JPS6384213A - Streo coding and decoding device - Google Patents

Abstract

PURPOSE:To obtain a pseudo stereo coder/decoder with less quantized noise by coding only a channel signal with large signal power and in regard to a channel signal with small signal power analyzing a predicting coefficient and sending the result, and correcting the spectrum of the decoded value at the reception side. CONSTITUTION:The sender side is provided with a power difference/time difference calculator 1 calculating the power difference and time difference of a left channel input signal XL and a right channel input signal XR and a switch 2 using the signal large power as a signal X1 and a signal having small power as a signal X2. The signal X1 is coded by a coding section 3 as an output I, a predicting coefficient A2 of the signal X2 is obtained by an analyzer 4, auxiliary information is arranged and outputted by a multiplexing circuit 5. Each information is separated by a separating circuit 6 from the reception signal at the reception side, a decoding section 7 outputs a monaural decoding signal Y1 and a monaural signal Y2 is reproduced by predicting coefficients A1, A2. A power difference/time difference detector 10 sending a control signal to a power correcting device 11 and a time difference corrector 12 is used to reproduce a pseudo stereo signal from the signal Y1, Y2.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention is directed to an apparatus for highly efficiently encoding and decoding two-channel stereo signals at a transmission rate equivalent to one channel.

[Conventional technology]

As a technology related to stereo encoding, there is a conventional literature such as [Rinji, [High-efficiency encoding of high-quality acoustic signals using adaptive prediction]
, Theory of Faith (A), vol.

J67-A, No. 4. 1984],
A first method of removing correlation by orthogonally transforming two-channel signals, and obtaining a stereo output signal by orthogonally inversely transforming a signal obtained by independently encoding and decoding the two-channel signals after the orthogonal transform, and the literature [ Minami, [Sound system of teleconverter 7T lens system using pseudo-stereo sound] IEICE Technical Report CAS8
6-29), extracts the power difference and time difference between two channel signals, sends the power difference and time difference as auxiliary information to a signal that encodes the sum signal of the two channel signals, and uses this auxiliary information to Two methods are known: the tlS2 method, which obtains a two-channel pseudo-stereo signal from the signal.

[Problem that the invention attempts to solve]

Among the conventional stereo encoding techniques mentioned above, in the former tlSl method, the stereo output signal is not pseudo, so the degree of separation between the left and right channel signals is high;
Since the encoded bits are distributed by 1!49 to the two-channel encoder, quality deterioration due to quantization noise becomes a problem.

On the other hand, in the second method, the sum signal of two channel signals is obtained and encoded in one channel, so the degree of separation of left and right channel signals is somewhat inferior, but the quantization noise, which was a drawback of the first method, is reduced. This solves the problem of quality deterioration due to In such a +Q-like stereo encoding system, it is important to faithfully reproduce the sound signal, and the power difference and time difference are considered to be the main factors for sound image localization, but on the other hand, the difference in frequency characteristics of the two channel signals It is also known that the sound image localization affects the localization position (Yamakoshi et al., [On changes in sound image localization due to the left and right f time difference, J
In fact, when a sound source is close to one of the microphones in a conference room or the like, a difference in frequency characteristics appears between the left and right channels because the sound reverberation characteristics are symmetrical. Therefore, the method of reproducing the mu stereo signal using only the power difference and time difference factors, such as the second method, has the drawback that the localized position of the sound image cannot be faithfully reproduced.

In order to solve these conventional drawbacks, the present invention provides pseudo-stereo encoding and decoding that can encode two-channel stereo signals with high quality using the transmission speed of one channel, and can localize sound images faithfully. The purpose is to provide equipment.

[Means for solving problems]

According to the invention, the above-mentioned objects are achieved by the means specified in the claims.

That is, the present invention encodes only the channel signal with higher signal power of the two channel stereo signals on the transmitting side, analyzes the prediction coefficients of the channel signal with lower signal power, and sends it out. and means for determining the channel signal with the smaller signal power by modifying the shape of the spectrum of the decoded value of the channel signal with the larger signal power using the prediction coefficient of the decoded signal and the received prediction coefficient. The present invention differs from conventional techniques in that it incorporates a function to correct differences in frequency characteristics into the pseudo-stereo encoding method.

〔Example〕

FIG. 1 is a block diagram showing one embodiment of the present invention.

In FIG. 1, there is a power difference/time difference calculation lit on the transmitting side that calculates the power difference and time difference between the left channel input signal XL and the right channel input signal XH, and the signal with the larger signal power between XL and XR is The signal with smaller signal power is
2, an encoding unit 3 that encodes the output signal X1 of the switch to obtain a coded output I, an analyzer 4 that obtains a prediction coefficient A2 of the output signal X2 of the switch, The signal (2) includes a multiplexing circuit 5 that arranges and outputs a signal S representing the power difference/time difference and auxiliary information of the signal C indicating which channel of the signal has been encoded.

The power difference/time difference calculator 1 calculates the cross-correlation function r(k) within 17 frames consisting of N signals shown in equation (1), and calculates the ratio of σR2 and σL2 shown in equation (3). The power difference and the value of k that maximizes r (k) are the time difference, and this power difference and time difference are output as a signal S, and σR2 and σL
' is compared, and information representing a channel with a large signal power is output as C.

At the knee, ・・・・・・・・・・・・・・・・・・(3) N: 1
Number of samples L in 7 frames: The subscript switch 2 that represents the Bi1% column of samples is
Based on information C representing a channel with high signal power,
The one with the larger signal power of the left channel input XL and the right input XR is switched to Xl, and the other is switched to X2.

This switching is performed at the beginning of encoding the 17-frame signal. The encoding unit 3 performs high-efficiency encoding on the monofull input signal X1 and outputs an encoded signal I. Selecting and encoding one channel out of two channel signals in this way reduces the number of encoded bits by one, compared to the case where two encoders are prepared and each of the two channel signals is encoded independently.
This is because high-quality encoding can be performed with less quantization noise since all the signals are allocated to each encoder.

The analyzer 4 performs linear predictive analysis on the monaural input signal X2 to obtain a prediction coefficient A2. The prediction order is 1, and the prediction coefficient to be output is A 2 = (azt r a2z+...
If v a2111, then the spectral envelope H2(Z) of the input signal X2 is H2(Z) =□ ・・・・・・・・・・・・・・・
(4) The coefficient A2 is determined by approximating the frequency response of the all-pole digital filter of Lm+. In other words, if the coefficients of the all-pole dinotal filter are set based on the prediction coefficient A2 obtained by the analyzer 4, the input signal X2 can almost be reproduced. In this sense, it can be said that the prediction coefficient aA2 represents spectrum information of the input signal X2.

On the receiving side, from the received signal, encoded signal processing, prediction coefficient A2,
A separation circuit 6 that separates encoded channel information C1 and power difference/time difference information S, a decoding unit 7 that decodes the encoded signal signal and outputs a monaural decoded signal Y1, and a prediction coefficient A1 obtained in decoding 5S7. a synthesizer 8 that reproduces a monaural signal Y2 using the output Y1 and the prediction coefficient A2, and a signal C
a switch 9 that switches the signals Y1 and Y2 to the right channel and the left channel based on the signal S; a power difference corrector 11 that extracts the power difference and time difference from the signal S and reproduces a pseudo stereo signal from the monaural signals Y1 and Y2; The power difference/time difference detector 10 sends a control signal to the time difference corrector 12.

FIG. 2 is a block diagram showing an example of the configuration of the synthesizer 8. As shown in FIG.

This synthesizer includes an all-zero dinotal filter consisting of a delay device 13, a variable coefficient water multiplier 14, an addition 1315, and a subtraction@16, a delay device 17, a variable coefficient multiplier 18, and an adder 19.
．． It is composed of 20 all-pole dinotal filters. The decoded output Y1 of the decoder 7 is sent to the delay device 13 and the subtracter 16.
is input. Zhu W, the variable coefficient of the device 14 is a decoder 7
The linear prediction coefficient A1 obtained by is used.

The prediction coefficient of the decoding unit is n s Al =l az ra
1□,...e azn l , the total Hl of this synthesizer (Z) = 1-Σra, i Z-'
・・・・・・・・・・・・・・・・・・(5) Represented by Lm+.

Here, r is a positive constant smaller than 1 (for example, 0°7 to 0.8
This has the effect of slightly lowering the filter Q to prevent abnormal noise from occurring in the reproduced output.

Since the filter expressed by equation (5) approximates the inverse characteristic of the spectral envelope of the decoded signal Yl, the spectrum of the output Y of the all-zero digital filter is whitened.

The prediction coefficient A2 of the signal X2 sent from the transmitting side is used as the variable coefficient of the multiplier 18 in the all-pole digital filter.

The transfer function H2 (
Z) is represented by 1.

The role of "in equation (6) is the role of ", which was mentioned in the explanation of equation (5).
The role is similar to that of

Since the filter expressed by equation (6) is sent from the transmitting side, it approximates the spectral envelope of the signal X2, so the whitened signal Y is shaped into the shape of the spectral envelope of the signal X2 and is output. The signal becomes Y2.

As described above, the signals Y1 and Y2 are two-channel signals in which the spectral envelopes of the input signals X1 and X2 are preserved, respectively, and the switch 9 selects the signal Y1 to the channel with the larger signal power based on the auxiliary information C. However, the signal Y2 is supplied to the channel with low signal power.

Next, in order to obtain pseudo-stereo output signals YL and YR, appropriate correction is performed by a power difference corrector 11 and a time difference corrector 12 using the power difference and time difference detected by the power difference/time difference detector 10 from the auxiliary information S. will be carried out.

In this embodiment, assuming that predictive coding is performed in the encoding section and the decoding section, the polar prediction coefficient A1 in the decoding section is sent to the synthesizer 8.
However, even if predictive coding is not performed in the encoding section and decoding section, if the analyzer 4 used on the transmitting side is added to the receiving side and the signal Y1 is analyzed, the polar predictive coefficient can be obtained. A
1 can be obtained.

〔Effect of the invention〕

As explained above, in the present invention, the transmitting side includes a circuit for determining the power difference and time difference between two channel stereo signals, a circuit for encoding the channel signal with higher signal power, and a circuit for encoding the channel signal with higher signal power. It is equipped with a circuit for decoding the encoded signal of one channel in the receiver (1[) and a circuit for decoding the encoded signal of one channel using the prediction coefficient.
The configuration includes a circuit that shapes the spectrum of a channel signal and synthesizes other channel signals, and a circuit that corrects the power difference and time difference between left and right channel signals, allowing high-quality encoding with little quantization noise. It is possible to realize a pseudo-stereo encoder/decoder that has accurate sound image localization characteristics.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention, wS2
The figure is a block diagram showing an example of the configuration of a synthesizer.

Claims (1)

[Claims] A detector that detects the power difference and time difference between two channel stereo signals, an encoder that predictively encodes the channel signal with higher signal power, and an analyzer that calculates the prediction coefficient of the channel signal with lower signal power. and a signal representing the power difference/time difference found by the detector, a prediction coefficient found by the analyzer, and a signal representing which channel the signal was encoded as auxiliary information for the encoder. a multiplexing circuit that multiplexes the output signal with the output signal and outputs it to the digital output terminal; a separation circuit that separates the encoder output signal and auxiliary information from the signal at the digital input terminal; a decoder that decodes the encoded signal; and a decoding circuit. A synthesizer that regenerates the signals of other channels by shaping the spectrum of the decoder output signal using the prediction coefficients obtained for the signal and the prediction coefficients obtained from the auxiliary information; A circuit that switches the decoder output and the synthesizer output to left and right channels using a signal indicating whether the signal is converted into a A stereo encoder/decoder comprising a corrector that performs appropriate correction.