JP2774518B2 - Stereo audio transmission system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention] (Industrial Application Field) The present invention relates to a stereo signal transmission system applied to, for example, a teleconferencing system.

(Prior Art) The present inventor has disclosed a transmission system capable of transmitting high quality stereo audio at low cost in Japanese Patent Application Laid-Open No. 61-82840. In summarizing the present invention, only the main audio signal and the residual signal are transmitted in the transmission of stereo audio, equivalent audio estimation is performed on the receiving side and the transmitting side, and the residual signal is converted to the estimated audio signal on the receiving side. The addition is performed.

Further, according to the present invention, a single utterance is encoded in high-quality stereo, and even when a plurality of utterances are used, the level of the residual signal of the sub-channel increases, but the audio level is increased by the adaptive quantizer. Since the encoding can be performed following, the deterioration of the inter-channel correlation component removal amount due to the increase of the quantization noise can be compensated. Therefore, the problem that the sound image is disturbed even at the time of plural remarks does not occur.

However, in this method, the sub-channel for encoding the residual signal from which the estimated signal output from the inter-channel predictor is removed and the main channel for encoding the input signal of the inter-channel predictor are fixed to the left and right microphones, respectively. For example, when a person close to the microphone on the sub-channel side speaks, a phenomenon occurs in which the power on the sub-channel side to which less quantization bits of about 2 bits are allocated is larger than the power on the main channel side. .

In the method of assigning the quantization step size at the time of adaptation according to the power of the input signal such as ADPCM, it is more advantageous to assign a larger bit length to the higher power, so in the above case, the inter-channel correlation component is removed. However, since the power on the sub-channel side is originally large, there is a problem that quantization noise increases.

(Problems to be Solved by the Invention) As described above, the stereo audio transmission system disclosed by the present inventor can transmit high-quality stereo audio at low cost, but the main channel and the sub-channel have left and right microphones respectively. , A phenomenon occurs in which the power on the sub-channel side is higher than the power on the main channel side, and there is a problem that quantization noise increases.

The present invention has been made based on such circumstances,
It is an object of the present invention to provide a system capable of transmitting high-quality stereo signals with reduced quantization noise.

[Configuration of the Invention] (Means for Solving the Problems) The present invention relates to a stereo signal transmission system having an encoding unit and a decoding unit for encoding and decoding a plurality of input signals input from a plurality of input devices. In the above, the encoding unit separates a signal having a high power among the plurality of input signals from a first signal, and a unit that distinguishes other input signals from a second signal, and encodes the first signal. Means for generating a main channel code, first signal generation means for predicting a second signal from the first signal to generate a first prediction signal, and means for generating the first signal from the second signal. Means for generating a prediction residual code by encoding a prediction residual signal obtained by subtracting the prediction signal of (i), wherein the decoding unit decodes the main channel code to generate a main channel decoded signal. Means and the generated master Second generating a second prediction signal by predicting a signal of the sub channel from the tunnel decoded signal
Means for generating a sub-channel decoded signal by adding the second prediction signal to a prediction residual decoded signal which is a signal obtained by decoding the prediction residual code; and Means for adaptively switching the speaker outputs of the main channel and the sub-channel based on a comparison between the power of the decoded signal and the power of the sub-channel decoded signal.

According to a second aspect of the present invention, in the stereo signal transmission system according to the first aspect, the first prediction signal generation unit includes:
Means for predicting the second signal by performing adaptive filter processing on the first signal, and generating a first prediction signal expressed by including a first prediction coefficient of the adaptive filter. And a unit for erasing the first prediction coefficient of the adaptive filter when the power of the plurality of input signals is reversed, wherein the second prediction signal generation unit outputs the main channel decoded signal to the main channel decoded signal. Means for predicting the signal of the sub-channel by performing adaptive filter processing on the signal and generating a second prediction signal expressed by including a second prediction coefficient of the adaptive filter; Means for erasing the second prediction coefficient of the adaptive filter when the power of the signal and the power of the sub-channel decoded signal are reversed.

(Operation) In the present invention, the side that encodes an input signal encodes at least one input signal of the plurality of input signals according to each level of the plurality of input signals, allocates the input signal to one channel, and The remaining input signal among the input signals is encoded and assigned to the other channel, and the side that decodes the encoded input signal outputs at least one of the input signals according to each level of the plurality of input signals. Since the input signal is allocated to at least one speaker and the remaining input signals of the input signals are allocated to the remaining speakers, quantization noise is reduced and a high-quality stereo signal can be transmitted.

(Example) Hereinafter, an example of the present invention will be described.

FIG. 1 is a diagram for explaining the principle of a stereo sound transmission system according to the present invention.

As shown in the figure, on the transmitting side of this system, switches 10001 and 1002 are interposed between left and right microphones 1 _R and 1 _L and an encoding unit (not shown). And the main channel 1
By monitoring the power of the input sound of 003 and the sub-channel 1004 by the power comparator 1005, when the power of the sub-channel 1004 exceeds the power of the main channel 1003, the left and right microphones 1 _R and 1 The correspondence between the _L output and the main and sub-channels 1003 and 1004 is switched.

Also on the receiving side, the power of the main and sub-channels 1003 and 1004 after decoding by the decoding unit (not shown) is monitored by the power comparator 1006. When the decoded speech subchannel 1004 is above the main channel 1003, a main switch 1007 switches the corresponding subchannel 1003, 1004 and left and right speakers 2 _R, 2 _L.

As described above, in the present invention, the control is performed such that the one with the maximum power is always the input of the main channel among the voices of the plurality of input channels.

In order to perform more accurate switching, switching in encoding may be performed by detecting the power of this code output when the encoder has a built-in multiplexer such as ADPCM. it can.

 Next, specific examples of the present invention will be described.

FIG. 2 is a diagram showing a configuration of a stereo sound transmission system according to this embodiment.

As shown in the figure, a transmitting unit 100 of this stereo sound transmission system includes one of left and right microphones 1 _R and 1 _R.
The main channel ADPCM encoding unit 30 that ADPCM encodes the audio input from _L and assigns it to the main channel, and converts the audio input from one of the left and right microphones 1 _R and 1 _L into ADPCM.
A sub-channel ADPCM encoding unit 31 that encodes and assigns to the sub-channel; an inter-channel correlation removal unit 34 that predicts sub-channel audio based on the main channel audio and removes inter-channel correlation from the sub-channel audio; A level detector 300 detects the level of the channel sound and controls switching between the outputs of the left and right microphones 1 _R and 1 _L and the correspondence between the main and sub channels.

The ADPCM signals D (k) and F (k) output from the transmission unit 100 are input to the reception unit 200 via the transmission paths 25 and 26.

On the other hand, the receiving unit 200 of this stereo transmission system decodes the main channel ADPCM code D (k) to reproduce the main channel audio x ₁ (t).
2. Decoding sub-channel ADPCM code F (k) to reproduce left channel sound y ₁ (t) Sub-channel ADPCM decoding 33. Predict sub-channel sound based on main channel sound and remove inter-channel correlation A channel correlation adding unit 35 for adding the inter-channel correlation again to the decoded ADPCM decoded voice,
And a level detector 400 for detecting the levels of the main channel audio and the sub channel audio, and controlling the switching of the correspondence between the main and sub channels and the left and right speakers 2 _R and 2 _L.

 Next, each part will be described in detail.

Main channel ADPCM encoder 30 The main channel ADPCM encoder 30 includes an A / D converter 101, a subtractor 103, an adaptive quantizer 105, an adaptive inverse quantizer 107, and an adder 1.
09, a predictor 111, and a switch 181.

The main channel sound x (k) input from one of the left and right microphones 1 _R and 1 _L and digitized and selected by the switch 181 is subtracted from the main channel predicted sound (k) predicted by the predictor 111, and the power is calculated by prediction. Is x
The prediction residual signal d (k), which is smaller than (k), is encoded into an ADPCM code of, for example, about 32 kbps by an adaptive quantizer 105 that adaptively changes the quantization step according to the amplitude.

Further, in the predictor 111, the reproduced prediction residual signal d ₁ (k) decoded by the adaptive inverse quantizer 107 and the output (k) of the predictor are added by the adder 109, so that the main channel of the receiving unit is added. A
The same main channel playback audio x ₁ (k) as the DPCM decoding unit 32
, And performs adaptive filter processing to minimize the power of the reproduction prediction residual signal d ₁ (k).

Sub-channel ADPCM encoder 31 The sub-channel ADPCM encoder 31 includes an A / D converter 113, delay units 115 and 180, a subtractor 117, an adaptive quantizer 119, an adaptive inverse quantizer 121, an adder 123, It comprises a predictor 127 and a switch 182.

A channel that is input from one of the left and right microphones 1 _R and 1 _L , digitized, delayed by a delay circuit 115 or a delay circuit 180, and whose correlation has been removed by the inter-channel correlation remover 34 from the sub-channel audio selected by the switch 182. ADPCM encoding processing similar to the main channel is performed on the inter-correlation removal signal e (k).

The above-described delay guarantees the causality of the main and sub channel audios x (k) and y (k) in the inter-channel correlation removing unit 34 (when the direct sound reaches the left microphone earlier than the right microphone). However, the input to the inter-channel correlation elimination unit 34 must be such that the left voice is always delayed.

Here, the auto-correlation component of the voice is removed by the predictor 127, and the correlation component with the main channel voice mixed in the sub-channel voice is removed by the inter-channel correlation removing unit 34, so that less information than the main channel is obtained. Amount of ADPCM code, for example, about 16 kbps.

Inter-channel correlation remover 34 The inter-channel correlation remover 34 is an inter-channel predictor 12
9, consisting of subtractor 131, main channel playback audio x ₁ (k)
And an inter-channel correlation component (k) is generated by the inter-channel predictor 129. The inter-channel predictor 129 performs adaptive filter processing, and adaptively changes the characteristics of the filter so as to minimize the power of the sub-channel prediction residual signal e ₁ (k) in the receiving unit 200.

Level detection unit 300 The level detection unit 300 includes the level detectors 161, 164 and the comparator 16
2, 163, a hangover processing circuit 165, an OR circuit 166, a multiplier 167, and an adder 168.

In the level detector 300, the main channel reproduced sound x ₁ (k) of the adder 109 is input, and the inter-channel predictor
The sub-channel reproduced signal y ₁ (k) obtained by adding the inter-channel correlation component (k) of 129 and the sub-channel prediction residual signal e ₁ (k) of the adder 123 by the adder 168 is input.

The levels of these input signals are detected by level detectors 161 and 164, and are input to comparators 162 and 163.

The comparator 162 detects whether the level of the main channel reproduced sound x ₁ (k) is equal to or lower than a predetermined threshold level TH. When the level is equal to or lower than the threshold level TH, the prediction output from the OR circuit 166 is output. The transmitting unit 100 and the receiving unit 200 simultaneously clear the prediction coefficients (specifically, the tap coefficients of the adaptive filter) of the inter-channel predictors 129 and 151 by the coefficient clear signal HCL.
And a synchronization section between the receiving section 200 and the receiving section 200.

On the other hand, the comparator 163 compares the levels of the reproduced signals of the main channel and the sub-channel, and the level of the reproduced signal of the sub-channel is determined by the multiplier 167 from the level of the reproduced signal of the main channel (in this embodiment, the coefficient is 0.5 Therefore, when it exceeds 6 dB), a switching signal RLFLAG for switching the correspondence between the left and right microphones 1 _R and 1 _L and the main and sub channels is generated. Note that the hangover processing circuit 165 performs hangover processing to prevent frequent switching,
When the switching occurs, the next switching is prohibited for the time T. Also at the time of this switching, the adaptive filter is cleared by the OR circuit 166, and the deterioration of the temporary inter-channel correlation removal performance due to the change of the speaker is reduced.

Switching between the left and right microphones 1 _R and 1 _L and the correspondence between the main and sub-channels is performed by the switches 181 and 182 described above.
Performed under the control of AG.

Main channel ADPCM decoding unit 32 The main channel ADPCM decoding unit 32 is an adaptive inverse quantizer 133,
Adder 135, predictor 137, delay units 139, 184, D / A converter 14
1, consisting of a switch 183.

The received ADPCM code D (k) is converted into a main channel reproduction prediction residual signal d ₁ (k) by the adaptive inverse quantizer 133,
The main channel prediction signal (k) output from the predictor 137 is added by the adder 135, and the main channel reproduction signal x
₁ (k).

Thereafter, in order to compensate for the delay added to the sub-channel by the transmission unit 100 by the delay unit 139, the same amount of delay is added to the sub-channel reproduced sound, and then the D / A converter
141 is outputted from the speaker 2 _R via the speaker 2 _R or D / A converter 149 via the.

The predictor 137 receives the main channel reproduced voice x ₁ (k) and performs an adaptive filter process for generating a main channel predicted signal (k). Learning of this filter is performed by the main channel reproduced predicted residual signal d _1. This is performed by minimizing the power of (k).

This predictor 137 is the same as the predictor 111 of the main channel ADPCM encoding unit 30, and furthermore, since the same signal is input, the transmitting unit 100 and the receiving unit 200 can output the same predicted signal x (k).

The delay unit 184 calculates the difference between the left and right delays newly generated by the delay processing of the delay unit 180 inserted in the main channel in order to give the causality of the inter-channel prediction filter on the transmission unit 100 side when the sub-channel becomes the main channel. This is for canceling, and the same delay as that of the sub-channel is inserted.

Sub-channel ADPCM decoding unit 33 The sub-channel ADPCM decoding unit 33 is an adaptive inverse quantizer 143,
It comprises an adder 145, a predictor 147, a D / A converter 149, and a switch 185.

Similarly to the main channel, a reception-side sub-channel prediction residual signal e (k) is generated from the received ADPCM code F (k).

Thereafter, this signal is added with the correlation component (k) from the inter-channel correlation adding section 35, and the sub-channel reproduced sound y ₁
(K) becomes, through switch 185, through the D / A converter 141 speaker 2 _R or from the D / A converter 149 through the speaker 2 _R
Output from

Inter-channel correlation adding section 35 Inter-channel correlation adding section 35 is an inter-channel predictor 15
1, consisting of adder 153, main channel playback audio x ₁ (k)
And the inter-channel correlation component (k) is generated by the inter-channel predictor 151.

This inter-channel predictor 151 is the same as the inter-channel predictor 129 of the transmission unit 100, and is an adaptive filter that learns to minimize the power of the reception-side sub-channel prediction residual signal e ₁ (k) similarly to the transmission unit 100. And the same predicted value (k) as the transmitting unit 100
Is obtained.

Level detector 400 The level detector 400 is composed of the level detectors 171 and 174 and the comparator 17
2, 173, a hangover processing circuit 175, an OR circuit 176, and a multiplier 177.

The level detecting section 400 has the same function as the level detecting section 300 described above. That is, the switching signal RLFLAG and the prediction coefficient clear signal HCL are changed by the level detection unit 400 at the same timing as the transmission unit 100 based on the levels of the reproduction signals of the main and sub channels detected by the level detectors 171 and 174. The inter prediction unit 151 and switches 183 and 185 for switching the correspondence between the main and sub channels and the left and right speakers 2 _R and 2 _L are provided. The hangover processing is also performed by the hangover processing circuit 175 as in the transmission unit 1000.

The clearing process of the adaptive filter based on the level of the main channel can be omitted by adopting a leak process (a process of gently forgetting past coefficients) for tap coefficients of the adaptive filter.

As described above, according to the present embodiment, since the main and sub-channels can be optimally assigned based on the output level of the microphone according to the position of the speaker, that is, even if the speaker is biased left and right, the bit length is always set. Since a signal having a large power is supplied to a quantizer having a long length, quantization noise is reduced, and the speech quality of the conventional stereo speech coding system is greatly improved.

[Effect of the Invention] As described above, according to the present invention, quantization noise is reduced and a high-quality stereo signal can be provided.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the principle of the present invention, and FIG. 2 is a configuration diagram of a stereo sound transmission system according to an embodiment of the present invention. 30 ... Main channel ADPCM encoder 31 ... Sub channel ADPCM encoder 32 ... Main channel ADPCM decoder 33 ... Sub channel ADPCM decoder 34 ... Inter-channel correlation remover 35 ... Channel correlation Additional unit 300, 400… Level detector

Claims (2)

(57) [Claims] 1. A stereo signal transmission system having an encoding unit and a decoding unit for encoding and decoding a plurality of input signals input from a plurality of input devices, wherein the encoding unit comprises a plurality of input signals. Means for classifying a high-power signal from the first signal and other input signals from the second signal; means for encoding the first signal to generate a main channel code; A first prediction signal generating means for generating a first prediction signal by predicting a second signal from the second signal; and a prediction residual signal obtained by subtracting the first prediction signal from the second signal. Means for encoding to generate a prediction residual code, wherein the decoding unit decodes the main channel code to generate a main channel decoded signal, and from the generated main channel decoded signal. Sub-channel signal And a second prediction signal generating means for generating a second prediction signal by predicting the second prediction signal by adding the second prediction signal to a prediction residual decoded signal which is a signal obtained by decoding the prediction residual code. Means for generating a sub-channel decoded signal, and means for adaptively switching the main channel and the sub-channel speaker output based on a comparison between the power of the main channel decoded signal and the power of the sub-channel decoded signal. A stereo signal transmission system, characterized in that: 2. The first predictive signal generating means predicts the second signal by performing adaptive filter processing on the first signal, and generates a first predictive coefficient of the adaptive filter. Means for generating a first prediction signal expressed by including: and means for erasing a first prediction coefficient of the adaptive filter when power levels of the plurality of input signals are reversed, The second prediction signal generation unit predicts the signal of the sub-channel by performing adaptive filtering on the decoded signal of the main channel,
Means for generating a second prediction signal expressed by including a second prediction coefficient of the adaptive filter; and when the power of the main channel decoded signal and the power of the sub channel decoded signal are reversed, 2. The stereo signal transmission system according to claim 1, further comprising: means for eliminating a second prediction coefficient of the adaptive filter.