CN101754086B - Decoder and decoding method for multichannel audio coder using sound source location cue - Google Patents

Abstract

The invention provides a multi-channel audio decoding device and method based on sound source location clues. The decoding device includes: a demultiplexer that receives a signal, and parses the received signal into an audio bit stream and an additional information bit stream; an audio decoder that restores the downmixed signal based on the audio bit stream, and passing the restored downmix signal to a synthesis upmixer without applying an analysis filter bank to said restored downmix signal; the synthesis upmixer, by using said downmix signal and said additional information bitstream, predicting a multi-channel signal, and upmixing the downmixed signal based on the multichannel signal to generate an upmixed signal; and a window opener of an integrated filter bank, performing an integrated filter bank on the upmixed signal to extract time domain signal, and perform windowing on the upmixed signal to extract the output signal.

Description

A decoding device and method for multi-channel audio based on sound source location clues

technical field

The invention discloses a multi-channel audio decoding device and method based on sound source location clues.

Background technique

The basic coding concept of Sound Source Location Cue Coding (SSLCC: Sound Source Location Cue Coding) starts from the method of Spatial Audio Coding (SAC: Spatial Audio Coding).

The SAC is a multi-channel audio compression technology based on sound source location cues, and deleting the redundancy of each channel signal based on the spatial cues (Spatial Cue) recognized by people in the space can maximize the compression efficiency.

Moreover, the multi-channel signal is basically processed through down-mixing, so that the transmitted audio download signal becomes a core signal. That is, reproduction can be achieved through existing stereo audio, which is the basic principle of the SAC method.

The SSLCC is one of the SAC methods and is a spatial cue recognized by people in space. It can predict the position of the sound source, and extract the position information of the sound source from the multi-channel signal to represent and transmit it.

At this time, because the amount of information extracted by the SAC encoding strategy is too small, it can be transmitted to the redundant field, so if the receiver receiving the information is not audio that supports SAC encoding, the existing stereo audio is used to reproduce only For the stereo signal, if the receiver receiving the information supports SAC coding audio, the transmitted redundant information is used to restore the multi-channel audio signal based on the location clue of the sound source from the downmixed stereo signal.

However, in order to restore a multi-channel audio signal based on sound source position cues from the downmixed stereo signal by using redundant information, it is necessary to use the T/F (time to frequency ) transformation method with the same T/F transformation method, so if the T/F transformation method used in the process of extracting information through the SAC coding strategy is not the most suitable T/F transformation method at the receiving end, it will give the transformation process bring bad influence.

Therefore, there is a need for an apparatus and method for recovering multi-channel audio signals based on sound source location clues that are most suitable for the T/F conversion method at the receiving end.

Contents of the invention

The present invention provides a method of receiving and compressing multi-channel audio signals, and compressing and transmitting the stereo signals via basic stereo coding, which can transmit multi-channel audio while maintaining backward compatibility with basic stereo audio coding. Signal decoding device and method thereof.

The present invention provides a decoding device and method thereof that can convert the T/F of an audio signal for stereo downmixing according to selection by using a Time Domain Aliasing Cancellation (TDAC: Time Domain Aliasing Cancellation) file bank.

Technical solutions

According to an exemplary embodiment of the present invention, a multi-channel audio decoding device based on sound source location clues is provided, which includes: a demultiplexer, receiving signals, and analyzing the received signals into audio bit streams and additional an information bit stream; an audio decoder, based on the audio bit stream, restores the downmixed signal; an integrated up-mixer, predicting a multi-channel signal by using the downmixed signal and the additional information bitstream, based on the The multi-channel signal upmixes the downmixed signal to generate an upmixed signal; and a windower of an integrated filter bank performs an integrated filter bank on the upmixed signal to extract a time domain signal, and the The upmix signal is windowed to extract the output signal.

And, according to another exemplary embodiment of the present invention, a method for decoding multi-channel audio based on sound source location clues is provided, which includes: analyzing the received signal into an audio bit stream and an additional information bit stream; based on the an audio bit stream, restoring the downmixed signal; using the downmixed signal and the additional information to predict a multi-channel signal; based on the multi-channel signal, upmixing the downmixed signal to generate an upmixed signal; performing a synthesis filter bank on the upmixed signal to decimate a time domain signal; and windowing the upmixed signal to decimate an output signal.

technical effect

According to the present invention, by receiving and compressing the multi-channel audio signal, and compressing and transmitting the stereo signal via the basic stereo coding, the multi-channel audio signal can be transmitted while remaining backward compatible with the basic stereo coding.

The present invention can transform the T/F of the stereo downmixed audio signal according to selection by using the TDAC filter bank.

Description of drawings

Fig. 1 shows the coding device of the multi-channel audio based on sound source position clue according to the exemplary embodiment of the present invention;

Fig. 2 shows the decoding device of the multi-channel audio based on sound source location clue according to the exemplary embodiment of the present invention;

FIG. 3 shows a multi-channel audio decoding device based on sound source position clues according to another exemplary embodiment of the present invention;

FIG. 4 shows an apparatus for decoding an additional information bitstream according to an exemplary embodiment of the present invention;

FIG. 5 shows a process in which an integrated up-mixer predicts the gain of each channel according to an exemplary embodiment of the present invention;

Figure 6 shows an inverse correlator according to an exemplary embodiment of the present invention;

Fig. 7 shows a method for decoding multi-channel audio based on audio source position clues according to an exemplary embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present invention will be specifically described below with reference to the accompanying drawings.

Fig. 1 shows an apparatus for encoding multi-channel audio based on audio source position clues according to an exemplary embodiment of the present invention.

According to the exemplary embodiment of the present invention, the multi-channel audio coding device 100 based on the sound source position clue is as a 5-channel multi-channel audio coding device based on SSLCC, as shown in FIG. 120, a downmix processor 130, an audio encoder 140 and a multiplexer 150.

At this time, the multi-channel audio encoding device 100 based on sound source location clues according to an exemplary embodiment of the present invention may be extended to more than 5 channels of multi-channel audio content based on sound source location clues.

The pre-processing filter bank unit 110 performs pre-processing on the multi-channel input audio signal input to the multi-channel audio encoding device 100 based on the sound source position clue, and converts the pre-processed input audio signal into a frequency domain through a filter bank. Signal. At this time, the filter bank performs T/F transformation based on sub-band analysis, which can be applied to MDCT, MDST, DFT, and the like.

Here, the input audio signal may include input signal LF (Left Front), input signal RF (Right Front), input signal C (Center Front), input signal Ls (Left Surround), input signal Rs (Right Surround) .

The analyzer 120 extracts spatial cues from the input audio signal transformed into the frequency domain in the pre-processing filter bank 110 , and expresses the spatial cues as a bit stream of additional information for transmission. At this time, the analyzer 120 transmits to the downmix processor 130 by compressing the input audio signal.

The downmix processor 130 may perform downmixing as the analyzer 120 downmixes the input audio signal in the frequency domain. In addition, the downmix processor 130 can perform downmixing according to ITU-T recommendations.

The audio signal downmixed in the downmixing processor 130 may be expressed as a bit stream through common stereo audio. The commonly used stereo audio can utilize MP3 (MPEG Layer III) or AAC (Advanced Audio Coding) and the like.

The audio encoder 140 may encode the audio signal downmixed from the downmix processor 130 .

The multiplexer 150 combines and transmits the encoded signal from the audio encoder 140 and the additional information bit stream transmitted from the analyzer 120 .

Fig. 2 shows an apparatus for decoding multi-channel audio based on sound source position clues according to an exemplary embodiment of the present invention.

The decoding device 200 of the multi-channel audio based on the sound source position clue according to the exemplary embodiment of the present invention is as the 5-channel multi-channel audio decoding device based on SSLCC, as shown in Figure 2, by demultiplexer 210, audio decoder 220 , a windowing filter bank 230 , an integrated upmixer 240 and an integrated filter bank windower 250 .

The demultiplexer 210 receives the signal transmitted from the demultiplexer 150, and parses the received signal into an audio bit stream and an additional information bit stream.

The audio decoder 220 restores the downmixed signal based on the audio bitstream.

The windowed filter bank 230 applies an analysis filter bank to the downmixed signal to perform T/F conversion, windowing and transmitting the T/F converted downmixed signal.

The integrated upmixer 240 generates an upmixed signal by upmixing the downmixed signal based on the multichannel signal by predicting a multichannel signal using the downmixed signal and the additional information bitstream.

Specifically, the integrated up-mixer 240 separates the amplitude information and the phase information from the down-mixed signal, and assigns a weighted value to the phase information based on the amplitude signal by windowing the stored random time series, based on the Amplitude information and phase information given by weighting values can predict multi-channel signals.

At this time, the integrated upmixer 240 complex transforms the downmixed signal, and separates amplitude information and phase information from the complex transformed downmixed signal.

Moreover, the integrated upmixer 240 is used to modify the spectral window of the phase information based on the envelope modeling of the amplitude information, and applies the spectral window to the stored random time sequence to open the window, by using the windowed The random timing of will assign weighted values to the phase information.

The integrated upmixer 240 combines the amplitude information and the phase information assigned by the weight value, and inversely transforms the combined amplitude information and the phase information assigned by the weight value to predict the multi-channel signal.

The window opener 250 of the integrated filter bank performs an integrated filter bank on the upmixed signal to extract a signal in the time domain, and performs windowing on the upmixed signal to extract an output signal.

Fig. 3 shows an apparatus for decoding multi-channel audio based on audio source position clues according to another exemplary embodiment of the present invention.

According to another embodiment of the present invention, the decoding device 300 of multi-channel audio based on sound source position clues is used as a decoding device 300 based on the multi-channel audio based on sound source position clues of real transform, as shown in FIG. 2 , which consists of A demultiplexer 310, a TDAC filter bank 320, an integrated upmixer 330 and a windower 340 for the integrated filter bank are formed.

SSLCC basically follows a DFT filter bank (transform). However, for interfacing with Kernel Stereo Audio, various filter banks are available.

Although the form of the filter bank is changed, the operation principle of the synthesis between the SSLCC analyzer 120 or the synthesis upmixer 240 and the windower 250 of the synthesis filter bank is the same.

Since no decorrelator is used for stereoscopic transmission, the actual transformation can be realized. According to another embodiment of the present invention, the multi-channel audio decoding device 300 based on sound source location clues uses the MDCT that can be actually transformed to interlock with the kernel codec.

The TDAC filter bank 320 restores the downmixed signal based on the audio bitstream, and then omits the process of applying the analytical filter bank and the windowing process to the downmixed signal, and transmits it to the integrated upmixer 330 . At this time, the signals transmitted by the TDAC filter bank 320 may be the frequency downmixed signal L and the frequency downmixed signal R.

The synthesis filter bank windower 340 applies a synthesis filter bank to the upmixed signal generated in the synthesis upmixer 330 to decimate the time domain signal for combining said upmixed signal with the kernel stereo audio The analysis windows of the two are paired with each other to extract the output signal.

At this time, the demultiplexer 310 and the integrated up-mixer 330 have the same functions as the demultiplexer 210 and the integrated up-mixer 240 of the multi-channel audio decoding device 200 based on sound source location clues according to an embodiment of the present invention. structure, so a detailed description is omitted.

According to the multi-channel audio decoding apparatus 300 based on sound source position clues according to an exemplary embodiment of the present invention, the T/F transformation of the stereo downmixed audio signal can be changed according to selection. For example, when the operation of the inverse correlator in the decoding device is off, the actual T/F conversion can also be applied. At this time, complex T/F conversion can also be realized, and even in the case of complex T/F conversion, phase information does not need to be used.

However, when the operation requiring phase information is turned on in the decoding device, that is, when an inverse correlator operation is required, complex T/F conversion must be used. In complex T/F transformation, DFT becomes the basis, and MDCT/MDST can also be applied as a complex transform pair.

FIG. 4 shows an apparatus for decoding an additional information bitstream according to an exemplary embodiment of the present invention.

According to the decoding device of the additional information bit stream of the exemplary embodiment of the present invention, as the additional information in the bit stream of the additional information analyzed by the demultiplexer 210, that is, VLSA (Virtual SoundLocation Angle), as shown in FIG. Huffman decoder 410 and inverse quantizer 420 constitute.

And, the decoding apparatus of the additional information bitstream according to the exemplary embodiment of the present invention may belong to the windowed filter bank 230 and the windowed filter bank 250 to synthesize the filter bank.

The Huffman decoder 410 uses the Huffman coding book to perform Huffman coding on the bit stream of the additional information to generate a differential index.

The Huffman decoder 410 includes an inverse differential encoder 411 , a differential encoder 412 , a mapper 413 and a Huffman encoder 414 to generate the Huffman codebook.

The inverse differential encoder 411 can decode the original index by performing inverse differential encoding based on the processed previous frame and Huffman coded information.

In addition, the differential encoder 412 deletes negative information from the original exponent according to the sine bit information, and then performs differential encoding to generate exponent information.

The mapper 413 is used to remove the negative information in the exponent, delete the offset (offset) information, and then map the exponent according to the frequency solution (solution), to be divided into the first sub-band (sub band) and divide the second Other frequency bands other than a sub-band.

Finally, the Huffman encoder 414 applies a Huffman encoding method to each of the first sub-bands and other frequency bands except the first sub-band to generate a Huffman codebook.

The Huffman decoder 410 decodes the first sub-band by referring to the Huffman coding book of Table 1.

[Table 1]

Index Num of bits Code word Index Num.of bits Codeword

0 5 0x17 16 5 0x1d 1 8 0x64 17 5 0x19 2 8 0x65 18 5 0x1c 3 8 0xf0 19 5 0x16 4 8 0xf1 20 5 0x18 5 7 0x33 twenty one 5 0x14 6 7 0x79 twenty two 5 0x13 7 6 0x18 twenty three 5 0x15 8 6 0x22 twenty four 5 0x1b 9 6 0x23 25 5 0x10 10 6 0x3d 26 5 0x0e 11 5 0x0b 27 5 0x0f 12 5 0x12 28 5 0x0d 13 5 0x1a 29 5 0x0a 14 4 0x04 30 2 0x00 15 5 0x1f

Moreover, when the signal received by the demultiplexer 210 is a 5-bit quantized signal, the Huffman decoder 410 performs Huffman decoding by referring to the Huffman coding book in Table 2.

[Table 2]

Moreover, when the signal received by the demultiplexer 210 is a 4-bit quantized signal, the Huffman decoder 410 performs Huffman decoding by referring to the Huffman coding book in Table 3.

[table 3]

The inverse quantizer 420 uses an inverse quantization table to inverse quantize the difference index to recover additional information. Specifically, the inverse quantizer 420 can perform inverse quantization by mapping VLSA (VirtualSound Location Angle) information in each frame with a quantization table corresponding to each VSLA. At this time, since the multi-channel audio based on the sound source location clues according to the exemplary embodiment of the present invention is basically decoded with frame-by-frame DFT or MDCT, the smoothing between frames is mainly performed by windowed overlapping Additional (overlap-add) mode is satisfied.

When the VLSA information is LHA (Left Half-plane Angle), the inverse quantizer 420 can perform quantization through the quantization table of the mapping table 4.

[Table 4]

At this time, in the step of restoring the additional information, when the VLSA information is RHA (Right Half-plane Angle), the inverse quantizer 420 can perform quantization through the quantization table of the mapping table 5.

[table 5]

At this time, in the step of restoring the additional information, when the VLSA information is LSA (Left Subsequent Vector Angle), the inverse quantizer 420 can perform quantization through the quantization table of the mapping table 6.

[Table 6]

Idx -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 LSA[idx] -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 Idx -4 -3 -2 -1 0 1 2 3 4 5 6 LSA[idx] -4 -3 -2 -1 0 1 2 3 4 5 6

Idx 7 8 9 10 11 12 13 14 15 LSA[idx] 7 8 9 10 11 12 13 14 15

At this time, in the step of restoring the additional information, when the VLSA information is RSA (Right Subsequent Vector Angle), the inverse quantizer 420 can perform quantization through the quantization table of the mapping table 7.

[Table 7]

Idx -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 RSA[idx] -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 Idx -4 -3 -2 -1 0 1 2 3 4 5 6 RSA[idx] -4 -3 -2 -1 0 1 2 3 4 5 6 Idx 7 8 9 10 11 12 13 14 15 RSA[idx] 7 8 9 10 11 12 13 14 15

And, the inverse quantizer 420 may extract a variable satisfying Mathematical Expression 1 from the VLSA information obtained from each index information.

[mathematical formula 1]

${\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; lh</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; LHA</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; idx</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; LSA</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; idx</span>}<span\; class="notranslate">\; ]</span>}{\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; LSA</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; idx</span>}<span\; class="notranslate">\; ]</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

${\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; rh</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; RHA</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; idx</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; 5</span>}}{\mathrm{<span\; class="notranslate">\; RSA</span>}<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; idx</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; P</span>}\mathrm{<span\; class="notranslate">\; 5</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\frac{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

gLs=sin(θ _lh )

gL＝cos(θ _lh )·sin((LSA[idx]-2π)×-3)

gCL＝cos(θ _lh )·cos((LSA[idx]-2π)×-3)

gRs＝cos(θ _rh )

gCR＝sin(θ _rh )·cos(RSA[idx]×3)

gR＝sin(θ _rh )·sin(RSA[idx]×3)

At this time, the number of sub-bands can be obtained from the definition variable of the number of sub-bands included in the additional information, namely bsFreqRes, and the inverse quantizer 420 can map the number of the transmitted VSLA information according to the number of sub-bands. In addition, the maximum number of frequency bands is based on 28 frequency bands (Mpar=28), and the resolution and the number of frequency bands are different depending on the bit rate (bit rate) or the frequency characteristics of the frame.

The inverse quantizer 420 can be mapped as shown in Table 8 through mapsubbands(bsFreqRes, Mpar).

[Table 8]

The inverse quantizer 420 can process Mathematical Formula 1 by using the resolution of each frequency band designed based on the ERB frequency band. In the case of Mpar=28 when mapping such as 8, the resolution of each frequency band can be as shown in Table 9.

[Table 9]

m M _par =28 kHz 0 0 0.0702 1 1 0.1639 2 2 0.2576 3 3 0.3512 4 4 0.4449 5 5 0.5385 6 6 0.6322 7 7 0.7259 8 8 0.9132 9 9 1.1005 10 10 1.2878 11 11 1.4751 12 12 1.8498 13 13 2.2244 14 14 2.599

15 15 2.9737 16 16 3.7229 17 17 4.4722 18 18 5.2215 19 19 5.9707 20 20 6.72 twenty one twenty one 7.4693 twenty two twenty two 8.5932 twenty three twenty three 9.7171 twenty four twenty four 11.2156 25 25 13.0888 26 26 15.3366 27 27 twenty four

At this time, the periodic prior symbol remover 210, the inverse binary Fourier transformer 220, and the guard band remover 230 have the same number as the number of receiving antennas used by the decoder to correspond to each receiving antenna .

After the T/F transform is performed by the windowed filter bank 230, the frequency band in the frequency domain can be defined as a processing band.

For example, when performing 2048-point DFT transformation, as shown in Table 10, with the positions of start bin and stop bin as the center, the frequency band in the frequency domain can be defined as a processing frequency band.

[Table 10]

The integrated upmixer 240 can restore the multi-channel signal through audio phase positioning within each sub-band in the downmixed signal based on the VSLA information restored from the bitstream of the side information. Specifically, as shown in Figure 5, the dynamic information in each sub-band is predicted by using the panning angle from the bit stream of additional information, and then the sub-channel signal of each channel can be predicted by applying the dynamic information.

FIG. 5 illustrates a process of predicting a gain of each channel by an integrated upmixer according to an exemplary embodiment of the present invention.

As shown in FIG. 5 , the integrated up-mixer 240 predicts the gain of each channel by recovering the audio phase information of each channel in stages.

First, the integrated upmixer 240 can restore LHA[idx] 510 and RHA[idx] 520 .

Synthetic upmixer 240 predicts gLs[idx] 530 from LHA[idx] 510 , restores LSA[idx] 511 , predicts gRs[idx] 540 from RHA[idx] 520 , and restores RSA[idx] 521 .

Then, the integrated upmixer 240 predicts gL[idx] 550 and gCL[idx] 512 from LSA[idx] 511 , and gRs[idx] 560 and gCR[idx] 522 from RSA[idx] 521 .

Finally, synthetic upmixer 240 predicts gCL[idx]/sqrt(2) 570 from gCL[idx] 512 and gCR[idx] 522 . At this time, gCL[idx]/sqrt(2) is gCL[idx] 512*0. 7071 may be the adjusted value of the gain of the center channel.

The integrated upmixer 240 may generate an upmixed signal by upmixing the downmixed signal based on the multi-channel signal predicted through the steps.

If X _dmxL (m, k) is the k-th frequency bin of the m-th sub-band of the transmitted Left downmixed signal, the 'Left upmixing Matrix' can satisfy Mathematical Formula 2.

[mathematical formula 2]

$\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; gCL</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; gL</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; wxya</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\end{array}\right]{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; CL</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; ls</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\end{array}\right]$

In addition, the 'Rightupmixing Matrix' of the Right downmixed signal can satisfy Mathematical Formula 3.

[mathematical formula 3]

$\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; gCL</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; GR</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; wxya</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>\end{array}\right]{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; CR</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; Rs.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>\end{array}\right]$

Also, the integrated upmixer 240 may include DFT-based decorrelators, ie, _DL and _DR .

The _DL and _DR can operate in a high complexity mode (high complexity mode) and a low complexity mode (low complexity mode) as a normal mode. At this time, the _DL and _DR are generated only in the decoder, and operate in a high-complexity mode when generating high-quality sound, and operate in a low-complexity mode when reproducing normal sound quality.

In high complexity mode, the _DL and _DR perform matrix transformation (matrixing) of Mathematical Formula 4 on L(m,k) and R(m,k) to generate an inverse correlation signal.

[mathematical formula 4]

In normal mode, the _DL and _DR satisfy Mathematical Formula 5, and no inverse correlation signal is generated.

[mathematical formula 5]

The integrated upmixer 240 upmixes the values calculated by the Mathematical Formulas 2 and 3 using Mathematical Formula 6.

[mathematical formula 6]

$\left[\begin{array}{cccccccc}0.7071& 0.7071& 0& 0& 0& 0& 0& 0\\ 0& 0& \mathrm{\&alpha;}\left(m\right)& 1-\mathrm{\&alpha;}\left(m\right)& 0& 0& 0& 0\\ 0& 0& 0& 0& \mathrm{\&alpha;}\left(m\right)& 1-\mathrm{\&alpha;}\left(m\right)& 0& 0\\ 0& 0& 0& 0& 0& \mathrm{\&delta;}& 0& 0\\ 0& 0& 0& 0& 0& 0& \mathrm{\&delta;}& 0\\ 0& 0& 1& 0& 1& 0& 0& 0\end{array}\right]\left[\begin{array}{c}\mathrm{CL}(m,k)\\ \mathrm{CR}(m,k)\\ L(m,k)\\ L_{\mathrm{wet}}(m,k)\\ R(m,k)\\ R_{\mathrm{wet}}(m,k)\\ \mathrm{ls}(m,k)\\ \mathrm{Rs.}(m,k)\end{array}\right]=\left[\begin{array}{c}C(m,k)\\ L(m,k)\\ R(m,k)\\ \mathrm{ls}(m,k)\\ \mathrm{Rs.}(m,k)\\ \mathrm{lfe}(m,k)\end{array}\right],$ m<4

$\left[\begin{array}{cccccccc}0.7071& 0.7071& 0& 0& 0& 0& 0& 0\\ 0& 0& 1-\mathrm{\&alpha;}\left(m\right)& \mathrm{\&alpha;}\left(m\right)& 0& 0& 0& 0\\ 0& 0& 0& 0& 1-\mathrm{\&alpha;}\left(m\right)& \mathrm{\&alpha;}\left(m\right)& 0& 0\\ 0& 0& 0& 0& 0& \mathrm{\&delta;}& 0& 0\\ 0& 0& 0& 0& 0& 0& \mathrm{\&delta;}& 0\\ 0& 0& 0& 0& 0& 0& 0& 0\end{array}\right]\left[\begin{array}{c}\mathrm{CL}(m,k)\\ \mathrm{CR}(m,k)\\ L(m,k)\\ L_{\mathrm{wet}}(m,k)\\ R(m,k)\\ R_{\mathrm{wet}}(m,k)\\ \mathrm{ls}(m,k)\\ \mathrm{Rs.}(m,k)\end{array}\right]=\left[\begin{array}{c}C(m,k)\\ L(m,k)\\ R(m,k)\\ \mathrm{ls}(m,k)\\ \mathrm{Rs.}(m,k)\\ \mathrm{lfe}(m,k)\end{array}\right],$ m≥4

At this time, α(m) may be a factor indicating a relationship between L and R signals of each frequency band. δ is a fixed coefficient, which may be a fixed coefficient for the demixing coefficient of the surrounding signal when the encoder is downloaded.

The α(m) is obtained by calculating the value calculated by the above-mentioned Mathematical Expressions 4 and 5 using Mathematical Expression 7.

[mathematical formula 7]

$\mathrm{<span\; class="notranslate">\; \&alpha;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>}{\sqrt{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; wxya</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; \&gamma;</span>}$

Inputting a coefficient as a weighting value may be a value used to adjust the mixing degree of the inverse correlation signal. Therefore, α(m) can be defined in the range of 0≤α(m)≤γ, where γ is in the range of 0≤γ≤1.

And, _{the wet} L(m,k) and _wet R(m,k) may be generated as inversely correlated signals via an inverse correlation process performed by an inverse correlator.

Fig. 6 shows an inverse correlator according to an exemplary embodiment of the present invention.

The inverse correlator 600 according to the present invention is included in the integrated up-mixer 240 to form an element of the inverse correlation signal, as shown in FIG. 6 , it may include a complex transformer 610, an amplitude information extractor 620, a phase information extractor 630 , random timing memory 640 , window opener 650 , phase converter 660 , synthesizer 670 and inverse converter 680 .

The complex transformer 610 may perform complex transformation on the downmixed signal.

The amplitude information extractor 620 and the phase information extractor 630 respectively extract amplitude information and phase information from the downmixed signal transformed in the complex transformer 610 to separate the downmixed signal.

The window opener 650 is based on the envelope of the amplitude information extracted by the amplitude information extractor 620 to modify the phase information modeling spectral window, by using the spectral type for the random timing stored in the random timing memory 640 window open window.

At this time, the number of random timings stored in the random timing memory 640 is determined according to the number of downmixed signals. That is, in order to generate the _wet L(m, k) and _wet R(m, k), different random time series are used, and the degree of correlation between the two random time series used at this time is close to zero.

The phase converter 660 may assign a weighted value to the phase information extracted from the phase information extractor 630 by using the random timing of opening a window from the window opener 650 .

The synthesizer 670 may combine the amplitude information extracted from the amplitude information extractor 620 and the phase information from the phase transformer 660 to apply a weighted value.

The inverse transformer 680 inversely transforms the information combined in the synthesizer 670 to calculate an inverse correlation signal.

Fig. 7 shows a method for decoding multi-channel audio based on audio source position clues according to an exemplary embodiment of the present invention.

In step S710, the demultiplexer 210 receives the signal transmitted by the multiplexer 150, and dissects the received signal into a bit stream of stereo audio and a bit stream of additional information.

In step S720, the audio decoder 220 may restore the downmixed signal based on the bitstream of the audio parsed in step S710.

In step S730, the Huffman decoder 410 performs Huffman decoding on the bit stream of the additional information parsed in step S710 using Huffman's coding book to generate a difference index.

In step S740, the inverse quantizer 420 uses the inverse quantization table to inverse quantize the difference index generated in S730 to restore the additional information. Specifically, the inverse quantizer 420 can perform inverse quantization by mapping the VSLA information of each frame to a quantization table corresponding to each VSLA.

In step S750, the integrated up-mixer 240 uses the downmixed signal restored in step S720 and the additional information restored in step S740 to predict a multi-channel signal, based on the multi-channel signal The downmixed signal is upmixed to generate an upmixed signal.

In step S760, the windower 250 of the integrated filter bank performs an integrated filter bank on the upmixed signal generated in step S750 to extract the signal in the time domain, and opens the upmixed signal window to extract the output signal.

As described above, the multi-channel audio decoding device and method based on sound source location clues according to the exemplary embodiments of the present invention receive and compress the multi-channel audio signal, and convert the stereo Signal compression and transmission, allowing the transmission of multi-channel audio while providing backward compatibility with existing stereo audio coding.

Although specific exemplary embodiments of the present invention have been described for illustrative purposes, various modifications, additions, and and replace. The scope of the invention should therefore be defined by the appended claims and equal claims.

Claims (10)

1. decoding device based on the multichannel audio of sound source position clue, it comprises:

Demodulation multiplexer receives signal, and the above-mentioned signal that will receive is parsed into audio bitstream and additional information bits stream;

Audio decoder based on said audio bitstream, restores mixing signal down, and the following mixed signal that will restore sends to and comprehensively mixed device, and analysis filterbank is not applicable to the following mixed signal of said recovery;

Comprehensive going up mixed device, through from the said mixed signal down of recovery, separating amplitude information and phase information, based on said amplitude information the random sequence of having deposited carried out windowing; Weighted value is given to said phase information, based on said amplitude information and the said phase information that is endowed weighted value, and based on the said additional information bits stream that restores; Prediction multichannel signal; Based on said multichannel signal said down mixed signal is gone up and to be mixed, generate mixed signal, wherein said additional information bits is flowed and carry out Hofmann decoding and generate the difference index; And with the quantization table said difference index is carried out inverse guantization (IQ), to restore said additional information; With

The mouthpart of windowing of synthesis filter group uses the synthesis filter group to extract time field signal to the said signal that upward mixes, and the said signal that upward mixes is carried out windowing, extracts the output signal.

2. the decoding device of the multichannel audio based on the sound source position clue as claimed in claim 1, said audio decoder have and use time domain to mix the three-dimensional kernel codec of repeatedly cancelling the TDAC bank of filters.

3. the decoding device of the multichannel audio based on the sound source position clue as claimed in claim 1, the mouthpart of windowing of said synthesis filter group shines upon mutually for the analysis windowing with the kernel stereo audio, and said going up mixed signal and carried out windowing.

4. the decoding device of the multichannel audio based on the sound source position clue as claimed in claim 1, said comprehensive going up mixed the said mixed signal down of device complex transformation, separated amplitude information and phase information the following mixed signal of complex transformation from said.

5. the decoding device of the multichannel audio based on the sound source position clue as claimed in claim 1; Said comprehensive going up mixed the envelope of device based on said amplitude information; For revising phase information, modeling frequency spectrum type window uses the random sequence of having deposited that said frequency spectrum type window is carried out windowing; And use the random sequence of windowing, weighted value is given to said phase information.

6. the decoding device of the multichannel audio based on the sound source position clue as claimed in claim 1; The said comprehensive device that upward mixes interosculates said amplitude information and the said phase information that is endowed weighted value; Said amplitude information and the said phase information that is endowed weighted value to combining carry out contrary complex transformation, predict the multichannel signal.

7. coding/decoding method based on the multichannel audio of sound source position clue, it comprises:

The signal that receives is parsed into audio bitstream and additional information bits stream;

Based on said audio bitstream, restore mixing signal down, but analysis filterbank is not applicable to said mixed signal down;

Said additional information bits stream is carried out Hofmann decoding, generate the difference index;

With the quantization table said difference index is carried out inverse guantization (IQ), to restore additional information;

Separation amplitude information and phase information from the said mixed signal down that restores; Based on said amplitude information the random sequence of having deposited is carried out windowing; Weighted value is given to said phase information; Based on said amplitude information and the said phase information that is endowed weighted value, and, predict the multichannel signal based on the said additional information of restoring;

Based on said multichannel signal, upward mix said mixed signal down, mix signal to generate;

Use the synthesis filter group to extract time field signal to the said signal that upward mixes; With

The said signal that upward mixes is carried out windowing, extract the output signal.

8. the coding/decoding method of the multichannel audio based on the sound source position clue as claimed in claim 7, said separating step comprises: the said mixed signal down of complex transformation; With separated amplitude information and phase information from said the following mixed signal of complex transformation.

9. the coding/decoding method of the multichannel audio based on the sound source position clue as claimed in claim 7, said step of giving weighted value comprises: based on the envelope of said amplitude information, be to revise phase information, modeling frequency spectrum type window; Use the random sequence of having deposited that said frequency spectrum type window is carried out windowing; With the random sequence of using windowing, weighted value is given to said phase information.

10. the coding/decoding method of the multichannel audio based on the sound source position clue as claimed in claim 7, the step of said prediction multichannel signal comprises: said amplitude information and the said phase information that is endowed weighted value are interosculated; With being carried out, said amplitude information that combines and the phase information that is endowed weighted value predict the multichannel signal against complex transformation.

Images