is not performed. Processing, direct processing.

And the difference stereo decoding can also be performed after the entropy decoding process and before the inverse quantization process, that is: when the language is obtained After the quantized value, if the signal type analysis result indicates that the letter~ type is consistent, it is judged according to the difference and the stereo control signal whether the quantized value of the spectrum needs to be performed and the difference stereo code; if necessary, according to each scale The flag bit determines whether the scale factor band needs to be decoded with the difference body sound, and if necessary, converts the quantized value of the spectrum of the difference channel in the scale dice band into the quantized value of the spectrum of the left and right channels, and then performs subsequent Processing; The word signal type is inconsistent or does not need to be performed and the difference stereo: ^ code, then the paired quantized value is not processed, r is followed by subsequent processing.

If the difference and the stereo decoding are after the entropy decoding and before the inverse, the left and right channels are obtained by using the following operations in the frequency factor of the scale factor band and the frequency domain of the difference channel.

The quantized sum channel frequency domain coefficients; S represents the quantized channel frequency domain coefficients; Z represents the quantized left channel frequency domain coefficients; ^ represents the quantized right channel frequency domain coefficients.

If the sum and difference stereo decoding are after inverse quantization, then the right channel is in the frequency domain of the inverse quantization of the subband.

I 1 1 m

According to the following matrix operation, the frequency domain coefficients of the difference channel and the difference channel are found: where:

1 one 1

Channel frequency domain coefficient; s represents the difference channel frequency domain coefficient; 7 table left channel frequency domain coefficient; r represents the right channel frequency domain system.

Fig. 9 is a view showing the construction of a second embodiment of the encoding apparatus of the present invention. This embodiment adds a frequency domain linear prediction and vector quantization module 57 to FIG. 5 , 7 , wherein the frequency domain linear prediction and vector quantization module 57 is located at the output of the time-frequency mapping module 52 and the input of the quantization and entropy coding sample 53 . The residual sequence is outputted to the quantization and entropy encoding module 53, and the quantized code index is output to the bit stream multiplexing module 55 as side information.

The frequency domain coefficients output by the time-frequency mapping module 52 are transmitted to the frequency domain linear prediction and vector quantization module 57. If the predicted gain value of the frequency domain coefficients satisfies a given condition, J5'J performs linear prediction filtering on the frequency domain coefficients to obtain The prediction coefficient is converted into a line coefficient LSF (Line Spec trum Frequency), and then the codeword index of each codebook is calculated by using the best distortion metric, and the codeword index is transmitted as side information to the ratio ^ The stream multiplexing module 55, and the residual sequence obtained by the predictive analysis is output to the quantization and entropy encoding module 53.

The frequency domain linear prediction and vector quantization module 57 is composed of a linear prediction analyzer, a linear prediction filter, a converter, and a vector quantizer. The frequency domain coefficients are input into a linear pre-conductor for prediction analysis, and the predicted soil and prediction coefficients are obtained. If the value of the prediction gain satisfies certain conditions, the frequency domain coefficients are output to the linear prediction filter to perform linear prediction error filtering. Get the pre-frequency of the frequency domain coefficient: The residual sequence is directly output to the quantity and entropy coding module 53, and the prediction coefficient is converted into a line spectrum pair frequency coefficient LSF by the converter, and then the LSF parameter is sent to the vector quantizer for multi-level vector quantization. The codeword index of the pair is transferred to the bitstream multiplexing module 55.

Frequency domain linear prediction processing of the audio signal can effectively suppress the pre-echo and obtain a large humming gain. Assuming the real signal Jf <3⁄4), its square H il ber t envelope e is expressed as: e(t) = F~ ^l

Among them, the edge language corresponding to the positive frequency component of the signal, that is, the Hi l ber t envelope of the signal is related to the autocorrelation function of the signal. The relationship between the power intensity function of the signal and the autocorrelation function of its time domain waveform is: PS (/) = _F fx(i:) ' jc* (r - t)^r}, so the square of the signal in the time domain Hi l ber t envelope and signal ^" power spectral density function in the frequency domain is mutually dual relationship. It can be seen that the bandpass signal in each certain frequency range, if its H i lber t envelope remains constant, |5, the autocorrelation of adjacent spectral values will also remain constant, this This means that the sequence of spectral coefficients is a sequence of 3⁄4fe states with respect to frequency, so that the spectral value can be processed by predictive coding techniques, and the signal is represented by a common set of prediction coefficients.

The encoding method based on the encoding device shown in FIG. 9 is basically the same as the encoding method based on the encoding device shown in FIG. 5, except that the following steps are added: standard linear predictive analysis is performed on the frequency domain coefficients to obtain prediction gain and prediction coefficients; Determining whether the prediction gain exceeds a threshold value of t, and if so, performing frequency domain linear prediction error filtering according to the prediction coefficient tooth frequency domain coefficient to obtain a prediction residual sequence of the domain coefficient; degenerating the prediction coefficient into a line spectrum pair frequency coefficient, and The multi-level vector quantization processing of the line spectrum on the frequency coefficient line is performed to obtain the side information; the tooth residual sequence is quantized and entropy encoded; if the prediction gain exceeds the set threshold, the frequency domain coefficients are quantized and entropy encoded.

When the frequency domain coefficients are obtained, the standard linear prediction analysis of the frequency domain coefficients is first performed, including the self-correlation matrix and the recursive Levinson-Durb in algorithm to obtain the prediction gain and the prediction coefficient. Then, it is judged whether the prediction gain of the if calculation exceeds a preset threshold, and if it exceeds, linear prediction error filtering is performed on the frequency domain coefficient according to the prediction coefficient; otherwise, the frequency domain coefficient is not processed, and the next step is performed, and the frequency domain coefficient is performed. Quantity ^ 匕匕 and 嫡 encoding.

Linear prediction can be divided into forward prediction and backward prediction. Forward prediction refers to the prediction of the current value by the value before a certain moment, while backward prediction refers to the prediction of the current value by the value after a certain moment. In the following, the linear prediction error filtering is explained as an example. The transfer function of the linear prediction error is ^( ) = 1 -^ '' , where

;=1

A represents the prediction coefficient, which is the prediction order. After passing through the frequency-domain coefficients of the Γ division-frequency transformation, the predicted error E (k) is obtained, which is also called the difference sequence, and the relationship between the two is satisfied.

E(k) = X(k) · A(z) = X{k) - a _t X(k - i).

Thus, after linear prediction error filtering, the frequency domain coefficient X (k) output by the time-frequency transform module can be represented by the residual sequence E 0> and a set of prediction coefficients _ai . After ^^, the set of prediction coefficients α,. is converted into a line spectrum pair frequency coefficient LSF, and multi-level vector quantization is performed thereon. Vectorization selects the best distortion metric (such as the most recent criterion), and searches for the codeword that best matches the temple-quantized LSF parameter vector (residual vector) in each codebook, and the corresponding codeword index is used. Side information is transmitted ±±!. At the same time, the residual sequence is quantized and encoded. According to the linear prediction analysis coding principle, the dynamic range of the residual sequence of the pedigree teaching is smaller than the dynamic range of the original syllabus, so less 1=匕 special number can be allocated in the quantization, or the condition of the same number of bits can be Improved coding gain is obtained.

10 is a schematic diagram of a second embodiment of a decoding apparatus. The decoding apparatus adds an inverse frequency domain linear prediction and vector quantization module 607 to the output and frequency of the inverse quantizer group 603 based on the decoding apparatus shown in FIG. Between the inputs of the time mapping module 604, and the bits: the demultiplexing module 601 outputs inverse frequency domain linear vector quantization control information thereto for inverse quantization processing and inverse linear prediction filtering on the inverse quantization spectrum (residual spectrum) The pre-predicted spectrum is obtained and output to the frequency-time mapping module 604.

In the encoder, frequency domain linear predictive vector quantization techniques are employed to suppress pre-echo and obtain a larger coding gain. Therefore, in the decoder, the inverse frequency domain linear pre-quantization control information outputted by the inverse quantization spectrum and the bit multiplex demodulation module 601 is input to the inverse frequency domain linear prediction and vector quantization module 607 to recover the linear prediction error. . The inverse frequency domain linear prediction and vector quantization module 607 includes an inverse vector quantizer, an inverse transformer, and an inverse linear prediction filter, wherein the inverse vector quantizer is used to inversely quantize the codeword index row to obtain a line spectrum pair frequency coefficient LSF; Then, the line spectrum is inversely converted into a prediction coefficient by a frequency coefficient LSF; the inverse linear prediction filter is used to inversely filter the inverse quantization spectrum according to the prediction coefficient, to obtain a pre-predicted spectrum, and output to the frequency-time mapping module 604.

The decoding method based on the decoding device shown in FIG. 10 is basically the same as the decoding method based on the decoding device shown in FIG. 6, except that the following steps are added: After obtaining the inverse spectrum, it is judged whether or not the inverse quantization is included in the control information. Inverse frequency domain linear prediction vector quantization information, if it is, then inverse vector quantization processing is performed to obtain prediction coefficients, and the inverse quantization spectrum is linearly predicted and synthesized according to the prediction coefficients to obtain a spectrum before prediction; Frequency-time mapping.

After obtaining the inverse quantization spectrum, determining whether the frame signal is subjected to frequency domain linear predictive vector quantization according to the control information, and if so, obtaining the quantized codeword index from the control information; and then obtaining the quantization according to the codeword index The line spectrum pairs the frequency coefficient LSF, and calculates the prediction coefficient; then the inverse quantization spectrum is linearly predicted and synthesized to obtain the spectrum before prediction.

The transfer function fe> used in the linear prediction error filtering process is: Α(ζ) = ί ~^ _α , where: is the prediction coefficient; is the prediction order. Therefore, the residual sequence 0 > is satisfied with the spectrum YU before prediction:

X k) = E(k) --i- = E(A) +∑ _ai X{k - ).

A ^z ) =1

Thus, the residual sequence and the calculated prediction coefficients are synthesized by frequency domain linear prediction, and the pre-predicted spectrum (k) can be obtained, and the pre-predicted spectrum J (k) is subjected to frequency-time ί mapping processing.

If the control information indicates that the frame signal has not undergone frequency domain linear predictive vector quantization, the inverse frequency domain linear predictive vector quantization process is not performed, and the inverse quantized spectrum is directly subjected to frequency-time mapping processing.

Figure 11 is a block diagram showing the structure of a third embodiment of the encoding apparatus of the present invention. This embodiment, based on FIG. 9, adds a difference and difference stereo coding module 56 between the output of the frequency domain linear prediction and vector quantization module 57 and the input of the quantization and entropy coding module 53, to which the psychoacoustic analysis module 51 The signal type analysis result is output, and the masked value of the difference channel is output to the quantization entropy encoding module 53.

The sum and difference stereo coding module 56 may also be located between the quantizer group and the encoder in the quantization entropy coding module 53 to receive the signal type analysis result output by the psychoacoustic analysis module 51.

In the present embodiment, the function and working principle of the and difference stereo coding module 56 are the same as those in FIG. 7, and will not be described here.

The encoding method based on the encoding apparatus shown in FIG. 11 is basically the same as the encoding method based on the encoding apparatus shown in FIG. 9, except that the following steps are added: Before the quantization and entropy encoding processing of the frequency domain system, it is determined whether the audio signal is Multi-channel signal, if it is a multi-channel signal, it is judged whether the signal types of the left and right channel signals are consistent. If the signal types are the same, it is judged whether the scale factor satisfies the coding condition, and if it is satisfied, the scale factor band is performed. And difference stereo coding; if not full, the difference stereo coding process is not performed; if it is a mono signal or a multi-channel signal of which the signal type is inconsistent, the difference stereo coding process is not performed.

And the difference stereo coding can be applied before the quantization and before the entropy coding, that is, after the quantization of the frequency domain coefficients, it is judged whether the audio signal is a multi-channel signal, if it is multi-channel. Signal, then determine whether the signal types of the left and right channel signals are consistent. If the signal types are consistent, determine whether the scale factor band satisfies the coding condition, and if so, perform a differential stereo coding on the ^-factor band; If it is satisfied, the difference and the stereo encoding processing are not performed; if it is a multi-channel signal in which the mono signal signal types are inconsistent, the difference and the stereo encoding processing are not performed.

Figure 12 is a block diagram showing a third embodiment of the decoding apparatus. The decoding apparatus adds a difference and difference stereo decoding module 606 on the basis of the decoding apparatus shown in FIG. 10, between the output of the inverse quantizer group 603 and the input of the inverse frequency domain linear prediction and vector quantization module 607, and the bit stream solution The multiplexing module 601 outputs and the difference stereo control signals thereto.

The sum and difference stereo decoding module 606 may also be located between the output of the entropy decoding module 602 and the input of the inverse quantizer group 603, and receive the sum and difference stereo planting signals output by the bit stream demultiplexing module 601.

In this embodiment, the function and working principle of the difference and stereo decoding module 606 are the same as those in FIG. 8, and details are not described herein again.

The decoding method based on the decoding apparatus shown in FIG. 12 is basically the same as the decoding method based on the decoding apparatus shown in FIG. 10, except that the following steps are added: After the inverse quantization is obtained, if the signal type analysis result indicates that the signal types are the same, Determining whether inverse quantization spectrum and differential stereo decoding are required according to the difference stereo control signal; if necessary, determining whether the scale factor band requires and difference stereo decoding according to the flag bits on each scale factor band, if necessary, The 3⁄4 quantized spectrum of the difference channel in the scale factor band is converted into the inverse quantized spectrum of the left and right channels, and then subjected to subsequent processing; if the signal type is inconsistent or does not need to perform and the difference stereo decoding, the inverse quantization is not processed. , directly follow-up processing.

And the difference stereo decoding can also be performed before the inverse quantization process, that is: after the obtained quantized value, if the signal type analysis result indicates that the signal type is consistent, it is determined according to the difference and the stereo control signal whether the quantized value of the spectrum needs to be Poor stereo decoding; if necessary, judging whether the scale factor band requires and difference stereo decoding according to the flag bit on each scale factor band, if necessary, the quantized value of the spectrum of the sum channel in the scale factor band Converted into the quantized values of the left and right channels, and then perform subsequent processing; if the signal types are inconsistent or do not need to perform and the difference stereo decoding, the quantized values of the spectrum are not processed, and the subsequent processing is directly performed.

Fig. 13 is a view showing the construction of a fourth embodiment of the encoding apparatus of the present invention. This embodiment adds a multi-resolution analysis module 59 on the basis of FIG. 5, wherein the multi-resolution analysis block 59 is located between the output of the time-frequency mapping module 52 and the input of the quantization and entropy coding module 53, psychoacoustic points. The Zhejiang module 51 outputs a signal type analysis result to it.

For fast-changing type signals, in order to effectively overcome the pre-acquisition phenomenon generated in the encoding process and improve the encoding quality, the encoding apparatus of the present invention increases the time resolution of the frequency-domain coefficients of the fast-changing signal by the multi-resolution analyzing module 59. The frequency domain coefficients output by the time-frequency mapping module 52 are input to the multi-resolution analysis module 59. If it is a fast-varying type signal, the frequency domain wavelet transform or the frequency domain modified discrete cosine transform (MDCT > is obtained, and the frequency domain coefficients are obtained. The multi-resolution representation is output to the quantization and entropy coding module 53. If it is a slowly varying odor signal, the frequency domain coefficients are not processed and are directly output to the quantization and entropy coding module 53.

The multi-resolution analysis module 59 performs time-frequency domain re-organization on the input frequency domain data, and improves the time resolution of the frequency domain data at the expense of the frequency precision, thereby automatically adapting the time-frequency characteristics of the fast-changing type signal. The effect of suppressing the pre-echo is achieved. The form of the filter bank in the frequency mapping module 52 can be adjusted at any time. The multi-resolution analysis module 59 includes a frequency domain coefficient transform module and a recombination patch, wherein the frequency domain coefficient transform module is configured to transform the frequency domain coefficients into time-frequency plane coefficients; the reassembly module is configured to use the 频"frequency plane coefficients according to certain rules. The frequency domain coefficient transform module may use a frequency domain wavelet transform filter bank, a frequency domain MDCT transform filter bank, and the like.

The following takes the frequency domain wavelet transform and the frequency domain MDCT transform as an example to illustrate the operation of the multi-segment analysis module 59. Cheng.

1) Frequency domain wavelet transform

Set the time series χ(ζ·), ζ· = 0,1,...,2Μ - 1 , and obtain the ό frequency domain coefficients after time-frequency mapping as X(k), 1=0, 1 M-\ _a The wavelet base of the frequency domain wavelet or wavelet packet transform may be fixed or adaptive.

Let's take the simplest Harr wavelet-based wavelet transform as an example to illustrate the process of multi-resolution analysis of frequency domain and number.

The scale factor of the Harr wavelet basis is, Figure 14 shows the adoption

Schematic diagram of the filtering structure of wavelet transform based on wavelet transform, in which. Indicates 氐 pass filtering (filter coefficient is

) , indicating high-pass filtering (filter coefficient is - =) ), ' ^c I 2' means 1 times lowering V2 V2 V2 V2

Sample operation. For the low-frequency part of the frequency domain coefficient = 0,... No, the wave-wave transform is performed, and the high-frequency part of the frequency-domain coefficient is subjected to Harr wavelet transform to obtain coefficients X ₂ (k) and X _{3 of} different time-frequency intervals ( k), X (k), 5 (;), X^ ^ X. ik) , the corresponding time-frequency plane is divided as shown in Figure 15. Different wavelet bases can be selected, and different wavelet transform structures can be selected for processing, and other similar time-frequency plane partitions are obtained. Therefore, the time-frequency plane division of the signal analysis can be arbitrarily adjusted as needed to meet the analysis requirements of different time and frequency resolutions.

The above-mentioned time-frequency plane coefficients are weighted according to certain rules in the recombination module. For example, the time-frequency plane coefficients may be organized in the frequency direction, the coefficients in each frequency band are organized in the time direction, and then the organized coefficients are according to the sub-window, The order of the scale factor bands.

2) Frequency domain MDCT transform

Let the frequency domain data of the input frequency domain MDCT transform filter bank be X (k), k= 0, 1 , ..., N-1, and perform MDCT transform of M point in sequence for the N point frequency domain data, so that time The frequency of the frequency domain data is reduced, and the time accuracy is correspondingly increased. Different frequency-frequency MDCT transforms can be used in different frequency domains to obtain different time-frequency plane partitions, that is, different time and frequency precisions. The recombination module reorganizes the time-frequency domain data outputted by the frequency MDCT transform filter bank. A recombination method is to first organize the time-frequency plane coefficients in the frequency direction, and the coefficients in each frequency band are organized in the time direction, and then the organization Good coefficients are arranged in the order of the sub-window and scale factor bands.

Based on the encoding method of the encoding apparatus shown in FIG. 13, the basic flow is the same as the encoding method based on the encoding apparatus shown in FIG. 5, except that the following steps are added: If it is a fast-changing type signal, multi-resolution is performed on the frequency domain coefficient. Analysis, then quantizing and entropy coding the multi-resolution representation of the frequency domain coefficients; if not fast-type signals, the frequency domain coefficients are directly quantized and entropy encoded.

The multi-resolution analysis may use a frequency domain wavelet transform method or a frequency domain MDCT transform method. The frequency domain wavelet analysis method comprises: performing wavelet transform on the frequency domain coefficients to obtain time-frequency plane coefficients; and recombining the above-mentioned time-frequency plane coefficients according to a certain rule. The MDCT transform method includes: performing n times MDCT transformation on the frequency domain coefficients to obtain a time-frequency plane coefficient; and recombining the above-mentioned time-frequency plane coefficients according to a certain rule. The method of recombination may include: firstly, the time-frequency plane coefficients are organized in the frequency direction, the coefficients in each frequency band are organized in the time direction, and then the coefficients of the woven fabric are arranged in the order of the sub-window and the scale factor band. Figure 16 is a block diagram showing the structure of a fourth embodiment of the decoding apparatus. The decoding device adds a multi-resolution synthesis module 609 to the base of the anti-horse device shown in FIG. The multi-resolution synthesis module 609 is located between the output of the inverse quantizer 纟JL 603 and the input of the frequency-time mapping module 604 for multi-resolution green combining of the inverse quantized spectrum.

In the encoder, a multi-resolution filtering technique is employed for the fast-changing type signal to improve the temporal resolution of the encoded fast-changing type signal. Accordingly, in the decoder, the multi-resolution synthesis module 609 is required to recover the frequency domain coefficients before the multi-resolution analysis for the fast-changing signal. The multi-resolution synthesis module 609 includes: a coefficient recombination module and a coefficient transformation module, wherein the coefficient transformation module can adopt a frequency domain inverse wavelet transform filter bank or a frequency domain IMDC Ί transform filter group.

Based on the decoding method of the decoding device shown in FIG. 16, the basic flow is the same as the decoding method based on the de-horse device shown in FIG. 6, except that the following steps are added: After the inverse quantization spectrum is obtained, the inverse quantization "^ Multi-resolution synthesis is performed, and the obtained prediction coefficients are subjected to frequency-time mapping.

The following is a description of the multi-resolution synthesis method by taking the frequency domain IMDCT transform as an example, which specifically includes: recombining the inverse quantized spectral coefficients; and performing multiple IMDCT transforms on each coefficient to obtain an inverse i-spectrum before multi-resolution analysis. The process is described in detail below with 128 IMDCT transforms (8 inputs, 16 outputs). First, the 3 inverse quantization spectral coefficients are arranged in the order of sub-window and scale factor bands; then recombined according to the frequency order, so that 128 coefficients of each sub-blind are organized in frequency order. Then, the coefficients arranged by the sub-window are grouped by frequency every 8 groups: ^ "to the organization, each group of 8 coefficients are arranged in time series, so that there are 128 sets of coefficients in the frequency direction. Each group of coefficients is converted by 16 points IMDCT, The 16 coefficients outputted by each group of IMDCT are superimposed and added to obtain 8 frequency domain data. 128 similar operations are performed from the low frequency to the high frequency direction, and 1024 frequency domain coefficients are obtained.

Fig. 17 is a view showing the configuration of a fifth embodiment of the encoding apparatus of the present invention. In this embodiment, based on the encoding apparatus shown in FIG. 13, a frequency domain linear prediction and vector quantization module 57 is added, which is located between the output of the multi-dividing rate analyzing module 59 and the input of the quantization and entropy encoding module 53. The acoustic analysis module 51 outputs a signal type analysis result thereto; the frequency domain linear prediction and vector quantization module 57 is configured to perform linear prediction and multi-level vector quantization on the frequency domain coefficients subjected to the multi-resolution analysis, and output the residual sequence to the quantization and entropy. The encoding module 53 simultaneously outputs the quantized codeword index to the bitstream multiplexing module 55.

Since the frequency domain coefficients are obtained by multi-resolution analysis with time-frequency coefficients divided by specific time-frequency planes, the frequency-domain linear prediction and vector quantization module 57 needs to linearly predict the frequency domain coefficients in each time period. And multi-level vector quantization.

The frequency domain linear prediction and vector quantization module 57 may also be located between the output of the time-frequency mapping module 52 and the input of the multi-resolution analysis module 59, performing linear prediction and multi-level vector quantization on the frequency domain coefficients, and outputting residuals; The resolution analysis module 59 simultaneously outputs the quantized codeword index to the bitstream multiplexing module 55.

The encoding method based on the encoding apparatus shown in Fig. 17 is basically the same as the encoding based on the encoding apparatus shown in Fig. 13; the difference is that the following steps are added: After performing multi-resolution analysis on the frequency domain coefficients, ^The frequency domain coefficient on the time period is subjected to standard linear prediction analysis; whether the prediction gain exceeds the set threshold value; if it exceeds, the frequency domain linear prediction error is filtered on the frequency domain coefficient to obtain the residual of the prediction coefficient and the frequency domain coefficient. The prediction coefficient is converted into a line spectrum pair frequency coefficient, and the line spectrum is subjected to multi-level vector quantization processing to obtain side information; the residual sequence is quantized and entropy encoded; if the prediction gain does not exceed the set threshold Then, the frequency domain coefficients are quantized and entropy encoded. Or before the multi-resolution analysis, the frequency domain coefficients are linearly predicted and multi-level vector quantized, and then the residual sequences are subjected to multi-resolution analysis.

Figure 18 is a block diagram showing the structure of a fifth embodiment of the decoding apparatus. The decoding apparatus adds an inverse frequency domain linear prediction and vector quantization module 607 on the basis of the numerator apparatus shown in FIG. 16, and is located in the inverse quantizer group 603. Between the inputs of the resolution synthesis module 609, and the bitstream demultiplexing module 601 outputs inverse frequency domain linearity vector quantization control information thereto for inverse quantization processing and linear prediction synthesis of the inverse quantization spectrum to obtain a prediction mouth. The spectrum is output to the multi-resolution synthesis module 609.

The inverse frequency domain linear prediction and vector quantization module 607 can also be located between the output of the multi-resolution synthesis module 609 and the input of the page rate-time mapping module 604 for linear pre-processing of the multi-resolution integrated inverse quantization :: 5 is synthesized.

The decoding method based on the decoding device shown in FIG. 18 is the same as the decoding method based on the decoding device shown in FIG. 16, except that the following steps are added: After obtaining the inverse quantization spectrum, it is determined whether or not the control information needs to include the quantized spectrum. After inverse frequency domain linear prediction vector quantization information, if it is included, the inverse vector quantization is used to obtain the prediction coefficients, and the inverse quantization 镨 is linearly predicted and synthesized to obtain the error before prediction; the pre-predictive spectral line multi-resolution synthesis is performed. . It may also be: after obtaining the inverse quantization spectrum, performing multi-resolution synthesis on the inverse quantization spectrum; whether the information of the inverse quantization spectrum needs to be quantized by the inverse frequency domain linear prediction vector is included in the break control information, and if there is, the inverse is performed. The vector quantization process obtains the prediction coefficients, and performs linear prediction synthesis on the inverse quantization spectrum to obtain the spectrum before the ij prediction; the frequency-time mapping is performed before the prediction.

Fig. 19 is a view showing the construction of a sixth embodiment of the encoding apparatus of the present invention. This embodiment is based on the encoding apparatus shown in Fig. 17, with an addition and difference stereo encoding module 56 between the output of the frequency domain linear prediction and vectoring module 57 and the input of the quantization and entropy encoding module 53. The signal type analysis result from the psychoacoustic analysis module 51 is received. The sum and difference stereo coding module 56 may also be located between the quantizer and the encoder in the quantization and entropy coding module 53. In this embodiment, the functions and working principles of the and difference stereo coding module 56 are the same as those in FIG. 11, and are not described herein again.

The encoding method based on the encoding device shown in FIG. 19 is basically the same as the encoding method based on the encoding device shown in FIG. 17, except that the following steps are added: after obtaining the residual sequence, according to whether the audio signal is a multi-channel signal and For signals of the same signal type and satisfying the encoding conditions, it is determined whether or not to perform stereo encoding on the difference; however, subsequent processing is performed. The specific process has been introduced above, and will not be described here.

Fig. 20 is a block diagram showing the structure of the sixth embodiment of the decoding apparatus. The decoding apparatus is based on the decoding apparatus shown in Fig. 18, and the sum and difference stereo decoding module 606 is added between the output of the inverse quantizer group 603 and the input of the inverse frequency domain linearity detecting and vector quantization module 607. The sum and difference stereo decoding module 606 can also be located between the output of the entropy decoding module 602 and the input of the inverse quantizer group 603. The function and working principle of the differential stereo decoding module 6 O 6 in this embodiment are the same as those in FIG. 12 , and details are not described herein again.

The decoding method based on the decoding apparatus shown in FIG. 20 is basically the same as the decoding method based on the decoding apparatus shown in FIG. 18, except that the following steps are added: After the inverse quantization spectrum is obtained, the result and the stereo control are analyzed according to the signal type. The information judges whether the inverse quantization spectrum needs to be subjected to differential stereo decoding; then the subsequent processing is described above, and will not be described here.

Figure 21 is a diagram showing a seventh embodiment of the encoding apparatus of the present invention, which is based on the tomb of Figure 13, with the addition and difference stereo encoding module 56, which is located at the output and quantize of the multiresolution analysis module 59. Between the input of the domain decoding module 53 and the input. The sum and difference stereo encoding module 56 can also be located between the quantizer group and the encoder in the quantization and entropy encoding module 53. The difference and stereo encoding module 56 has been previously detailed and is no longer referred to herein.

The encoding method based on the encoding apparatus shown in FIG. 21 is basically the same as the encoding method based on the encoding apparatus shown in FIG. 13, and the difference is that the following steps are added: After the multi-resolution analysis of the frequency domain coefficients, whether the audio signal is more or not The channel signal is a signal of the same signal type and satisfies the encoding condition, and determines whether it is subjected to and differential stereo encoding; and then performs subsequent processing. The specific process has been introduced above, and will not be described here. Figure 22 is a diagram showing the seventh embodiment of the decoding apparatus. The decoding apparatus adds a difference stereo decoding module 606 between the output of the inverse quantizer group 603 and the input of the multi-resolution synthesis module 609 based on the decoding apparatus shown in FIG. The and difference stereo decoding module 606 can also be located between the output of the entropy decoding module 602 and the input of the inverse quantizer group 603. The and difference stereo decoding module 606 has been previously detailed and will not be obscured here.

The decoding method based on the decoding device shown in FIG. 22 is basically the same as the decoding method based on the decoding device shown in FIG. 16, except that the following steps are added: After the inverse quantization spectrum is obtained, the result and the difference stereo control information are analyzed according to the signal type. It is judged whether or not the inverse quantization spectrum needs to be subjected to differential stereo decoding; then, subsequent processing is performed. The specific process has been introduced above and will not be described here.

Figure 23 is a diagram showing an eighth embodiment of the encoding apparatus of the present invention. The embodiment is based on the encoding apparatus shown in Figure 13, and a signal property analyzing module 510 is added for outputting the resampling module 50. The signal is subjected to signal type analysis, and the resampled signal is output to the time-frequency mapping module 52 and the psychoacoustic analysis module 51, and the signal type analysis result is output to the bit stream multiplexing module 55.

The signal property analysis module 510 performs front and back masking effect analysis based on the adaptive threshold and the waveform prediction to determine whether the signal type is a slow-changing signal or a fast-changing signal, and if it is a fast-changing type signal, continues to calculate related parameter information of the abrupt component. Such as the location of the mutation signal and the strength of the mutation signal.

The encoding method based on the encoding apparatus shown in Fig. 23 is basically the same as the encoding method based on the encoding apparatus shown in Fig. 13, except that the following steps are added: The type of the resampled signal is analyzed as a part of signal multiplexing.

The signal type is determined based on adaptive threshold and waveform prediction for front and back masking effect analysis. The specific steps are: decomposing the input audio data into frames; decomposing the input frame into multiple sub-frames, and searching for each sub-frame. a local maximum point of the absolute value of the PCM data; selecting a subframe peak value in a local maximum point of each subframe; for a certain subframe peak, using a plurality of (typically 3) subframe peaks in front of the subframe to predict relative A typical sample value of a plurality of (typically 4) subframes of the forward delay of the subframe; calculating a difference and a ratio of the peak value of the subframe to the predicted typical sample value; if the predicted difference and the ratio are both greater than The set threshold value is determined to be a sudden signal in the sub-frame, and it is confirmed that the sub-frame has a local maximum peak point of the back masking pre-echo capability, if between the sub-end and the mask peak, 2. 5 ms If there is a sub-frame with a sufficiently small peak, it is judged that the frame signal belongs to the fast-changing type signal; if the predicted difference value and the ratio are not greater than the set threshold, the above steps are repeated until the judgment The frame signal is broken out to be a fast-changing type signal or reaches the last sub-frame. If the last sub-frame is not determined to be a fast-changing type signal, the frame signal belongs to a slowly-changing type signal.

Figures 24 to 28 are schematic views showing the ninth to thirteenth embodiments of the encoding apparatus of the present invention. The above embodiment is based on the coding apparatus shown in FIG. 17, FIG. 19, FIG. 21, FIG. 9 and FIG. 11, respectively, and a signal property analysis module 510 is added for performing signal type analysis on the signal output by the resampling module 50. And outputting the resampled signal to the time-frequency mapping module 52 and the psychoacoustic analysis module 51, and outputting the signal type analysis result to the bit stream multiplexing module 55.

The encoding method based on the encoding apparatus shown in FIGS. 24 to 28 is basically the same as the encoding method based on the encoding apparatus shown in FIGS. 17, 19, 21, and 11, except that the following steps are added: analysis of the resampled The type of signal that is part of the signal multiplexing.

Figure 29 is a diagram showing a fourteenth embodiment of the encoding apparatus of the present invention. In this embodiment, on the basis of the encoding apparatus shown in FIG. 23, the gain control module 511 is added, the audio signal outputted by the signal property analysis module 510 is received, the dynamic range of the fast-changing type signal is controlled, and the pre-echo in the audio processing is eliminated. Its output is connected to a time-frequency mapping module 52 and a psychoacoustic analysis module 51.

The gain control module 511 only controls the fast-change type signal, but does not perform the slow-change type signal. Direct, direct output. For the fast variable type signal, the gain control module 511 adjusts the time domain energy envelope of the signal, and increases the gain value of the signal before the fast change point, so that the time domain signal amplitudes before and after the fast change point are relatively close; then the time domain is adjusted. The time envelope signal of the energy envelope is output to the time-frequency mapping module 52, and the gain adjustment amount is output to the bit stream multiplexing module 55.

The encoding method based on the encoding apparatus shown in Fig. 29 is basically the same as the encoding method based on the encoding apparatus shown in Fig. 23, except that the following steps are added: Gain control is performed on the signal subjected to signal type analysis.

FIG. 30 is a schematic structural diagram of Embodiment 8 of the decoding apparatus. The embodiment is an inverse gain control module 610 added to the output of the frequency-time mapping module 604 and the band expansion module 605. Between the inputs, the signal type analysis result and the gain adjustment amount information output by the bit stream demultiplexing module 601 are received, and the gain of the time domain signal is adjusted to control the pre-echo. After receiving the reconstructed time domain signal output by the frequency-time mapping module 604, the inverse gain control module 610 controls the fast-change type signal, and does not process the slowly-varying type signal, and directly outputs it to the band extension module 605. For the fast variable type signal, the inverse gain control module 610 adjusts the energy envelope of the reconstructed time domain signal according to the gain adjustment amount information, reduces the amplitude value of the signal before the fast change point, and adjusts the energy envelope back to the original front low and high height. State, such that the magnitude of the quantization noise before the fast change point is correspondingly reduced with the amplitude value of the signal, thereby controlling the pre-echo.

The decoding method based on the decoding apparatus shown in FIG. 30 is basically the same as the decoding method based on the decoding apparatus shown in FIG. 16, except that the following steps are added: Before the frequency band expansion of the reconstructed time domain signal is performed, the reconstructed time domain signal is performed. Inverse gain control.

31 to 35 are views showing the fifteenth to nineteenth embodiments of the encoding apparatus of the present invention. The five embodiments are respectively added to the encoding device shown in FIG. 24 to FIG. 28, and a gain control module 511 is added for controlling the dynamic range of the audio signal analyzed by the signal type, and eliminating the pre-echo in the audio processing. Its output is connected to a time-frequency mapping module 52 and a psychoacoustic analysis module 51.

The encoding method based on the above five encoding means is basically the same as the encoding method based on the encoding apparatus shown in Figs. 24 to 28, except that the following steps are added: Gain control is performed on the signal subjected to signal type analysis.

36 to 40 are diagrams showing the configuration of Embodiments 9 to 13 of the decoding apparatus, and the above 5 decoding apparatuses are respectively based on the decoding apparatus shown in Figs. 18, 20, 22, 10 and 12. The inverse gain control module 610 is added between the output of the frequency-time mapping module 604 and the input of the band expansion module 605, and receives the signal type analysis result output by the bit stream demultiplexing module 601 for adjusting the time domain signal. Gain, control pre-echo.

The decoding methods based on the above five decoding devices are also substantially the same as the decoding methods based on the decoding devices shown in Figs. 18, 20, 22, 10, and 12, respectively, except that the following steps are added: The inverse time gain is controlled on the reconstructed time domain signal before the time domain signal is reconstructed.

It should be noted that the above embodiments are only intended to illustrate the technical solutions of the present invention and are not intended to be limiting, and the present invention will be described in detail with reference to the preferred embodiments. The modifications and equivalents of the present invention are intended to be included within the scope of the appended claims.

Claims

Rights request

An enhanced audio encoding apparatus, comprising a psychoacoustic analysis module, a time-frequency mapping module, a quantization and chirp encoding module, and a bitstream multiplexing module, further comprising a band extension module and a resampling module; wherein the frequency band An expansion module, configured to analyze the original input audio signal over the entire frequency band, extract a spectral envelope of the high frequency portion, and correlate parameters related to the correlation between the low and high frequency spectra, and output the same to the bit stream multiplexing module;

The resampling module is configured to resample the input audio signal, change the sampling rate of the audio signal, and output the audio signal after changing the sampling rate to the psychoacoustic analysis module and the time-frequency mapping module;

The psychoacoustic analysis module is configured to calculate a masking threshold and a mask ratio of the input audio signal, and output to the quantization and entropy encoding module;

The time-frequency mapping module is configured to transform a time domain audio signal into a frequency domain coefficient;

The quantization and entropy coding module is configured to quantize and entropy encode frequency domain coefficients under control of a mask ratio output by the psychoacoustic analysis module, and output the same to the bitstream multiplexing module;

The bitstream multiplexing module is configured to multiplex the received data to form an audio coded code stream.

2. The enhanced audio encoding apparatus according to claim 1, wherein the resampling module comprises a low pass filter and a down sampler; wherein the low pass filter is for limiting a frequency band of the audio signal, the lower The sampler is used to downsample the audio signal of the limited frequency band and reduce the sampling rate of the signal.

The enhanced audio encoding apparatus according to claim 1, wherein the frequency band expansion module comprises a parameter extraction module and a spectral envelope extraction module; and the parameter extraction module is configured to extract input signals in different time-frequency regions. a parameter for inputting a spectral characteristic of the signal; the spectral envelope extraction module is configured to estimate a spectral envelope of a high frequency portion of the signal with a certain time-frequency resolution, and then output a spectral characteristic parameter of the input signal and a spectral envelope of the high frequency portion To the bitstream multiplexing module.

4. The enhanced audio encoding apparatus of claim 1, further comprising a frequency domain linear prediction and vector quantization module between the output of the time-frequency mapping module and the input of the quantization and entropy encoding module Specifically composed of a linear predictive analyzer, a linear predictive filter, a converter, and a vector quantizer;

The linear predictive analyzer is configured to perform predictive analysis on frequency domain coefficients, obtain prediction gains and prediction coefficients, and output frequency domain coefficients satisfying certain conditions to the linear prediction filter; for frequency domain coefficients that do not satisfy the condition Directly output to the quantization and entropy coding module;

The linear prediction filter is configured to perform linear prediction error filtering on the frequency domain coefficients to obtain a residual sequence, and output the residual sequence to the quantization and entropy coding module, and output the prediction coefficients to the converter;

The converter is configured to convert a prediction coefficient into a line spectrum pair frequency coefficient;

The vector quantizer is configured to perform multi-level vector quantization on the line spectrum on the frequency coefficient, and the related side information is transmitted to the bit stream multiplexing module.

5. The enhanced audio encoding apparatus of claim 1, further comprising a multi-resolution analysis module located between an output of the time-frequency mapping module and an input of the quantization and entropy encoding module, The signal type analysis result output by the psychoacoustic analysis module is used for multi-resolution analysis of the fast variable type signal; specifically, the frequency domain coefficient transform module and the recombination module, wherein the frequency domain coefficient transform module is a frequency domain wavelet transform filter a group or a frequency domain MDCT transform filter bank for transforming frequency domain coefficients into time-frequency plane coefficients; and the reassembly module is configured to recombine time-frequency plane coefficients according to a certain rule.

6. The enhanced audio encoding apparatus according to claim 5, further comprising a frequency domain linear pre- And a vector quantization module, located between an output of the time-frequency mapping module and an input of the multi-resolution analysis module, for performing linear prediction and multi-level vector quantization on frequency domain coefficients, and outputting a residual sequence to the The multi-resolution analysis module simultaneously outputs side information to the bit stream multiplexing module.

7. The enhanced audio encoding apparatus of claim 6 further comprising a sum and difference stereo encoding module between or located between an output of said multi-resolution analysis module and an input of said quantization and entropy encoding module Between the quantizer group and the encoder in the quantization and entropy coding module, the signal type analysis result output by the psychoacoustic analysis module is received.

8. An enhanced audio coding method, comprising the steps of:

Step 1: analyzing the input audio signal over the entire frequency band, extracting its high spectral envelope and signal characteristic parameters; Step 2, resampling the input audio signal;

Step 3: calculating a signal mask ratio of the signal after resampling;

Step 4: performing time-frequency mapping on the resampled signal to obtain a frequency domain coefficient of the audio signal; Step 5: performing quantization and entropy coding on the frequency domain coefficient;

Step 6: multiplexing the high spectral envelope, the signal spectral characteristic parameter and the encoded audio signal to obtain a compressed audio code stream.

The enhanced audio coding method according to claim 8, wherein the quantization of the step 是 is scalar quantization, specifically comprising: performing nonlinear compression on frequency domain coefficients in all scale factor bands; The scale factor of the band quantizes the frequency domain coefficients of the subband to obtain a quantized spectrum represented by an integer; selects the first scale factor in each frame signal as a common scale factor; and other scale factors are differentially processed with the previous scale factor;

The entropy coding specifically includes: entropy coding the quantized words and the scaled factors after the difference processing, obtaining a codebook serial number, a scale factor coded value, and a lossless coded value of the quantized spectrum; entropy coding the code book serial number to obtain a code book serial number The encoded value.

10. The enhanced audio coding method according to claim 8, wherein between step 4 and step 5, the method further comprises: performing standard linear prediction analysis on the frequency domain coefficients to obtain prediction gain and prediction coefficients; Whether the predicted gain exceeds a set threshold, if exceeded, frequency domain linear prediction error filtering is performed on the frequency domain coefficient according to the prediction coefficient, and a prediction residual sequence of the frequency domain coefficient is obtained; the prediction coefficient is converted into a line spectrum pair frequency coefficient, and Multi-level vector quantization processing is performed on the frequency spectrum to obtain the side information; the residual sequence is quantized and entropy encoded; if the prediction gain does not exceed the set threshold, the frequency domain coefficients are quantized and entropy encoded.

The enhanced audio encoding method according to claim 8, wherein the step 4 and the fifth step further comprise: if it is a fast variable type signal, performing multi-resolution analysis on the frequency domain coefficients; Instead of fast-changing type signals, the frequency domain coefficients are directly quantized and encoded;

The multi-resolution analysis step includes: performing MDCT transform on the frequency domain coefficients to obtain time-frequency plane coefficients; and reconstructing the time-frequency plane coefficients according to a certain rule;

The recombination includes: firstly, the time-frequency plane coefficients are organized in a frequency direction, the coefficients in each frequency band are organized in a time direction, and then the organized coefficients are arranged in the order of the sub-window and the scale factor band.

The enhanced audio coding method according to claim 11, wherein before performing the multi-resolution analysis step, the method further comprises: performing standard linear prediction analysis on the frequency domain coefficients to obtain prediction gain and prediction coefficients; Whether the predicted gain exceeds the set value, if it exceeds, frequency domain linear prediction error filtering is performed on the frequency domain coefficient according to the prediction coefficient, and the prediction residual sequence of the frequency domain coefficient is obtained; and the prediction coefficient is converted into a line spectrum pair frequency coefficient, And multi-level vector quantization processing on the frequency spectrum of the line spectrum to obtain the side information; Multi-resolution analysis of the line; multi-resolution analysis of the frequency domain coefficients if the predicted gain does not exceed the set threshold.

The enhanced audio encoding method according to claim 12, wherein the step 5 further comprises: quantifying the residual sequence; determining whether the audio signal is a multi-channel signal, and if it is a multi-channel signal, determining Whether the signal types of the left and right channel signals are consistent. If the signal types are the same, it is judged whether the scale factor bands corresponding to the two channels satisfy the difference and the stereo coding conditions. If yes, the spectral coefficients in the scale factor band are satisfied. Performing and difference stereo coding to obtain a residual sequence of the difference channel; if not, the spectral coefficients in the scale factor band are not subjected to and differential stereo coding; if it is a mono signal or a signal type inconsistent multi-channel Signal, then the residual sequence is not processed; the residual sequence is entropy encoded;

The method for judging whether the scale factor band satisfies the coding condition is: K-L transformation, specifically: calculating a correlation matrix of spectral coefficients of the left and right channel scale factor bands; performing KL transformation on the correlation matrix; if the absolute value of the rotation angle α Deviation; ζ74 is small (eg 3 / 16 <

And the scale factor band corresponding to the two-channel signal can be subjected to and the difference stereo coding; wherein the sum and the difference stereo coding are: wherein: the quantized sum channel frequency domain is represented

Coefficient; represents the quantized difference channel frequency domain coefficient; £ represents the quantized left channel frequency domain coefficient; represents the quantized right channel frequency domain coefficient.

An enhanced audio decoding device, comprising: a bit stream demultiplexing module, an entropy decoding module, an inverse quantizer group, and a frequency-time mapping module, further comprising a band extension module;

The bitstream demultiplexing module is configured to demultiplex the compressed audio data stream, and output corresponding data signals and control signals to the entropy decoding module and the band extension module;

The entropy decoding module is configured to perform decoding processing on the foregoing signal, recover a quantized value of the spectrum, and output the result to the inverse quantizer group;

The inverse quantizer group is configured to reconstruct an inverse quantization spectrum and output to the frequency-time mapping module; the frequency-time mapping module is configured to perform frequency-time mapping on the spectral coefficients to obtain a low-band time domain audio. Signal

The band extension module is configured to receive the band extension control information output by the bit stream demultiplexing module and the frequency-time mapping module, and the time domain audio signal of the low frequency band, reconstruct the high frequency signal part, and output the broadband audio signal.

15. The enhanced audio decoding device of claim 14, further comprising a sum and difference stereo decoding module, located between an output of the inverse quantizer group and an input of the frequency-time mapping module or located at Between the output of the entropy decoding module and the input of the inverse quantizer group, the sum and difference stereo control signals output by the bit stream demultiplexing module are received, and are used to inverse the sum channel and the difference channel according to the difference stereo control information. The quantized value of the quantized spectrum/spectrum is converted into a quantized value of the inverse quantized spectrum/spectrum of the left and right channels.

16. The enhanced audio decoding device of claim 14, further comprising an inverse frequency domain linear prediction and vector quantization module, located at an output of the inverse quantizer group and an input of the frequency-time mapping module And receiving the inverse frequency domain linear predictive vector quantization control information output by the bit stream demultiplexing module, performing a linear prediction synthesis process on the inverse quantized spectrum, obtaining a pre-predicted spectrum, and outputting to the frequency-time mapping module Specifically comprising an inverse vector quantizer, an inverse converter and an inverse linear prediction filter; the inverse vector quantizer is configured to inverse quantize the codeword index to obtain a line spectrum pair frequency coefficient; and the inverse converter is used for the line Language pair frequency system The inverse of the number is converted into a prediction coefficient; the inverse linear prediction filter is configured to perform a linear prediction synthesis process on the inverse quantization spectrum according to the prediction coefficient to obtain a spectrum before prediction.

17. The enhanced audio decoding device of claim 16, further comprising a sum and difference stereo decoding module located at an output of the inverse quantizer group and an input of the inverse frequency domain linear prediction and vector quantization module Or between the output of the entropy decoding module and the input of the inverse quantizer group, receiving the sum and difference stereo control signals output by the bit 3⁄4⁄4 demultiplexing module, for sum and difference according to the sum and difference stereo control information The quantized value of the channel humble quantization 镨/spectrum is converted into the quantized value of the inverse quantized borrow/spectrum of the left and right channels.

18. The enhanced audio decoding device of claim 14, further comprising a multi-resolution synthesis module located between an output of the inverse quantizer group and an input of the frequency-time mapping module, The signal type analysis result output by the bit stream demultiplexing module is used for multi-resolution synthesis of the inverse quantization spectrum; the multi-resolution synthesis module specifically includes: a coefficient recombination module and a coefficient transformation module; Frequency domain inverse wavelet transform filter bank or frequency domain inverse modified discrete cosine transform filter bank.

19. The enhanced audio decoding device of claim 18, further comprising an inverse frequency domain prediction and vector quantization module, the output of the inverse quantizer group and the multi-resolution synthesis module Between input or between an output of the multi-resolution synthesis module and an input of the frequency-time mapping module, receiving inverse frequency-domain linear prediction vector quantization control information output by the bit stream demultiplexing module, The inverse quantization process and the linear prediction synthesis process are performed on the inverse quantization spectrum after the inverse quantization error/multiresolution synthesis, and the prediction before the prediction is obtained.

20. The enhanced audio decoding device of claim 19, further comprising a sum and difference stereo decoding module, located at an output of the inverse quantizer group and an input of the inverse frequency domain linear prediction and vector quantization module or Between the multi-resolution synthesis modules, or between the output of the entropy decoding module and the input of the inverse quantizer group, receiving the sum and difference stereo control signals output by the bit stream demultiplexing module, The quantized values of the inverse quantized spectrum/spectrum of the difference channel and the difference channel are converted into inverse quantized spectral/spectral 量化 quantized values of the left and right channels.

21. An enhanced audio decoding method, comprising the steps of:

Step 1: Demultiplexing the compressed audio data stream to obtain data information and control information;

Step 2: Entropy decoding the above information to obtain a quantized value of the spectrum;

Step 3: performing inverse quantization processing on the quantized value of the spectrum to obtain inverse quantization borrowing;

Step 4: performing frequency-time mapping on the inverse quantization spectrum to obtain a time domain audio signal;

Step 5: reconstruct a high frequency portion of the time domain audio signal according to the band extension control information to obtain a broadband audio signal.

The enhanced audio decoding method according to claim 21, wherein the step 4 further comprises: performing an inverse modified discrete cosine transform on the inverse quantized spectrum to obtain a transformed time domain signal; The domain signal is windowed in the time domain; superimposing the windowed time domain signal to obtain a time domain audio signal; wherein the window function in the windowing process is:

w{ -k) =cos {pill* ( +0. 5) / -0. 94* s in (2 (1+0. 5) ) / (2* i) ) ) , where pi is the pi, k = 0. . . 71 ^-1 ; represents the kth coefficient of the window function, with w (k) = w (2*7^1- D ; N represents the number of samples of the encoded frame.

The enhanced audio decoding method according to claim 21, further comprising: between the step 2 and the step 3, further comprising: if the signal type analysis result indicates that the signal type is consistent, determining according to the difference and the stereo control signal Whether it is necessary to perform inverse stereo quantization on the inverse quantization spectrum; if necessary, according to each scale The flag bit on the factor band determines whether the scale factor band requires and the difference stereo decoding, and if necessary, converts the inverse quantized spectrum of the difference channel in the scale factor band into the inverse quantized spectrum of the left and right channels, and proceeds to the step If the signal type is inconsistent or does not need to perform and the difference stereo decoding, the inverse quantization spectrum is not processed, and the process proceeds to step 3. In the above step, the sum difference stereo decoding is: , where: represents inverse quantization

The following sum channel frequency domain coefficients; s represents the inverse quantized difference channel frequency domain coefficients; z represents the inverse quantized left channel frequency domain coefficients; f represents the inverse quantized right channel frequency domain coefficients.

The enhanced audio decoding method according to claim 21, further comprising: determining whether the inverse quantization spectrum needs to pass the inverse frequency domain linear prediction vector in the control information between step 3 and step 4 The quantized information, if included, is subjected to inverse vector quantization processing to obtain a prediction coefficient, and linearly predictively synthesizes the inverse quantized spectrum to obtain a spectrum before prediction; and performs frequency-time mapping on the spectrum before prediction; wherein the inverse vector The quantization process further includes: obtaining a codeword index of the prediction coefficient vector quantization from the control information; obtaining the quantized line spectrum pair frequency coefficient according to the codeword index, and calculating the prediction coefficient from the line coefficient to the frequency coefficient.

The enhanced audio decoding method according to claim 24, further comprising: between the step 2 and the step 3, further comprising: if the signal type analysis result indicates that the signal type is consistent, determining according to the difference and the stereo control signal Whether it is necessary to perform differential and stereo decoding on the quantized values of the spectrum; if necessary, determine whether the scale factor band requires and difference stereo decoding according to the flag bits on each scale factor band, if necessary, in the scale factor band And the quantized value of the spectrum of the difference channel is converted into the quantized value of the spectrum of the left and right channels, and the process proceeds to step 3; if the signal type is inconsistent or does not need to perform the difference and the stereo decoding, the quantized value of the spectrum is not processed, and the process proceeds to Step three.

The enhanced audio decoding method according to claim 21, further comprising: step of performing multi-resolution synthesis on the inverse quantization spectrum between the step 3 and the step 4; The inverse quantized spectral coefficients are arranged in the order of sub-window and scale factor bands, and then recombined according to the frequency sequence, and then multiple inverse modified cosine transforms are performed on the recombined coefficients to obtain an inverse quantized spectrum before multi-resolution analysis.

The enhanced audio decoding method according to claim 26, further comprising: determining whether the inverse quantization spectrum needs to undergo inverse frequency domain linear prediction vector quantization before performing the multi-resolution synthesis step The information, if included, is subjected to inverse vector quantization processing to obtain prediction coefficients, and the residual spectrum is linearly predicted and synthesized to obtain a spectrum before prediction; and the spectrum before prediction is subjected to multi-resolution synthesis.

The enhanced audio decoding method according to claim 26, further comprising: determining whether the inverse quantization spectrum needs to be subjected to inverse frequency domain linear prediction vector quantization before the step 4 is performed. If it is, the inverse vector quantization process is performed to obtain a prediction coefficient, and the residual spectrum is subjected to line>!·sheng prediction synthesis to obtain a spectrum before prediction; and the spectrum before prediction is subjected to frequency-time mapping.

The enhanced audio decoding method according to any one of claims 26 to 28, characterized in that, between the step 2 and the step 3, the method further comprises: if the signal type analysis result indicates that the signal types are consistent, according to And the difference stereo control signal determines whether it is necessary to perform the difference and the stereo decoding on the borrowed quantized value; if necessary, judge whether the scale factor band needs the difference and the stereo decoding according to the flag bit on each scale factor band, if necessary, The quantized value of the spectrum of the difference channel in the scale factor band is converted into the quantized value of the spectrum of the left and right channels, and the process proceeds to step 3; if the signal type is inconsistent or does not need to perform and the difference stereo decoding, the quantized value of the spectrum Do not process, go to step three.