JPS6251844A - Stereo sound transmission system - Google Patents

Abstract

PURPOSE:To transmit stereo sounds of high quality even if a transmission line having a low transmission speed is used, by transmitting additional information which synthesizes the sound signal of one channel with sound signals of the other channels. CONSTITUTION:In the transmission side, sounds of the right channel are transmitted as they are, and a transfer function G(omega) or an impulse response H(k) is estimated in an estimating circuit 22 in accordance with sound signals of right and left channels. In this case, a part of the impulse response is extracted by a part extracting circuit 23 and is transmitted to raise the transmission speed. In the reception side, the received sound signal of the right channel is reproduced as it is and the sound signal of the left channel is synthesized and reproduced in a synthesizing circuit 24 in accordance with the sound signal of the right channel and an approximate transfer function G'(omega) or an approximate impulse response H'(k).

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an audio transmission system, and particularly to a stereo audio signal transmission system.

[Technical Background of the Invention and Problems Therewith] In recent years, with the advancement of communication technology, the demand for remote conference systems that conduct conferences between remote locations by communication has increased.

Since teleconferencing systems generally exchange image information such as television images, electronic whiteboards, and faxes, and audio information between remote locations, it is desired to reduce the cost of transmitting this information. In particular, 64k that can be transmitted on current general subscriber lines
If this information can be transmitted at a transmission speed of about bps,
A remote conference system can be realized at a much lower cost than a high-quality remote conference system using optical fiber or the like. This is the l5DN (
When the comprehensive service digital network (Comprehensive Service Digital Network) is completed, it is thought to hold the key to the spread of remote conferencing systems, including to medium-sized and small-sized businesses and even ordinary households.

In a remote conference system using a transmission path of such a low transmission speed, for example, 54 kbps, it is necessary to compress a large amount of information such as images and audio to the extent that it does not interfere with the operation of the conference.

FIG. 8 is a configuration diagram of a conventional teleconferencing system, which includes a microphone 1 and a speaker 2 as an audio man-machine interface, a television camera 3 as an image man-machine interface, a television monitor 4, an electronic whiteboard 5, and a fax machine. 6. Telewriting device 7 and audio data of 16k
an audio unit 8 that encodes and decodes the image signals specified by the control unit 9; It is comprised of an image unit 11 that encodes and decodes at 48 kbps, and a transmission unit 13 that transmits and receives these audio and image signals via a 64 kbps transmission line 12.

In a remote conference system that uses such a low transmission speed transmission path, even monaural audio can be transmitted using, for example, ADPC
It is necessary to compress the audio data to a low transmission capacity such as 16 kbps through audio data compression processing such as M, and therefore stereo audio is not normally used.

On the other hand, it is well known that it is desirable to employ stereo audio in remote conference systems in order to create a sense of realism and to identify the speaker to know who is speaking.

Now, when attempting to introduce stereo audio, which has these advantages, into a remote conference system that uses a low transmission speed transmission line as described above, according to the conventional stereo audio transmission method, as shown in Figure 9, A stereo audio transmission section is required. In this stereo audio transmission section, audio information is transmitted in two channels, left and right, and requires twice as much transmission capacity as when monaural.

Therefore, with conventional methods, low transmission rates, e.g. 64kb
Methods for introducing stereo audio into a remote conference system that uses a PS transmission path include (a) a method of compressing the amount of information transmitted in one channel of audio to 8 kbps; and (b) a method of compressing the amount of information transmitted of images from 48 kbps to 32 kbps.
One possible method is to reduce the number to s.

However, the problem of (a) is that the quality of the audio deteriorates, and the problem of (b) is that the quality of the image or the service deteriorates.

As described above, with the conventional stereo audio transmission method, it is very difficult to introduce stereo audio into a remote conference system that uses a transmission path with a low transmission speed.

[Purpose of the Invention] The present invention has been made based on the above circumstances, and provides a stereo audio transmission method that enables transmission of high-quality stereo audio at low cost even when used on a transmission line with a low transmission speed. This is what we provide.

[Summary of the Invention] The present invention provides a stereo audio transmission system that transmits audio signals of multiple channels, in which a transmitting side receives main information consisting of an audio signal of at least one channel among the audio signals of multiple channels, and the main information.

The additional information necessary to synthesize the audio signals of the remaining channels is encoded and transmitted from the information, and the receiving side reproduces the audio signals of each channel from which the main information was transmitted by decoding, and By synthesizing and reproducing the channel from the main information and the additional information, it is possible to transmit high-quality stereo audio at low cost even when using a transmission line with a low transmission speed.

[Embodiments of the Invention] Hereinafter, the present invention will be described using a remote conference system as an example with reference to the drawings.

Note that in the following description, only one transmission direction will be shown to simplify the description. In addition, in a normal conference, multiple speakers in one conference room (A1-A4, Bl-84 in Figure 9)
) speaking at the same time, and even if multiple speakers in one conference room speak at the same time, there is no need to transmit these sounds in stereo because only one speaker is speaking. is considered to be lower than that of For this reason, in the following embodiments, only a case will be described in which stereo audio is transmitted for individual speeches that make up most of the conference.

First, a qualitative explanation of the principle of the present invention will be briefly given.

First, the voice ×(ω) of a single utterance is input to the microphones of the left and right channels. However, echoes from walls etc. are ignored, and ω is defined as each frequency.

At this time, the transfer function for each left and right channel is GR(ω)
, GL (ω), the input audio YL (ω) and YR (ω) of each left and right channel are YL ((c)) = GL
((c)) X (ω) ”” (1) YI! (ω
)=G* (ω)
(ω) ・・・・・・(3)=G(ω)Yg(ω)
(4), and as long as the transfer function G(ω) is known, the left channel sound can be reproduced from the right channel sound.

Based on this analysis, the present invention proposes that in the case of stereo transmission, even if the audio of both channels is not sent independently, if the audio of one channel and the transfer function are sent, the receiving side can transmit the audio of both channels. Audio can be reproduced and stereo transmission can be achieved. Moreover, since the transfer function in this case is defined by mere delay and attenuation, the amount of information is
This is much less than the information amount of (ω), and stereo transmission is possible with a smaller amount of transmission.

FIG. 1 is a schematic diagram of a stereo audio transmission system implementing a transmission method according to a first embodiment of the present invention.

This stereo audio transmission system consists of left and right channel microphones 1β, 1r, speakers 2β, 2r, and a transfer function G(ω
), and a synthesis circuit 21 that synthesizes the left channel audio signal YL (ω) from the transfer function G(ω) and the right channel audio signal YI2(ω).

In the figure, the voice X uttered by speaker A1 as described above
(ω) is the audio signal YR (
ω), the left channel microphone 1λ receives an audio signal Y
It is transmitted as L (ω). On the transmitting side, the right channel audio signal YR(ω) is transmitted as is, and the left channel audio signal YL(ω) is transmitted as the right channel audio signal YR.
(ω) is input to the estimation circuit 20, and the estimation circuit 20
From G(ω)=YL(ω)/Yt(ω), the transfer function G
After (ω) is estimated, this G(ω) is transmitted.

On the receiving side, the right channel audio signal YR (ω) is reproduced as is, and the left channel audio signal YL (ω)
is the transfer function G(ω) and the right channel audio signal YR(
ω) is input to the synthesis circuit 21, and in the synthesis circuit 21, YL (ω)=G(ω)Y$! (ω) Synthesized and reproduced.

At this time, from equations (3) and (4), G(ω)
=GL(ω)/GR(ω)...(5) Here, GR(ω)
), GL(ω) are determined by the voice characteristics in the room and the position of the speaker, and are not affected by the voice X(ω).

Therefore, from equation (5), it is considered that the transfer function G(ω) is steady as long as the speaker does not change, and that the duration of this steady state is mostly several QQ msec or more.

On the other hand, the audio X(ω) is unsteady, and from equations (1) and (2), the left and right channel audio signals YL(ω) and YR(ω
) are both unsteady. Therefore, unless the transfer function G(ω) is very complex, that is, the reverberation time in the room is long, the amount of information in the transfer function G(ω) is greater than YL (ω).
This embodiment, which transmits the transfer function G(ω), is more advantageous than the conventional stereo audio transmission system which transmits YL(ω).

FIG. 2 is a schematic configuration diagram of a stereo audio transmission system implementing a transmission method according to a second embodiment of the present invention. This stereo audio transmission system has one microphone for the left and right channels.
β, 1r, speakers 2β, 2r, and transfer function G(ω)
or the estimation circuit 22 that estimates the impulse response 1-((k), and the transfer function G(ω) or the impulse response H(k
), a partial extraction circuit 23 extracts the approximate transfer function G(ω) or the approximate impulse response 1-1(k), and
Audio signal YL of the left channel from (ω) or H(k)
(ω).

However, let the impulse response at the sampling point of the transfer function G(ω) be l-1(k), and H(k) = (h-M
(k), between h-++(k).

......, hN(k))', Wang indicates the transposed matrix, Mho, N>01ho (k)
is the center tap.

In this embodiment, as in the previous embodiment, the right channel audio signal is transmitted as is on the transmitting side, and the transfer function G(ω) or impulse response H(k) is determined by the estimation circuit 22 from the left and right channel audio signals. Presumed.

It is assumed that the impulse response H(k) estimated at this time has a shape that continues for 11000 to +2000 samples as shown in FIG. 3(A), and for the sake of simplicity,
It is assumed that the speaker continues to speak words with an average length of 500 m5ec.

At this time, if each sample value is quantized with 8 bits and transmitted, the transmission rate required to transmit this impulse response is 8 x 3000 x 2 = 48 kbps, which is a very large value.

Therefore, as shown in FIG. 3(B), a part of the impulse response, for example, -20 to +80 samples, is extracted by the partial extraction circuit 23 and transmitted.

In this case, the transmission speed is 1.6kbps, which is 15kbps.
This method is more advantageous than pS ADPCM.

Roughly speaking, considering that the three-dimensional effect of stereo is determined by the phase and delay of the sound, the position and size of the main tap, which is the largest of all the taps, is determined as shown in Figure 3 (C). Even if extracted and transmitted, the sense of realism is somewhat inferior, but if the size is set to 8 bits and the position is set to 8 bits, the speed becomes 32 bps, making it possible to significantly reduce the transmission capacity.

On the receiving side, the transmitted right channel audio signal is reproduced as is, and the left channel audio signal is reproduced in the synthesis circuit 24 from the right channel audio signal and the approximate transfer function G(ω) or the approximate impulse response H(k). Synthesize and reproduce audio signals.

As mentioned in the previous embodiment, when the transfer function G(ω) is simple, it is more advantageous to transmit G(ω) than to transmit the left channel audio signal @YL (ω), but G( ω
) is very complex, it becomes disadvantageous to transmit G(ω). Therefore, in this example, the transfer function G(
By transmitting approximation information of the transfer function G(ω) or impulse response H(k) instead of ω), the amount of transmitted information is reduced.

Figure 4 shows a transmitting side circuit (A) and a receiving side circuit (B) that implement the transmission method of this embodiment by transmitting approximate impulse responses.
) is a block configuration diagram of.

Transmitting side circuit (A> is microphone 1λ, 1r, amplifier 25
, 2.25r, A/D converters 26Jl, 26r, adaptive filter 27, and delay circuit 28f! , 28r, the subtracter 29, the approximation circuit 30, the threshold detection circuit 31,
Level detection circuits 32a and 32b and level ratio detection circuit 3
3, an ADPCM circuit 34, and a transmission circuit 35.

Further, the receiving side circuit (B) includes a separation circuit 36 and an ADPCM
circuit 37, D/A converters 38β, 38r, and amplifier 3
9β and 39r, speakers 2p and 2r, a coefficient generation circuit 40, and a filter 41.

The audio signals of each left and right channel were input from microphones 1λ and 1r, and were amplified to an appropriate level by amplifier 25J2.25r.
The /D converters 26 and 26r convert the signals into digital values XL(k) and XR(k) at the sampling time, respectively.

After that, the audio signal xi (k) of the right channel is 25
The audio signal XL (k) of the left channel is inputted to a 6-tap adaptive filter, and the left channel audio signal XL (k) is delayed by d samples in a delay circuit 28 to adjust the center tap position in the adaptive filter, and then sent to a subtracter as XL (k-d). 29. The subtracter 29 outputs the adaptive filter output XL (k)
is subtracted from Xt(k-d) to generate an error signal e(k). The adaptive filter uses the power of this error signal as an evaluation function, and adjusts the tap coefficients to minimize the error power using, for example, a well-known method such as the theoretical identification method.

Therefore, if learning of the adaptive filter progresses, XL (k
-d)=Xt (k), and the left channel audio signal can be synthesized from the right channel audio signal.

The approximation circuit 30 compresses the information amount of the tap coefficients of the adaptive filter obtained through the above processing in accordance with the transmission speed. Various methods for this approximation can be considered as shown in FIG. 3, but in this embodiment, the method shown in FIG.
Find the tap coefficient of the main tap, for example hloo(k), from k) to h256(k), and calculate h(k) to h1 according to the transmission speed l kbps of this embodiment.
The 23 tap coefficients before and after 11(k) are quantized to 8 bits, and the 5th tap is quantized to 8 bits, along with the main tap position encoded in 8 bits and an 8 bit header indicating the beginning of a series of information. As shown in the figure, it is framed and sent to the transmission circuit 35 as additional information.

Note that the approximation of the tap coefficients by the adaptive filter only needs to be performed when the speaker changes, and the threshold value detection circuit 31 detects the error e.
When the level of (k) exceeds the threshold and then falls below the threshold, it can be determined that the speaker has changed.

This will be roughly explained in detail with reference to FIG.

In the same figure, the speaker is moving from A to B at a time of 400 m5ec.
The transfer function changes from GA(ω) to Ge(ω
), the impulse response changes from HA(k) to He(k). At this time, the adaptive filter learns to follow changes in the impulse response and the tap coefficients change, so XL(k-d)≠XL(k) temporarily holds, and the level of error e(k) increases. Therefore e(k)
The additional information may be updated at the time when the level increases, then decreases, and becomes stable, which is 60Q m5ec in this figure.

In the transmission side circuit (A>) shown in Figure 4, the learning time of the adaptive filter is set to 200 m5ec or less, and the
Additional information is updated after 00m5ec.

Therefore, the delay circuit 28r causes the main information audio signal XR to
(k) is also delayed by 2 Q □m5ec and synchronized with the additional information.

Furthermore, since this additional information includes only a part of the tap coefficients of the adaptive filter, if it is transmitted as is, the level of the left channel voice signal XL (k) synthesized on the receiving side will be low. To prevent this, the level detection circuits 32a and 32b and the level ratio detection circuit 33 detect the level ratio of XR(k) and XL(k), and the approximation circuit 30 corrects the approximate tap coefficients. ) level is appropriate. Thereafter, it is sent to the transmission circuit 35 as a frame at a transmission rate of 1 kbps.

The transmission circuit 35 transmits this additional information and the right channel audio signal obtained from the delay circuit 28r to the ADPCM circuit 3.
4 and the main information converted into an ADPCM code with a transmission rate of 15 kbps to form 16 kbps stereo audio information, which is transmitted to the receiving side circuit (B) via the transmission path.

This stereo audio information sent to the receiving circuit (B) is separated by the separation circuit 36 into a 15 kbps right channel audio signal of main information and 1 kbps additional information. The right channel audio signal is decoded by the ADPCM circuit 37, then analog-converted by the D/A converter 38r, amplified by the amplifier 39r, and output from the speaker 2r.

On the other hand, the additional information is converted into 256 tap coefficients by the coefficient generation circuit 40 and provided to the filter 41. Next, the filter 41 synthesizes a left channel audio signal from the given tap coefficients and the input right channel audio signal. The synthesized left channel audio signal is sent to the D/A converter 381 in the same way as the right channel audio signal.
．． The signal is converted into an analog signal and output via an amplifier 39β and a speaker 2β.

As described above, this embodiment can be easily implemented using current technology, and can significantly compress the amount of information transmitted in a stereo signal in a teleconferencing system.

Note that in this embodiment, the adaptive filter 27 and the filter 41
Although transversal type filters are used for time domain processing, similar effects can be expected even if these are replaced with filters that process in the frequency domain.

FIG. 7 is a schematic configuration diagram of a stereo audio transmission system implementing a transmission method according to a third embodiment of the present invention. This stereo audio transmission system includes a microphone 1,911r for each left and right channel, speakers 2λ, 2r, and a transfer function G.
(ω) or impulse response H(k); a partial extraction circuit 23 that extracts approximate transfer function G(ω) or approximate impulse response H(k); and approximate transfer functions collected in advance. G(ω) or approximate impulse response H(k) tables 42a, 42b, encoding circuit 43, code decoding circuit 44, approximate transfer function G(ω) or approximate impulse response 'H'' (,k)
and a synthesis circuit 24 that synthesizes a left channel audio signal from a right channel audio signal.

In this embodiment, the transmitting side uses a table 42a of approximate transfer functions G(ω) or approximate impulse responses H(k) collected in advance and approximate transfer functions G(ω) or approximate impulse responses H output from the partial extraction circuit 23. A pattern comparison with (k) is performed by the encoding circuit 43, and the code q of the closest pattern is transmitted. And on the receiving side,
Approximate transfer function G(ω
) or an approximate impulse response H(k) is created by the code decoding circuit 44, and the synthesis circuit 24 synthesizes this with the left channel audio signal from the right channel audio signal. As described above, according to this embodiment, if the acoustic characteristics of the conference room are known, it is possible to maintain a sense of realism while reducing the amount of transmitted information.

In the above embodiments, a two-channel stereo audio transmission system has been described, but the present invention can also be applied to a three-channel or more stereo audio transmission system.

Further, the stereo audio transmission system of the present invention is not limited to remote conference systems, but can also be applied to other stereo audio transmission systems.

[Effects of the Invention] As explained in detail above, according to the present invention, main information consisting of an audio signal and additional information required to reproduce the audio signals of the remaining channels from this main information in stereo audio transmission. As a result, stereo audio can be transmitted without significantly increasing the amount of information compared to monaural audio transmission, and high-quality stereo can be transmitted at low cost even on low-speed transmission lines. Audio transmission becomes possible.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram of a stereo audio transmission system that implements the transmission method of the first embodiment of the present invention, and FIG. 2 shows a stereo audio transmission system that implements the transmission method of the second embodiment of the invention. Schematic configuration diagram, Figure 3 is a graph of impulse response, Figure 4 is a block configuration diagram of a circuit implementing the same example, Figure 5 is a frame configuration diagram in the same example, and Figure 6 is a time diagram in the same example. Chart diagram, FIG. 7 is a schematic configuration diagram of a stereo audio transmission system implementing the transmission method of the third embodiment of the present invention, FIG. 8 is a configuration diagram of a conventional remote conference system, and FIG. 9 is a diagram of a conventional stereo audio transmission system. FIG. 1 is a schematic diagram of an audio transmission system. 20.22... Estimation circuit 21.24... Synthesis circuit 23... Partial extraction circuit 43...
...... Encoding circuit 44 ......
...Code decoding circuit applicant: Toshiba Corporation Patent attorney Sa Suyama - Figure 1 Left ← Scroll down Figure 2 (A) (B) Seiwo Tatsua CC') Figure 3 Figure 4 Figure 5 Figure 6 Figure 7

Claims (3)

[Claims] (1) In a stereo audio transmission system that transmits audio signals of multiple channels, from the transmitting side to the receiving side, main information consisting of the audio signal of at least one channel among the audio signals of the multiple channels, and the remaining channels based on this main information The additional information necessary to synthesize the audio signals of the main information is encoded and transmitted, and the receiving side reproduces the audio signal of the channel to which the main information was transmitted by decoding, and the audio signals of the remaining channels are encoded and transmitted. A stereo audio transmission system characterized in that main information and the additional information are synthesized and reproduced. (2) The stereo audio transmission system according to claim 1, wherein the additional information is a transfer function, an impulse response, or approximate information thereof. (3) The stereo audio transmission system according to claim 1, wherein the additional information is identification information that identifies data that has been collected and accumulated in advance about a transfer function, an impulse response, or their approximation information.