JPH0319417A - Stereo sound coding device - Google Patents

Abstract

PURPOSE:To reduce the deterioration in the reproduced sound quality of a sub channel caused at the changeover of main and sub channels by providing independently an adaptive control section when a right channel is a main channel and an adaptive control section when a left channel is a main channel. CONSTITUTION:Right main channel adaptive control sections 5, 15 and left main channel adaptive control sections 6, 16 are provided respectively to sender and receiver sides. For example, when some of plural talkers use right side microphone 1 and others of the talkers use left side microphone 2 alternately, the right main channel adaptive control sections 5, 15 reproduce adaptively the sound in the indoor acoustic state at that time in the case of using the right side microphone and the left main channel adaptive control sections 6, 16 reproduce adaptively the sound in the indoor acoustic state at that time in the case of using the left side microphone. when the left and right channels arc replaced, the device is at a halt to keep the preceding state and when the right and left channels are replaced, the device is in immediate adaptive operation as the continuity of the preceding state. Thus, the sound quality deterioration in the sub channel at the channel changeover is reduced.

Description

[Detailed Description of the Invention] [Summary of the Invention] A stereo system in which a stereo audio signal captured in two channels is encoded by dividing it into a main channel signal with high power and a compressed signal (residual signal) of a sub channel with low power. Regarding the audio encoding method, in order to prevent deterioration of sound quality on the secondary channel side due to switching between the main and secondary channels, the voice signal obtained by the left and right sound collectors uses the higher power as the primary channel and the other as the secondary channel. A pseudo signal is calculated from the main channel signal using an adaptive filter, and the residual signal obtained by manually inputting the pseudo signal and the sub channel signal into an adder and the main channel signal are encoded and transmitted. In the stereo audio encoding device, two sets of independent adaptive control units including the adaptive filter and the adder calculate and compare the powers of the audio signals obtained from the left and right sound collection devices, and determine the power. The reason why is large is that a power calculation/comparison section that generates an output indicating either left or right is provided, and a control switching section that selects a right adaptive control section or a left adaptive control section according to the output of the comparison section. Constructed.

[Industrial application field]

The present invention takes a stereo audio signal captured in two channels, sets the one with higher power as the main channel, and the one with lower power as the sub-channel, and uses the former as is and the latter with a pseudo signal derived from the main channel signal. This invention relates to a stereo audio encoding method that takes a difference and encodes the difference.

In recent years, with the spread of video conferencing systems, the development of audio encoding devices that provide high-quality and realistic call quality is progressing. In order to increase the sense of reality with this system,
It is necessary to provide the speaker's voice in stereo to make the direction of the speaker clear. [Prior Art] FIG. 3 is a block diagram of a conventional stereo audio encoding device, in which left and right microphones 1. 2 to the multiplexing section 10 is the transmitting side, and the section from the demultiplexing section 11 to the left and right beam power 18.19 is the receiving side. There are two encoders 8 and 9 on the transmitting side, and one encoder 8 encodes the main channel audio signal input from the right microphone 1 as it is. At the same time, an adaptive filter (A
F) At 5A, create a pseudo signal that cancels (expects) most of the sub-channel signal from the main channel signal, subtract this from the sub-channel signal to convert it into a residual signal with less information, and convert this into a residual signal with less information. The other encoding unit 9 encodes. Note that the main channel signal refers to a signal with high power; if the speaker is near the right microphone 1, the right channel is the main channel, and if the speaker is near the left microphone 2, the left channel is the main channel. ．． The multiplexer lO multiplexes the output codes of the encoder 8.9,
The dividing part l1 separates this again. A transmission line is provided between the multiplexer 10 and the demultiplexer 11. The decoding section 12.13 corresponds to the encoding section 8.9, and one decoding section l2 decodes the main channel signal. Further, the other decoding unit l3 decodes the residual signal. The decoded main channel signal is sent directly to the right speaker l.
However, the residual signal is used to drive the left speaker 19 after adding a pseudo signal with the adder 15B and returning it to the sub-channel signal. 15A is an adaptive filter that multiplies the decoded main channel signal by a filter coefficient similar to that on the transmitting side to eliminate spurious signals.

(Problems to be Solved by the Invention) In the above-described system, switching between the main and sub channels is necessary because the speaker changes or moves and the speaker with higher power becomes the main channel.

In FIG. 3, this switching system is omitted from illustration.

When such main/sub channel switching is performed, the filters 5A. Since the coefficients of l5A are optimized by the residual signal before switching, it takes time to adapt to the state after switching, and noise is generated during that time. In addition, switching between the main and sub channels itself causes discontinuous changes in the audio level, so it is inevitable that the sound quality on the sub channel side will temporarily deteriorate due to these factors.

An object of the present invention is to provide two sets of adaptive control sections each consisting of an adaptive filter and an adder, and to use them properly depending on the combination of main and subchannels, thereby preventing deterioration in sound quality when switching between main and subchannels. It is.

[Means to solve the problem]

FIG. 1 is a diagram showing the principle of the present invention, in which (a) shows the transmitting side and (b) shows the receiving side. In (a), 1.2 is a sound collection device such as a left and right microphone, 3 is a power calculation/comparison unit that calculates the power of the left and right audio signals captured by these and compares the power, and 5 is a power calculation/comparison unit that calculates the power of the left and right audio signals captured by these devices, and 5 is the main channel for the right channel. 6, a right main channel adaptive control section used when the left channel is the main channel; 4.7 a control switching section for switching these human outputs; 8. 9 is an encoding section that encodes the main channel signal and the residual signal, respectively, as in the conventional case, and 10 is a multiplexing section.

In (b), ti is a separation unit, 12.13 is a decoding unit that decodes the main channel signal and the residual signal, respectively;
15.16 is the adaptive control diagram of the right main channel and left main channel corresponding to 5.6 on the transmitting side; 14.
17 is a control switching unit that switches these inputs and outputs; 18.1
9 are left and right speakers.

[Effect]

The left and right audio signals taken in by the sound collectors 1.2 placed a certain distance apart are subjected to a power calculation/comparison section 3 which calculates the signal power for each channel and compares the magnitude thereof. As a result of this comparison, the channel with the highest power is selected as the main channel. The control switching unit 4 on the input side selects the right main channel adaptive control unit 5 and the left main channel adaptive control unit 6 based on the main channel switching information obtained by the power calculation/comparison unit 3. The right main channel adaptive control section 5 performs adaptive control with the right channel as the main channel and the left channel as the sub channel. On the other hand, the left main channel adaptive control section 6 performs adaptive control using the left channel as the main channel and the right channel as the sub channel. The control switching section 7 on the output side switches the outputs (main channel signal and residual signal &[l) of the right main channel adaptive control section 5 and the left channel adaptive control section 6. Encoding units 8 and 9 encode the main channel signal and residual signal selected by the control switching unit 7. The multiplexing unit 10 multiplexes the encoded codes from the encoding units 8 and 9 and the main channel switching information from the power calculation/comparison unit 3 and transmits the multiplexed code.

On the other hand, on the receiving side, a main channel signal, a residual signal, and main channel switching information are separated in a separation unit l1, and the main channel signal and residual signal are decoded in a decoding unit 12.13. The control switching unit 14 on the input side switches between the right main channel adaptive control unit 15 and the left main channel adaptive control unit 16 based on the main channel switching information. The right main channel adaptation@control section 15 or the left main channel adaptive control section 16 calculates a sub-channel signal from the main channel signal and the residual signal, and outputs one of the adaptive control sections (a set of main channel signal and sub-channel signal). Only the output signal is selected by the control switching unit l7 on the output side and reproduced by the speaker 18 and 19. The adaptive control units 5 and 6 on the transmitting side are both composed of an adaptive filter and an adder, and the filter coefficients are optimized according to the main channel signal. If this adaptive control section 5.6 is used depending on the combination of main and sub-channels, and the filter coefficients of the other are left as they were while one is in use, the optimized filter coefficients will be activated immediately after switching the main and sub-channels. You can use it. For example, if we consider an example where multiple speakers speak alternately, one using the right microphone 1 and the other using the left microphone 2, when speaking using the right microphone, the right main channel is adapted to the room acoustic conditions at that time. The control units 5 and 15 are designed to adapt and reproduce the left main channel adaptive control units 6 and 16, so that when speaking using the left microphone, the left main channel adaptive control units 6 and 16 are configured to adapt and reproduce the room acoustic conditions at that time. When switching to the right, it pauses and maintains the previous state, and when switching to the left and right, it can immediately perform an adaptive operation as a continuation of the previous state. Therefore, it is possible to reduce the deterioration in sound quality of the subchannel when switching channels.

〔Example〕

FIG. 2 is a block diagram showing an embodiment of the present invention, in which (a) shows the transmitting side and (b) shows the receiving side. This example concretely shows the structure shown in FIG.
6 are all ADPCM systems. Control switching section 4, 7
, 14.17 are shown as interlocking two-way switches, but of course 5A, 17 can be realized with electronic switches.
6A, 15A. 16A is an adaptive control unit 5, 6, 15.16
It is an adaptive filter of 5B, 6B, 15
B and 16B are their adders. However, 5B, 6B
is used as a subtracter by reversing the input phase.

The power calculation/comparison unit 3 calculates the power (signal power) of each channel using the following equation for each frame of a certain period of time.

Pa(n) = (1/L), : , (P(n − i
+1) * exp(-s * (+-1))...
...(1) Here, the power ratio G (n) is defined as follows.

G (n) = Par (n) / Pa 1 (n)
・” ・” (2) This G (n)
is used as the main channel switching information, and the control switching unit 4.
At step 7, the right main channel adaptive control section 5 and the left main channel adaptive control section 6 are switched.

The right main channel adaptive control section 5 has G (n)≧1.0
, that is, it operates when the right channel is determined to be the main channel, and the left main channel adaptive control section 6 operates when G(n)<
1.0, that is, it operates when the left channel is determined to be the main channel. In the illustrated state, one adaptive control section 5 is operating, and the other adaptive control section 6 is at rest, holding the previous filter coefficient.

The same applies to the receiving side.

Assuming that the main channel signal vector at the j-th sample is Xj, the coefficient of the adaptive filter 5A is Hj, and the sub-channel signal vector is Yj, the residual signal ZJ is expressed by the following equation.

ZJ=Yj-HJ'・Xj ・・・・・・
(3) Here Hj' is the vector transpose of coefficient Hj.

Further, the coefficient Hj is successively modified by the following equation based on the output of the adder 5B.

Hj. +=Hj+ΔHj ・・・・・・
(4) Adaptive rules for equation (4) include a learning identification method and a correlation removal method.

The main channel signal and residual signal obtained as described above pass through the control switching section 7 and are encoded by the ADPCM encoding section 8.9. Then, main channel switching information and ADP
The output codes of the CM encoders 8 and 9 are multiplexed by a multiplexer 10 and transmitted.

In the receiving section, the main channel signal, the residual signal, and the main channel switching information are separated in the separating section l1, and the main channel signal and the residual signal are each separated by the ADPCM decoding section 12.
13 to be decrypted. Furthermore, the control switching unit l4 selects the right main channel adaptive control unit 15 and the left main channel adaptive control unit l6 based on the main channel switching information.

In the adaptive control unit 15, an adder 15B calculates a sub-channel decoded signal Yj from the adaptive filter 15A output (pseudo signal) of the main channel decoded signal Xj and the residual decoded signal Zj. Main channel decoded signal X obtained in this way
j and the sub-channel decoded signal YJ are outputted from the speaker 18.19 through the control switching section l7.

During the above-described operation, the adaptive control units 6, 16 stop operating and maintain the coefficients of the filters 6A, 16A as before. Then, when the control switching sections 4, 7, 14, 17 are switched to the side opposite to that shown in the figure, the adaptive control sections 5, 15 become inactive, and the adaptive control section 6.16 starts operating instead.

This is a case where the power calculation/comparison unit 3 determines that the left channel is the main channel.

By using two independent systems of adaptive control units 5 and 6 in which each filter coefficient is optimized in this way, a stable residual signal can be obtained immediately by simply selecting the adaptive control unit according to the determination result of the main and subchannels. can be obtained. Therefore, even if, for example, the speaking positions of multiple speakers distributed in an asymmetrical room change frequently, the two systems of adaptive control units maintain the spatial characteristics from different angles and improve the sound quality when switching channels. The speech of each speaker can be encoded in two-channel stereo without any deterioration.

〔Effect of the invention〕

As described above, according to the present invention, since the adaptive control diagram when the right channel is the main channel and the adaptive control section when the left channel is the main channel are independently provided, each control section has a certain degree of adaptation. The coefficients of the adaptive filter can be retained. Therefore, it is possible to reduce the deterioration in reproduction sound quality of the sub-channel that occurs when switching between the main and sub-channels.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the principle of the present invention, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is a block diagram of a conventional stereo audio encoding device. +m+ transmission unit In Figure 1, 1.2 is a sound collection device, 3 is a power calculation/comparison unit, 4.7 is a control switching unit, 5 and 6 are adaptive control units, and 8.9 is an encoding unit.

Claims (1)

[Claims] 1. The one with the higher power of the audio signals obtained by the left and right sound collectors is set as the main channel, and the other one is set as the sub-channel, and a pseudo signal is calculated from the signal of the main channel using an adaptive filter. A stereo audio encoding device that encodes and transmits a residual signal obtained by inputting a pseudo signal and a signal of the sub-channel to an adder, and a signal of the main channel, comprising: the adaptive filter (A); calculating and comparing the powers of the audio signals obtained by two sets of independent adaptive control units (5, 6) including the adder (B) and the left and right sound collecting devices (1, 2); The power calculation/comparison unit (3) produces an output indicating whether the power is left or right, and the right adaptive control unit (5) follows the output of the comparison unit (3).
Or a control switching unit (
4, 7).