CN103000179B - Multichannel audio coding/decoding system and method - Google Patents

Abstract

The invention relates to a multi-channel audio codec system and its method. The multi-channel signal is converted into two pairs of stereo signals before and after, and the dynamic information of simulating the slight rotation of the human head is added, and the two pairs of stereo signals before and after modulation are changed accordingly, without affecting the The listening effect of ordinary stereo after mixing, at the same time, the receiving end can also solve the dynamic information, distinguish the two pairs of stereo before and after, and then distribute it to the multi-channel system for playback, so as to achieve good surround sound playback, which can be used in broadcasting systems, in homes, cars, etc. Private spaces enable good surround sound reproduction. The present invention embeds the dynamic information simulating the slight rotation of the human head during encoding, on the one hand, it does not affect the listening effect of the down-mixed stereo signal; source location.

Description

A multi-channel audio codec system and method thereof

technical field

The invention relates to the technical field of acoustic signal processing, in particular to a multi-channel audio codec system and method thereof.

Background technique

With the wide application of multi-channel digital audio codec technology and the enrichment of multi-channel surround sound programs, how to enjoy high-quality surround sound programs through traditional transmission channels (such as broadcasting) in private spaces such as homes and cars is gradually being paid attention to .

Since the 1970s, Matrix Surround codec was proposed for the compatibility between multi-channel signals and two-channel stereo. Through the coding matrix, the relationship between the left and right channels of the down-mixed stereo corresponds to the spatial information of the sound source. , the decoding matrix restores the multi-channel signal with the same spatial positioning as the original surround sound as much as possible. Based on this, a series of commercial systems such as "Dolby Pro Logic", "Lexicon Logic7", "SRS Circle Surround", "DTS Neo 6" and so on were developed.

MPEG Surround (ISO/IEC 23003-1) that has appeared in recent years also uses matrix surround codec as a working mode. Based on a large number of audio sample tests, it restores multi-channel by calculating the sound level difference and correlation between channels. audio signal, but still cannot avoid the defects of matrix surround codec. After years of development, the matrix surround sound system has many forms, but its encoding method is similar. The following is an example of a 5.1 system to illustrate its principle.

d=Ms (1)

in,

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>\left(\begin{array}{ccccc}\mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}}& \mathrm{<span\; class="notranslate">\; j</span>}\frac{\sqrt{\mathrm{<span\; class="notranslate">\; 3</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}& \mathrm{<span\; class="notranslate">\; j</span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; j</span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; j</span>}\frac{\sqrt{\mathrm{<span\; class="notranslate">\; 3</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}\right)<span\; class="notranslate">\; ,</span>$ $\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\\ \mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\end{array}\right]$ $\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{c}{\mathrm{<span\; class="notranslate">\; l</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

M is the encoding matrix, which converts the multi-channel signal into a two-channel stereo signal; s is the multi-channel signal, l(t), r(t), c(t), l _s (t), rs _s (t) represent Left front, right front, center front, left rear, and right rear 5 audio channel signals; d is the down-mixed two-channel stereo signal, l _t (t), r _t (t) respectively represent the left and right channel signals of the stereo.

Assuming that the virtual sound image in a certain direction is played by the nearest two speakers, the signal distribution follows the principle of stereo signal distribution, and then the amplitude ratio of the left and right stereo channels after down-mixing is calculated, as shown in Figure 1, and the abscissa represents a virtual sound image. An azimuth angle, and the ordinate represents the amplitude ratio of the left and right channels.

It can be seen from Figure 1 that the amplitude ratio of the left and right signals after downmixing by the encoding matrix corresponds to the position of the set virtual sound source one by one, and the spatial information of the original sound source is preserved. source localization, but only in the case of a single source. However, when there are multiple sound sources and background sounds in practice, the stereo signal mixed down by the matrix codec technology often suffers from serious crosstalk interference and loss of positioning information, especially the sound quality and positioning information of the original surround channel. Damage, there is too much crosstalk between the channels of the decoded output signal, resulting in poor sound source localization, sound coloring, and no clear localized background surround sound. In addition, if the sound image is in the rear half plane, the phases of the left and right channels of the down-mixed stereo signal are reversed, which affects the normal listening of ordinary users.

Contents of the invention

The purpose of the present invention is that the present invention proposes a multi-channel audio codec system and its method, which converts the multi-channel signal into two pairs of stereo signals before and after, simulates the dynamic information of the slight rotation of the human head, and makes corresponding changes to the two pairs of stereo signals before and after modulation. , the receiving end decodes the dynamic information, distinguishes the front and back stereo pairs, and then distributes them to the multi-channel system for playback.

Generally speaking, sound source localization is mainly determined by the following factors: Interaural Time Difference (ITD for short), Interaural Level Difference (ILD for short), spectral factor (SpectralCue), dynamic factor (Dynamic Cue), etc. . Among them, the dynamic factor depends on the change of ITD and ILD caused by head rotation to localize the sound source. If the change conforms to the changing law of the front sound source, people can identify the front sound image, and vice versa.

When people listen to the sound, there will be some small rotations of the head unconsciously, which will cause changes in ITD (binaural time difference) and ILD (binaural sound level difference). According to such dynamic information, the front and rear sound sources can be distinguished without affecting listening Effect. Also in encoding, in order to distinguish the front and back without affecting the listening effect, the stereo signal corresponding to the sound source can be modulated to simulate the slight rotation of the human head, so as to achieve more accurate sound source positioning for multi-channel encoding and decoding. This application simulates the slight rotation of the human head by modulating the amplitude values of the left and right channels of the front and rear stereo pairs.

When converting stereo sound to multi-channel audio, some researchers (Christof Faller, Matrix Surround Revisited, the AES 30th International Conference, Saariselka, Finland, 2007) proposed to separate the direct sound from the background sound first, improve the signal-to-noise ratio of the direct sound, and then Then convert the direct sound from stereo to multi-channel.

The main steps are as follows:

⑴ Encoding end:

Divide the multi-channel audio signal into the front signal and the rear signal, and then according to the signal distribution principle, distribute the front signal to the front pair of virtual stereos, and the rear signal to the rear pair of virtual stereos. Get two pairs of stereo front and rear.

·Set the head rotation mode (direction, sequence), and the front and rear stereo pairs correspond to their respective rotation modes. The left and right channel amplitudes of the front and rear stereo signals are modulated to simulate the tiny rotation of the human head back and forth. If only two frames of signals before and after can be distinguished, the amplitude changes of the two pairs of stereo left and right channels must be different before they can be distinguished. If the signals of several frames before and after are used for discrimination, the variation trends of the amplitudes of the two pairs of stereo left and right channels in these frames must be different and cannot be completely the same.

·Add the left channel and the right channel of the front and rear stereo pairs respectively, and mix down to form a stereo pair.

⑵ Decoder:

·Convert the stereo signal to the time-frequency domain for processing, separate the direct sound and the background sound in the down-mixed dual-channel signal (each obtain a pair of stereo signals), and improve the signal-to-noise ratio of the direct sound.

At the same time, the original dual-channel stereo signal is divided by the left and right channels at each time-frequency point, observe the changes in the ratio of the left and right channels in the two frames (or more frames) before and after, and then compare the ratio of the two pairs of stereo left and right channels set before and after The change of the frequency point is judged to belong to the front or the rear, and the time-frequency point of the original dual-channel stereo is replaced by the time-frequency point of the corresponding direct sound. Finally, the time-frequency domain signal is inversely transformed into the time-domain signal, and the corresponding Two pairs of stereo signals front and rear of the direct sound.

·Signal distribution: distribute the two pairs of direct sounds to the nearest two channels according to their orientation; add the background sound to the left front and right front channels directly; the background sound is delayed and distributed through a low-pass filter To the left and right rear passages.

On this basis, the present invention provides a kind of multi-channel audio codec system, comprising: multi-channel audio sound source, loudspeaker, acoustic signal AD/DA converter, power amplifier, multi-channel audio encoder, multi-channel audio decoding Device and acoustic signal transmission equipment, characterized in that:

The multi-channel audio sound source is used to generate a multi-channel audio signal, which is output to the input end of a multi-channel audio encoder;

The multi-channel audio encoder is used to divide the multi-channel audio signal into a front signal and a rear signal, distribute the front signal to a pair of virtual stereo signals in the front, and distribute the rear signal to a pair of virtual stereo signals in the back, to obtain front and rear Two pairs of stereo signals; then, to simulate the slight rotation of the human head, the amplitudes of the left and right channels of the front and rear pairs of stereo signals are respectively modulated according to the set front and rear head rotation mode. , there must be a difference; finally, the two pairs of stereo signals before and after are down-mixed into a pair of stereo signals and output to the sound signal transmission equipment;

The sound signal transmission device is used to transmit a pair of stereo signals output by the multi-channel audio encoder to the input end of the multi-channel audio decoder;

The multi-channel audio decoder decomposes a pair of stereo signals into two pairs of front and rear stereo signals according to the front and rear rotation mode set in the multi-channel audio encoder, and then restores the two pairs of stereo signals according to the speaker configuration by using the principle of signal distribution. Output a multi-channel audio signal with the same spatial positioning as the original surround sound;

The power amplifier is used to amplify the multi-channel audio signal output by the decoder and the acoustic signal DA converter, and output the signal to the speaker.

As an improvement of the above technical solution, the multi-channel audio encoder and the multi-channel audio decoder adopt common PCs or digital signal processors.

As an improvement of the above technical solution, the loudspeaker includes two or more loudspeaker arrays for playing the decoded multi-channel audio signal.

The present invention also provides a multi-channel audio encoding and decoding method. At the encoding end, the multi-channel signal is converted into two pairs of stereo signals before and after, and then the dynamic information simulating the slight rotation of the human head is added to modulate the two pairs of stereo signals to make corresponding The change of the signal is mixed down into a pair of stereo signal output, and the dynamic information is solved according to the stereo signal at the decoding end, and the front and rear pairs of stereo are distinguished, which specifically includes the following steps:

(1) Output the multi-channel audio source signal to the input end of the multi-channel audio encoder;

(2) Divide the multi-channel audio signal into the front signal and the rear signal through the multi-channel audio encoder, and then according to the signal distribution principle, distribute the front signal to the front pair of virtual stereo signals, and the rear signal to the rear pair of virtual stereo signals signal, to obtain two pairs of stereo signals before and after;

(3) Simulate the tiny rotation of the human head, according to the set front and rear head rotation mode, modulate the amplitude of the left and right channels of the front and rear pairs of stereo signals according to a certain frequency and rotation mode; the amplitude of the front and rear pairs of stereo left and right channels The variation values cannot be the same, there must be a difference.

(4) Add the left channel and the right channel of the two pairs of stereo signals before and after, mix them down into a pair of stereo signals, and transmit them to a multi-channel audio decoder using an acoustic signal transmission device;

(5) The multi-channel audio decoder decomposes a pair of stereo signals into two pairs of front and rear stereo signals according to the front and rear rotation mode set in the multi-channel audio encoder, and then restores the two pairs of stereo signals according to the speaker configuration by using the signal distribution principle Output a multi-channel audio signal with the same spatial positioning as the original surround sound;

(6) Through the sound signal DA converter and power amplifier, it is output to the speaker for playback.

As an improvement of the above technical solution, the rotation frequency of the simulated human head in step 3) is generally above 50 Hz, so that the listener does not feel the movement of the sound image.

As an improvement of the above technical solution, the step 5) converts the two-channel stereo signal to the time-frequency domain through the multi-channel audio decoder, performs time-frequency transformation on the signal, and converts the left and right channels at each time-frequency point In addition, observe the changes in the left and right channel ratios of two or more frames before and after, and then compare the changes in the ratios of the two pairs of stereo left and right channels set before, to determine whether the frequency point belongs to the front or the rear, and finally reverse the time-frequency domain signal Convert to time-domain signals, solve two pairs of stereo signals, front and rear, and convert the two pairs of stereo signals into multi-channel audio signals according to the speaker configuration.

As an improvement of the above technical solution, the step 5) converts the two-channel stereo signal to time-frequency domain processing through a multi-channel audio decoder, and adopts a signal processing method to convert the direct sound component in the down-mixed two-channel signal Separate from the background sound components to obtain a pair of stereo signals, which can improve the signal-to-noise ratio of the direct sound; at the same time, the original two-channel stereo signal divides the left and right channels at each time-frequency point, and observes the ratio of the left and right channels of two or more frames before and after Then compare the changes in the ratio of the two pairs of stereo left and right channels set before, and judge that the frequency point belongs to the front or rear, and replace the time-frequency point of the original dual-channel stereo with the time-frequency point of the corresponding direct sound , and finally inversely transform the time-frequency domain signal to the time-domain signal, and solve the two pairs of stereo signals corresponding to the direct sound, front and rear; transform the two pairs of direct sound stereo and one pair of background sound stereo into multi-channel audio signals according to the speaker configuration.

As an improvement of the above technical solution, the step 5) assigns the two pairs of direct sounds to the nearest two channels according to their orientation through the principle of stereo signal distribution; directly adds the background sound to the left front and right front channels; the background sound Do a certain delay, pass through a low-pass filter, and distribute it to the left and right rear channels.

As a kind of improvement of above-mentioned technical scheme, described method comprises steps:

Step 301: Multi-channel audio encoder, digital signal processor 202 reads 5.1 format multi-channel audio source 201, left front, right front, center front, bass, left rear, right rear channel signals are L, R, C, LFE respectively , Ls, and Rs, the sampling frequency is 44100Hz;

Step 302: Assuming that there are 4 virtual speakers 401, 402, 403 and 404 in the directions of ±45 degrees and ±135 degrees, according to the simple signal distribution principle, convert the 5.1-channel signal to the hypothetical 2 pairs of virtual speakers, two pairs before and after Stereo are:

${\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; *</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; LFE</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

${\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; *</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; LFE</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

L _b = L _s ;

R _b = R _s ;

Step 303: Set the rotation mode of the front and rear heads to simulate the slight rotation of the human head. When the head rotates counterclockwise, it is equivalent to the simultaneous rotation of four virtual speakers 401, 402, 403, and 404 clockwise, and the corresponding signals of each channel changes to:

Let the stereo of the front half plane L _f ′=a ₁ *L _f , R _f ′=a ₂ *R _f ,

Stereo L _b ′=a ₃ *L _b ,R _b ′=a ₄ *R _b for the rear half plane;

Among them, the selection of a ₁ , a ₂ , a ₃ , and a ₄ should conform to the fact that the orientation of the sound image after modulation is basically unchanged, but the width of the sound image is slightly wider, which is not much different from the perception of the human ear before modulation. In order to make the sound The intensity and effect of the two do not change greatly, and the value is between 0.9-1.1. (a ₁ , a ₂ ) one of the values is greater than 1, and the other is less than 1. (a ₃ , a ₄ ) is the same . Who is big and who is small depends on the direction of rotation of the speaker;

When the head turns clockwise, the changes of the front and rear pairs of stereo left and right channels are opposite;

Stereo L _f '=a ₂ *L _f for the front half plane, R _f '=a ₁ *R _f ,

Stereo L _b ′=a ₄ *L _b ,R _b ′=a ₃ *R _b for the rear half plane;

Step 304: Mix down two stereo pairs into a stereo pair: L=L _f ′+L _b ′, R=R _f ′+R _b ′,

Step 305: Use the broadcast transmitter 203 to convert the (L, R) stereo pair into a wireless signal and transmit it;

Step 306: The user uses the broadcast receiving terminal 205 to receive the (L, R) stereo pair, demodulates it into a digital signal, and inputs it to the multi-channel audio decoder;

Step 307: The multi-channel audio decoder adopts the second decoding method, and the digital signal processor 206 performs time-frequency conversion on the stereo signal to separate the direct sound component (L _d , R _d ) and the background sound component (L _a , R _a ), improving the signal-to-noise ratio of the direct sound;

Step 308: Simultaneously observe the changes in the left and right channel ratios of two or more frames before and after the downmixed stereo (L, R), and then compare the changes in the ratios of the two pairs of stereo left and right channels set before to determine the frequency point Belonging to the front or rear, replace the time-frequency points of the mixed stereo (L, R) with the time-frequency points of the corresponding direct sound (L _d , R _d ), and finally inversely transform the time-frequency domain signal to the time-domain signal, so Obtain stereo signals of two pairs of direct sound components before and after;

Step 309: Convert two pairs of direct sound stereo signals and a pair of background sound stereo signals into 5.1 multi-channel audio signals according to the speaker playback system 209;

Distribute each channel signal of the direct sound to the nearest two speaker channels through the principle of stereo signal distribution; add the background sound directly to the left front L and right front R channels; do a certain delay for the background sound, and distribute it to the left rear Ls and right rear Rs channel: low-pass filter the front pair of direct sounds, add them, and distribute them to the bass channel LFE.

The method and system of the present application embed the dynamic information that simulates the slight rotation of the human head during encoding, on the one hand, it does not affect the listening effect of the down-mixed stereo signal, on the other hand, when restoring the multi-channel signal, it can solve the dynamic information and distinguish the sound source Before and after the sound source positioning.

The advantage of the present invention is that the micro-rotation of the simulated human head of the present invention realizes the conversion between multi-channel audio signals. Compared with the prior art, the advantages are:

1. Compared with the existing matrix codec technology, the present invention can better localize the rear sound source and improve the sense of spatial localization of the sound source.

2. The present invention is well compatible with the playback of stereo signals, and solves the problems caused by the phase distortion of matrix encoding and decoding.

In short, the present invention aims at the deficiencies of the traditional Matrix Surround encoding and decoding, converts the multi-channel signal into two pairs of stereo signals before and after, simulates the dynamic information of the small rotation of the human head, and makes corresponding changes to the two pairs of stereo signals before and after modulation , which does not affect the listening effect of ordinary stereo after down-mixing, and at the same time, the receiving end can also solve the dynamic information, distinguish the two pairs of stereo before and after, and then distribute it to the multi-channel system for playback to achieve good surround sound playback, which can be used in broadcasting systems. It can realize very good surround sound playback in private spaces such as homes and cars.

Description of drawings

Fig. 1 is a schematic diagram of the amplitude ratio of the left and right channels after down-mixing in the existing matrix surround coding;

Fig. 2 is a schematic diagram of an embodiment of the codec system of the present invention;

Fig. 3 is a flowchart of an embodiment of the encoding and decoding method of the present invention;

Fig. 4 is a schematic diagram of converting 5.1 multi-channel signals into two pairs of stereo before and after;

Fig. 5 is a human head rotation model;

Fig. 6 present invention uses the ITU recommended matrix encoding to mix the left and right channel amplitude ratio schematic diagram;

Fig. 7 is the stereo signal after down-mixing in the simulation example;

Fig. 8 is the background sound separated from the stereo in the simulation example;

Figure 9 is the front and rear sound sources separated from the direct sound in the simulation example.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings. The above-mentioned and other features and advantages of the present invention will become clearer by referring to the detailed description of the embodiments of the present invention below with reference to the accompanying drawings, wherein:

In order to realize one of the purposes of the present invention, a set of practical system of multi-channel audio codec is provided, comprising: multi-channel audio sound source, loudspeaker, sound signal AD/DA converter, power amplifier, multi-channel audio coder, multiple Channel audio decoder, acoustic signal transmission equipment.

The multi-channel audio sound source can be collected on-site by a microphone array or artificially produced in a later stage, and output to the input end of a multi-channel audio encoder.

The multi-channel audio encoder first divides the multi-channel audio signal into a front signal and a rear signal, and then converts them into two corresponding pairs of stereo signals; simulates the slight rotation of the human head, and sets the front and rear head rotation modes (sequence, direction), respectively modulate the amplitudes of the left and right channels of the front and rear pairs of stereo signals; finally, in order to be compatible with stereo transmission and playback, the front and rear pairs of stereo signals are down-mixed into a pair of stereo signals.

The sound signal transmission device may be a transmitting and receiving channel of wireless broadcasting or a transmission circuit such as a network, which transmits the stereo signal output by the multi-channel audio encoder to the input end of the multi-channel audio decoder.

The multi-channel audio decoder decomposes a pair of stereo signals into two pairs of front and rear stereo signals according to the front and rear rotation mode set in the encoder, and then transforms the two pairs of stereo signals into multi-channel audio signals according to the speaker configuration.

The power amplifier is used to amplify the multi-channel audio signal output by the decoder and the acoustic signal DA converter, and connect the signal to the loudspeaker.

The loudspeaker includes two or more loudspeaker arrays for playing the decoded multi-channel audio signal.

In the above technical solution, the multi-channel audio encoder and multi-channel audio decoder adopt common PC or digital signal processor.

The new multi-channel audio encoding and decoding method of the present invention simulates the dynamic information of the slight rotation of the human head at the encoding end, modulates the multi-channel signal to make corresponding changes, mixes it downwards into a pair of stereo signal output, and then decodes it according to the stereo signal at the decoding end. Dynamic information, to distinguish the front and rear 2 pairs of stereo. Including the following steps:

(1) Output the multi-channel audio source signal to the input end of the multi-channel audio encoder.

(2) Encoder: first divide the multi-channel audio signal into the front signal and the rear signal, and then use the signal distribution principle to convert them into two corresponding pairs of stereo signals;

(3) Simulate the slight rotation of the human head, set the front and rear head rotation modes (order, direction), and the rotation modes corresponding to the front and rear stereo can be the same or different. In order for the listener not to feel the movement of the sound image, the rotation frequency is generally above 50Hz. According to a certain frequency and rotation mode, the amplitudes of the left and right channels are respectively modulated for the front and rear pairs of stereo signals;

(4) Down-mix the two pairs of front and rear stereo signals into a pair of stereo signals. Transmit to a multi-channel audio decoder using an acoustic signal transmission device.

(5) Decoder: There are two modes (can be selected according to requirements)

1. Convert the dual-channel stereo signal to the time-frequency domain for processing, perform time-frequency transformation on the signal, divide the left and right channels at each time-frequency point, observe the changes in the ratio of the left and right channels in the two frames (or more frames) before and after, and then Comparing the changes in the ratio of the two pairs of stereo left and right channels set before, it is judged that the frequency point belongs to the front or the rear, and finally the time-frequency domain signal is inversely transformed into a time-domain signal. Solve two pairs of stereo signals, front and rear. Converts two stereo pairs into a multi-channel audio signal according to the speaker configuration.

2. Convert the dual-channel stereo signal to time-frequency domain processing, and use the signal processing method to separate the direct sound component and the background sound component in the down-mixed dual-channel signal (to obtain a pair of stereo signals for each), which can improve the direct sound signal noise ratio. At the same time, the original dual-channel stereo signal divides the left and right channels at each time-frequency point, observes the changes in the ratio of the left and right channels in the two frames (or more frames) before and after, and then compares the changes in the ratio of the two pairs of stereo left and right channels set before and after. , it is judged that the frequency point belongs to the front or the rear, and the time-frequency point of the original two-channel stereo is replaced by the time-frequency point of the corresponding direct sound, and finally the time-frequency domain signal is inversely transformed into a time-domain signal to solve the corresponding direct sound Two pairs of stereo signals front and rear. Converts two direct stereo pairs and one background stereo pair into a multi-channel audio signal according to the speaker configuration.

(6) After the sound signal DA converter, the power amplifier outputs to the speaker for playback.

The specific implementation plan is as follows:

As shown in Figure 2, an embodiment system of the present invention has included the multi-channel audio sound source 201 of a group of 5.1 systems, 5.1 channel loudspeaker playback system 209, power amplifier 208, acoustic signal AD/DA converter 207, here The AD/DA converter adopts a common computer built-in sound card or an external sound card, the broadcast transmitter 203, the broadcast receiver 205, the sound signal transmission device 204 is wireless transmission (medium wave, short wave, FM), multi-channel audio coding The decoder 202 and decoder 206 utilize digital signal processors within the broadcasting station.

As shown in Figure 3, the specific steps of the multi-channel audio codec system in this example are as follows:

Step 301: Encoder, digital signal processor 202 read 5.1 format multi-channel audio source 201, left front, right front, center front, bass, left rear, right rear channel signals are L, R, C, LFE, Ls, respectively Rs. The sampling frequency of this embodiment is 44100Hz.

Step 302: As shown in Figure 4, assuming that there are 4 virtual speakers 401, 402, 403, 404 in the directions of ±45 degrees and ±135 degrees, according to the simple signal distribution principle, convert the 5.1-channel signal into the hypothetical 2 pairs of virtual speakers On the loudspeaker, in this embodiment, since the rear channel Ls and Rs already constitute a pair of stereo signals, there is no need to convert the rear channel, and the original rear channel signal can be used directly.

The front and back stereo pairs are:

${\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; *</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; LFE</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

${\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; f</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\sqrt{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; *</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; LFE</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 45</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; *</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 180</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

L _b = L _s ;

R _b =R _s .

Step 303: Set the rotation mode (steering, sequence) of the front and rear heads, simulating the small rotation of the human head, as shown in Figure 5, when the head rotates counterclockwise, it is equivalent to the speakers 401, 402, 403, 404 clockwise at the same time Rotate, the corresponding signal of each channel also changes:

Let the stereo of the front half plane L _f '=a ₁ *L _f , R _f ′=a ₂ *R _f ,

Stereo L _b ′=a ₃ *L _b ,R _b ′=a ₄ *R _b for the rear half plane;

Among them, the selection of a ₁ , a ₂ , a ₃ , and a ₄ needs to conform to the fact that the orientation of the sound image after modulation is basically unchanged, but the width of the sound image is slightly wider, which is not much different from the perception of the human ear before modulation. In order to keep the intensity and effect of the sound from changing greatly, the value is generally between 0.9-1.1. (a ₁ , a ₂ ) one of the values is greater than 1, and the other value is less than 1, and (a ₃ , a ₄ ) is the same. Who is big and who is small depends on the rotation direction of the speaker. In this embodiment, if the speaker rotates clockwise, a _{1 <a 2 , a 3 >} a ₄ . In this embodiment, a ₁ =0.95, a ₂ =1.05, a ₃ =1.05, a ₄ =0.95.

When the head is turned clockwise, the changes in the anterior and posterior channels are reversed.

Stereo in the front half plane L _f '=a ₂ *L _f , R _f ′=a ₁ *R _f ,

Stereo L _b ′=a ₄ *L _b ,R _b ′=a ₃ *R _b for the rear half plane;

The down-mixing process is equivalent to using the ITU recommended matrix $m=\left(\begin{array}{ccccc}1& 0& 1/\sqrt{2}& a& 0\\ 0& 1& 1/\sqrt{2}& 0& a\end{array}\right),$ where a=1.

As shown in Figure 6, when a=1, the relationship between the ratio of the amplitude of the left and right channels and the angle of the sound source. Based on this stereo signal, it is impossible to locate the front and rear positions of the sound source, but adding dynamic information and solving this problem at the decoding end can solve this problem. This method utilizes head dynamic information to first distinguish the front and rear sound sources, and then according to Figure 6, the ratio of the front and rear sound source angles to the left and right channels is one-to-one correspondence, thereby determining the specific location of the sound source.

In this embodiment, we set the head to turn clockwise first, then counterclockwise, and turn back and forth in turn. The head rotation pattern corresponding to the front and rear stereo is the same. When the rotation frequency is greater than 50 Hz, the human ear can hardly feel the movement of the sound image. In this embodiment, the sampling frequency is 44100 Hz, and the rotation is selected every 448 points in the time domain.

Step 304: Mix down two stereo pairs into a stereo pair: L=L _f ′+L _b ′, R=R _f ′+R _b ′,

Step 305: Convert the stereo pair (L, R) into a wireless signal by the broadcast transmitter 203 and transmit it.

Step 306: The user uses the broadcast receiving terminal 205 to receive the (L, R) stereo pair, demodulates it into a digital signal, and inputs it to the multi-channel audio decoder.

Step 307: Decoder (using the second decoding method), the digital signal processor 206 performs time-frequency conversion on the stereo signal, and separates the direct sound component (L _d , R _d ) and the background sound component (L _a , R _a ), The signal-to-noise ratio of the direct sound is improved.

Step 308: Simultaneously observe the changes in the left and right channel ratios of the downmixed stereo (L, R) two frames (or more frames) before and after, and then compare the changes in the ratios of the two pairs of stereo left and right channels set before and after, to determine the The frequency point belongs to the front or the rear, replace the time-frequency point of (L, R) with the time-frequency point corresponding to the direct sound (L _d , R _d ), and finally inversely transform the time-frequency domain signal into the time-domain signal, so that the front and rear Stereo signal of two pairs of direct sound components.

Step 309: Convert two pairs of direct sound stereo signals and a pair of background sound stereo signals into 5.1 multi-channel audio signals according to the speaker playback system 209 .

Distribute each channel signal of the direct sound to the nearest two speaker channels through the principle of stereo signal distribution; add the background sound directly to the left front L and right front R channels; make a certain delay for the background sound and distribute it to the left rear Ls and right rear Rs channel: low-pass filter the front pair of direct sounds, add them, and distribute them to the bass channel LFE.

The simulated 5.1 system sound source signal in this embodiment is also applicable to the simulation of other multi-channel audio sound sources, and the above steps can be used to realize multi-channel audio encoding and decoding.

The human head rotation mode set in this embodiment is the same mode at the front and rear, that is, it rotates clockwise first, then counterclockwise, and then rotates back and forth in turn. It is also possible to set different rotation modes for the front and rear heads, such as setting the direction and order of rotation according to the value of the pseudo-random sequence (M sequence, Gold code, etc.). The selection of the head rotation mode should be able to distinguish the front and rear sound sources (the changes of the front and rear two pairs of stereo left and right channels are different).

Simulation

This experiment uses Matlab software simulation. Use dual sound sources for testing: Assume that the front stereo is a female voice at a 30-degree orientation, the rear stereo is a male voice at a 40-degree orientation, and white noise is used as the background sound.

Encoder: Modulate the front and rear stereo left and right channel signals respectively, and then down-mix them into a pair of stereo, as shown in Figure 7.

Decoder: first separate the direct sound and the background sound (the background sound is shown in Figure 8), and at the same time distinguish a pair of direct sounds into two pairs of stereo before and after according to the dynamic information of the head at the encoding end, as shown in Figure 9.

Conducting subjective listening experiments, the result is: the background sound is separated from the direct sound to a certain extent, and the signal-to-noise ratio of the direct sound is improved; the two pairs of stereo before and after listening to it are obviously able to distinguish between male and female sound sources, but there are still slight differences. When using a multi-channel playback system to replay, due to the masking effect, almost no crosstalk can be heard, and the front and rear pairs of stereo are well distinguished.

To sum up, the present invention utilizes the surround sound codec that simulates the dynamic information of the human head to achieve high-quality conversion compatibility between multi-channel and dual-channel stereo, and can also adopt matrix surround encoding after adding dynamic information, which is compatible with matrix surround decoding. At present, the principle verification and methods of the two working modes have been preliminarily realized.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than limit them. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent replacements to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all of them should be included in the scope of the present invention. within the scope of the claims.

Claims (8)

1. a multi-channel audio coding/decoding system, comprising: multi-channel audio sound source, loudspeaker, acoustical signal AD/DA converter, power amplifier, multi-channel audio decoder, multi-channel audio demoder harmony signal transmission apparatus, is characterized in that:

Described multi-channel audio sound source, for generation of multi-channel audio signal, outputs to the input end of multi-channel audio decoder;

Described multi-channel audio decoder, for multi-channel audio signal is divided into front signal and rear signal, a pair of virtual three-dimensional acoustical signal, rear signal that front signal is assigned to are above assigned to a pair of virtual three-dimensional acoustical signal below, two stereophonic signal before and after obtaining; Then, the minor rotation of phantom bead, the amplitude of front and back two stereophonic signal being modulated respectively to its left and right passage according to the forward-and-rearward number of people rotation mode of setting, the changes in amplitude trend of the two pairs of stereo left and right passages in front and back is different; Finally, become a stereophonic signal to output to acoustical signal transmission equipment front and back two stereophonic signal downmixes;

Described acoustical signal transmission equipment, for being transferred to a stereophonic signal of multi-channel audio decoder output the input end of multi-channel audio demoder;

Described multi-channel audio demoder, according to the front and back rotation mode of setting in multi-channel audio decoder, two stereophonic signal before and after one stereophonic signal is resolved into, according to speaker configurations, utilize signal distribution principle by two pairs of stereo multi-channel audio signals of locating with original surround sound same space that recover again;

Described power amplifier, for amplifying the multi-channel audio signal of demoder and the output of acoustical signal DA converter, and outputs to loudspeaker by signal;

The step of the minor rotation of the phantom bead of described multi-channel audio decoder comprises:

Step 301: multi-channel audio decoder, digital signal processor (202) reads 5.1 form multi-channel audio sound sources (201), left front, right front, in before, bass, left back, right back channel signal be respectively L, R, C, LFE, Ls and Rs;

Step 302: supposing has 4 virtual speakers (401,402,403 and 404) in ± 45 degree, ± 135 degree orientation, according to simple signal distribution principle, 5.1 channel signals are transformed on 2 pairs of virtual speakers of hypothesis to stereo being respectively in two pairs of front and back:

$L_f=L*\cos ((30-45)*\mathrm{pi}/180)+\sqrt{2}/2*(C+\mathrm{LFE})+R*\cos ((30+45)*\mathrm{pi}/180)$

$R_f=L*\cos ((30+45)*\mathrm{pi}/180)+\sqrt{2}/2*(C+\mathrm{LFE})+R*\cos ((30-45)*\mathrm{pi}/180)$

L _b＝L _s；

R _b＝R _s；

Step 303: the rotation mode of setting front and back head, phantom bead minor rotation, when head rotates counterclockwise, be equivalent to 4 virtual speakers (401,402,403,404) and clockwise rotate simultaneously, accordingly the signal of each passage change into:

The stereo L of half-plane before order _f'=a ₁* L _f, R _f'=a ₂* R _f,

The stereo L of rear half-plane _b'=a ₃* L _b, R _b'=a ₄* R _b;

Wherein, a ₁, a ₂, a ₃, a ₄choose need to meet modulation after acoustic image orientation substantially constant, just dimension slightly broadens, than the impression of modulation forefathers ear without very large difference, in order to make intensity and the effect of sound not produce large variation, value between 0.9-1.1, a ₁or a ₂in a value be greater than 1, another value is less than 1, whose large whose rootlet determines according to the rotation direction of loudspeaker; a ₃or a ₄in a value be greater than 1, another value is less than 1, whose large whose rootlet determines according to the rotation direction of loudspeaker;

When head clockwise rotates, the variation of front and back passage is contrary;

The stereo L of front half-plane _f'=a ₂* L _f, R _f'=a ₁* R _f,

The stereo L of rear half-plane _b'=a ₄* L _b, R _b'=a ₃* R _b;

Step 304: two pairs of downmixes are stereo is a pair of stereo: L=L _f'+L _b', R=R _f'+R _b';

Step 305: utilize broadcast transmission end (203) to be converted into wireless signal transmission to stereo L and R this and go out;

Step 306: user uses broadcast receiving terminal (205) to receive this to stereo L and R, and is demodulated into digital signal and inputs to multi-channel audio demoder;

Step 307: multi-channel audio demoder is decoded, does time-frequency conversion by digital signal processor (206) stereophonic signal, separated direct sound wave composition L _dand R _dwith background sound composition L _aand R _a, improved the signal to noise ratio (S/N ratio) of direct sound wave;

Step 308: simultaneously observe stereo L and R front and back two frames or the more variation of multiframe left and right passage ratio after lower mixing, the variation of the two pairs of stereo left and right passage ratios in front and back that then set before contrast, while judging this, frequency belongs to the place ahead or rear, changes the stereo L after lower mixing and the time frequency of R into corresponding direct sound wave L _dand R _dtime frequency, finally by time-frequency domain signal inverse transformation to time-domain signal, the stereophonic signal of two pairs of direct sound wave compositions before and after obtaining like this;

Step 309: two pairs of direct sound wave stereophonic signals and a pair of background sound stereophonic signal are converted to 5.1 multi-channel audio signals according to loudspeaker Play System (209);

By each channel signal of direct sound wave, by stereophonic signal distribution principle, be assigned to two nearest loudspeaker channel; Background sound is directly added to left front L, right front R passage; Background sound is done certain delay, is assigned to left back Ls, right back Rs passage; After a pair of direct sound wave in the place ahead is done to low-pass filtering treatment, be added, be assigned to bass channel LFE.

2. multi-channel audio coding/decoding system according to claim 1, is characterized in that, described multi-channel audio decoder and multi-channel audio demoder adopt ordinary PC or digital signal processor.

3. multi-channel audio coding/decoding system according to claim 1, is characterized in that, described loudspeaker is for comprising two or more loudspeaker arrays, for the multi-channel audio signal after broadcast decoder.

4. a multi-channel audio decoding method, the method is two stereophonic signal before and after coding side is first converted into multi channel signals, then the multidate information that adds phantom bead minor rotation, modulate two stereophonic signal and make corresponding variation, downmix becomes a stereophonic signal output, in decoding end, according to stereophonic signal, solve multidate information, distinguish two pairs of front and back stereo, specifically comprise the steps:

(1) multi-channel audio sound-source signal is outputed to the input end of multi-channel audio decoder;

(2) by multi-channel audio decoder, multi-channel audio signal is divided into front signal and rear signal, then according to signal distribution principle, front signal is assigned to a pair of virtual three-dimensional acoustical signal above, rear signal is assigned to a pair of virtual three-dimensional acoustical signal below, two stereophonic signal before and after obtaining;

(3) minor rotation of phantom bead, according to the forward-and-rearward number of people rotation mode of setting, the amplitude of front and back two stereophonic signal being modulated respectively to its left and right passage according to certain frequency and rotation mode, the changes in amplitude trend of the two pairs of stereo left and right passages in front and back is different;

(4) front and back two stereophonic signal are carried out respectively to left passage addition, the addition of right passage, downmix becomes a stereophonic signal, uses acoustical signal transmission equipment to be transferred to multi-channel audio demoder;

(5) multi-channel audio demoder is according to the front and back rotation mode setting in multi-channel audio decoder, two stereophonic signal before and after one stereophonic signal is resolved into, according to speaker configurations, utilize signal distribution principle by two pairs of stereo multi-channel audio signals of locating with original surround sound same space that recover again;

(6) through acoustical signal DA converter, power amplifier, output to loudspeaker and play;

Described step (3) comprising:

Step 301: multi-channel audio decoder, digital signal processor (202) reads 5.1 form multi-channel audio sound sources (201), left front, right front, in before, bass, left back, right back channel signal be respectively L, R, C, LFE, Ls and Rs;

Step 302: supposing has 4 virtual speakers (401,402,403 and 404) in ± 45 degree, ± 135 degree orientation, according to simple signal distribution principle, 5.1 channel signals are transformed on 2 pairs of virtual speakers of hypothesis to stereo being respectively in two pairs of front and back:

$L_f=L*\cos ((30-45)*\mathrm{pi}/180)+\sqrt{2}/2*(C+\mathrm{LFE})+R*\cos ((30+45)*\mathrm{pi}/180)$

$R_f=L*\cos ((30+45)*\mathrm{pi}/180)+\sqrt{2}/2*(C+\mathrm{LFE})+R*\cos ((30-45)*\mathrm{pi}/180)$

L _b＝L _s；

R _b＝R _s；

Step 303: the rotation mode of setting front and back head, phantom bead minor rotation, when head rotates counterclockwise, be equivalent to 4 virtual speakers (401,402,403,404) and clockwise rotate simultaneously, accordingly the signal of each passage change into:

The stereo L of half-plane before order _f'=a ₁* L _f, R _f'=a ₂* R _f,

The stereo L of rear half-plane _b'=a ₃* L _b, R _b'=a ₄* R _b;

Wherein, a ₁, a ₂, a ₃, a ₄choose need to meet modulation after acoustic image orientation substantially constant, just dimension slightly broadens, than the impression of modulation forefathers ear without very large difference, in order to make intensity and the effect of sound not produce large variation, value between 0.9-1.1, a ₁or a ₂in a value be greater than 1, another value is less than 1, whose large whose rootlet determines according to the rotation direction of loudspeaker; a ₃or a ₄in a value be greater than 1, another value is less than 1, whose large whose rootlet determines according to the rotation direction of loudspeaker;

When head clockwise rotates, the variation of front and back passage is contrary;

The stereo L of front half-plane _f'=a ₂* L _f, R _f'=a ₁* R _f,

The stereo L of rear half-plane _b'=a ₄* L _b, R _b'=a ₃* R _b;

Step 304: two pairs of downmixes are stereo is a pair of stereo: L=L _f'+L _b', R=R _f'+R _b';

Step 305: utilize broadcast transmission end (203) to be converted into wireless signal transmission to stereo L and R this and go out;

Step 306: user uses broadcast receiving terminal (205) to receive this to stereo L and R, and is demodulated into digital signal and inputs to multi-channel audio demoder;

Step 307: multi-channel audio demoder is decoded, does time-frequency conversion by digital signal processor (206) stereophonic signal, separated direct sound wave composition L _dand R _dwith background sound composition L _aand R _a, improved the signal to noise ratio (S/N ratio) of direct sound wave;

Step 308: simultaneously observe stereo L and R front and back two frames or the more variation of multiframe left and right passage ratio after lower mixing, the variation of the two pairs of stereo left and right passage ratios in front and back that then set before contrast, while judging this, frequency belongs to the place ahead or rear, the time frequency of stereo L after lower mixing and R is changed into the time frequency of corresponding direct sound wave Ld and Rd, finally time-frequency domain signal inverse transformation is arrived to time-domain signal, the stereophonic signal of two pairs of direct sound wave compositions before and after obtaining like this;

Step 309: two pairs of direct sound wave stereophonic signals and a pair of background sound stereophonic signal are converted to 5.1 multi-channel audio signals according to loudspeaker Play System (209);

By each channel signal of direct sound wave, by stereophonic signal distribution principle, be assigned to two nearest loudspeaker channel; Background sound is directly added to left front L, right front R passage; Background sound is done certain delay, is assigned to left back Ls, right back Rs passage; After a pair of direct sound wave in the place ahead is done to low-pass filtering treatment, be added, be assigned to bass channel LFE.

5. multi-channel audio decoding method according to claim 4, is characterized in that, the rotational frequency of phantom bead is more than 50Hz in described step (3), and imperceptible acoustic image moves when allowing hearer use normal stereo to listen to.

6. multi-channel audio decoding method according to claim 4, it is characterized in that, described step (5) is transformed into time-frequency domain by multi-channel audio demoder by dual-channel stereo signal and processes, signal is done to time-frequency conversion, at each time-frequency left and right passage of naming a person for a particular job, be divided by, two frames or the more variation of multiframe left and right passage ratio before and after observing, the variation of the two pairs of stereo left and right passage ratios in front and back that then set before contrast, while judging this, frequency belongs to the place ahead or rear, finally time-frequency domain signal inverse transformation is arrived to time-domain signal, solve front and back two stereophonic signal, according to speaker configurations by two pairs of stereo multi-channel audio signals that are transformed into.

7. multi-channel audio decoding method according to claim 4, it is characterized in that, described step (5) is transformed into time-frequency domain by multi-channel audio demoder by dual-channel stereo signal and processes, adopt signal processing method by the direct sound wave composition in lower mixed double-channel signal and background sound component separating, respectively obtain a pair of stereophonic signal, can improve direct sound wave signal to noise ratio (S/N ratio); Simultaneously former dual-channel stereo signal is divided by each time-frequency left and right passage of naming a person for a particular job, two frames or the more variation of multiframe left and right passage ratio before and after observing, the variation of the two pairs of stereo left and right passage ratios in front and back that then set before contrast, while judging this, frequency belongs to the place ahead or rear, when former binary channels is stereosonic, frequency changes the time frequency of corresponding direct sound wave into, finally time-frequency domain signal inverse transformation is arrived to time-domain signal, solve front and back two stereophonic signal of corresponding direct sound wave; According to speaker configurations by two pairs of stereo multi-channel audio signals that are transformed into of the stereo and a pair of background sound of direct sound wave.

8. multi-channel audio decoding method according to claim 7, is characterized in that, described step (5) by its orientation, is assigned to nearest two passages by stereophonic signal distribution principle by two pairs of direct sound waves; Background sound is directly added to left front, right front passage; Background sound is done certain delay, by low-pass filter, is assigned to left back, right back passage.