JPH11243396A - Audio coding transmission system - Google Patents

Abstract

(57) [Problem] It is difficult to connect a silence compression transmission network to an existing transmission network that does not perform silence compression without making any improvements to the existing speech communication system. SOLUTION: A transmitting end 10 for encoding a voiced section of an original audio signal and outputting a first audio code generated by silence compression to a first transmission line A, and a first transmission line A A relay node 20 that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to the second transmission path B based on the first speech code received from the second transmission path B; And a receiving end 30 that decodes the second audio code received from the transmission path B to reproduce the original audio signal, the internal state when the transmitting end 10 performs the encoding process, and the receiving end 30 performs the decoding process. It is characterized by matching the internal state when performing.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice coded transmission system for compressing and transmitting a voice signal with high efficiency, and more particularly to a voice coded transmission system capable of improving voice quality.

[0002]

2. Description of the Related Art Conventionally, digital transmission technology of voice signals using a high-efficiency voice coding technology for highly efficient coding excluding redundant components of voice signals has been used to reduce communication costs such as intra-company communication and international communication. It is actively used in fields where reduction is strongly desired. As an efficient speech coding method, ITU (Inte
rnational Telecommunications Union) Recommendation G.726 ADPC
A predictive differential coding scheme typified by the M (Adaptive Differential Pulse Code Modulation) coding scheme, and ITU Recommendation G.728 LD-CELP (Low-De
lay Code Excited Linear Prediction (low-delay code-excited linear prediction) coding method (see Fig. 21), and ITU recommendations
G.729CS-ACELP (Conjugate-Structure Algebraic-Code-E
xcited Linear Prediction: There is a coding method based on “Analysis-by-Synthesis” as represented by a conjugate structure algebraic code excitation linear prediction) coding method.

[0003] For details of these encoding methods, refer to the following documents.・ CCITT Recommendation G.726, 40,32,24,16kbit / s
adaptive differentialpulse code modulation (ADPCM)
■ ・ ITU-T Recommendation G.728, ■ Coding of speech a
t 16kbit / s using Low-Delay Code Excited Linear Pre
diction (LD-CELP) ■ ・ ITU-T Recommendation G.729, ■ Coding of speech a
t 8kbit / s using conjugate-structure algebraic-code
-excited linear prediction ■

Here, the prediction difference encoding method is a method of predicting a current signal from a past signal sequence, quantizing a difference signal between the prediction signal and an actual signal, and transmitting the quantized signal. Also, the encoding method based on “analysis by synthesis”
Using parameters obtained by analysis based on past and current audio signals, synthesize signals of a plurality of patterns as candidates, select a signal pattern closest to the input voice from among them, and select This is a method in which the parameters that are the basis of the synthesized signal are quantized as optimal parameters and transmitted.

[0005] Each of these systems is realized by an encoder provided on the transmitting side and a decoder provided on the receiving side. As for the parameters of the encoder obtained by so-called forward adaptation based on the current input speech signal component, there is no means for obtaining the information by the decoder. Therefore, in order for the encoder and the decoder to hold the same parameters, it is necessary to transmit information from the encoder to the decoder. On the other hand, for parameters obtained by so-called backward adaptation based on past speech signals, both the encoder and the decoder have common parameter calculation means, so that the encoder and the decoder can operate normally. .

That is, in the predictive difference coding method, the past signal sequence is shared as the internal states of the encoder and the decoder, and the encoding and decoding operations of speech based on this are realized. Also, in the coding method based on “analysis by synthesis”, the same set of parameters is shared as the internal state of the encoder and decoder in transmission and reception obtained by backward adaptation, so that speech based on these parameters is Encoding and decoding are realized. Therefore, it is premised that the decoder performs speech decoding processing while maintaining the same internal state as the encoder.
Conversely, if the past audio signals are different between the transmitting side and the receiving side for some reason, the internal states will not match, and the accuracy of the audio reproduced by the decoding unit may not be guaranteed.

On the other hand, in a communication network, there is an increasing demand for multimedia transmission such as transmission of images and computer data in addition to voice represented by a telephone. Asynchronous transfer mode (ATM: Asynchronous Transfer Mode)
In recent years, a communication system called sfer mode) has been actively introduced. In an ATM transmission network, an information signal (voice, image, etc.) to be transmitted is digitally encoded, the obtained bit string is divided into fixed-length blocks called cells, and asynchronously transmitted to a communication path. As a result, information signals having different transmission speeds can be handled in multiple ways.

Recently, due to the demand for higher efficiency as described above, a silent compression technique (a method of discarding a silent part of an audio signal and transmitting the same, taking advantage of the characteristics of the asynchronous and variable speed of the ATM network). ) Is being used together.
This silence compression technique can reduce the total amount of audio signals transmitted to a transmission path without deteriorating the audio quality of a sound section, and enables more efficient audio transmission by the statistical multiplexing effect.

However, in this silence-compressed audio transmission system, since there is no audio information transmitted during silence, the operation of the decoder for receiving and decoding the audio code, which is a highly efficient encoded audio signal, is silent. Sometimes undefined. Therefore, the backward adaptation described above does not work properly,
When transiting from a silence state (the talk spurt may be referred to as “absent” state) to a sound state (a talk spurt may be referred to as the “presence” state), an encoder that generates a speech code The internal state does not match the internal state of the decoder that decodes the audio code. Therefore, even if a correct high-efficiency code having no transmission path error is given, the decoding unit cannot always correctly decode a speech signal. This phenomenon often appears as an unpleasant abnormal sound in the reproduced sound at the receiving end, such as a click sound or an oscillating sound.

FIG. 22 is a block diagram showing a configuration of a first speech coded transmission system for solving this problem.
This figure is based on the configuration shown in Japanese Patent Application Laid-Open No. 2-181552. In this audio transmission system, a transmitting end 60 and a receiving end 70 form a pair. In the state with the talk spurt, that is, in the sound period, the transmitting end 60 encodes the audio signal by the high-efficiency audio encoder 601 and sends out the audio code to the transmission line A via the switch 604. Since the changeover switch 604 of the transmitting end 60 is switched so that there is no talk spurt, that is, nothing is transmitted to the transmission path A in the silent period, the voice code compressed with silence is transmitted from the transmitting end 60. Become. The voice detector 603 detects the presence or absence of voice / non-voice of the voice signal, and switches the changeover switch 604.

On the other hand, at the receiving end 70, the speech code from the transmission path A is decoded into a speech signal by the decoder 702 and output. During silence compression, the changeover switch 705 is switched to the pseudo background noise generation unit 706, and a pseudo signal imitating the background noise of the speaker (hereinafter, referred to as pseudo background noise signal) is received from the receiving end 70. ) Is output. The presence / absence information extraction unit 707 detects a presence / absence of voice based on the voice code, and switches the changeover switch 705.

In this system, an encoder 60 is provided at a transmitting end 60.
1 has a memory 630 storing a predetermined internal state, and the receiving end 70 has a memory 730 storing the same contents.
have. Then, at the time of transition from a silent section to a sound section of an audio signal in which the above-described problem occurs,
It is output to the voice detector 603 and the presence / absence information extraction unit 70
7 are detected synchronously, and the memory 63 is
From 0, the initial value of the internal state for the encoding process is set in the encoder 601, and at the receiving end 70, the initial value of the internal state for the decoding process is set in the decoder 702 from the memory 730.

As described above, the timing at which the talk spurt is detected at the transmitting end 60 and the receiving end 70 is synchronized, and at that time, the internal states of both are reset to the same state. For this reason, the internal states of the encoder 601 and the decoder 702 always match in the voiced section of the voice, and it is possible to avoid occurrence of abnormal noise at the head of the talk spurt.

By using such a technique, it is possible to realize silent compression transmission using high-efficiency coding. However, at present, transmission networks that have been constructed so far and do not perform silence compression already exist. These transmission networks are often built at high cost as infrastructure, and it is economically difficult to immediately replace or upgrade to a silent compression transmission network. Therefore, when it is desired to construct a large network that covers the range covered by these conventional transmission networks, for the time being, a network that does not perform silence compression must be used as it is and coexist with a network that performs silence compression.

That is, coexistence of a network that does not perform silence compression and a network that performs silence compression is realized by connecting these two types of networks with a relay node. As the method, FIG.
And two methods shown in FIG. These figures illustrate transmission from a network that performs silence compression to a network that does not perform silence compression.

FIG. 23 is a configuration diagram of a transmission system in which two types of networks are connected in tandem via a relay node. In this figure, components having the same functions as those in FIG. 22 are denoted by the same reference numerals, and description thereof will be omitted. The encoder 601 of the transmitting end 60 of this system performs silence compression, and then sends out a speech code to the transmission line A.

The relay node 80 receives the voice code (only the sound part) of the transmitting end 60 from the transmission path A, and decodes it into a voice signal by the decoder 802. Since the signal at the transmitting end 60 is silently compressed, the audio signal to be decoded is only a sound part. The pseudo background noise signal is inserted into a silent section of the audio signal, encoded by the encoder 801 again, and transmitted to the receiving end 70.

Here, the processing of the audio signal between the transmitting end 60 and the relay node 80 is performed, for example, by the silent compression transmission system using the synchronous reset described with reference to FIG. As described above, since the signal is once decoded by the relay node 80 and then re-encoded, from the viewpoint of the voice encoding process, a so-called tandem connection in which two mutually independent transmission paths A and B are connected in series. It is a form called.

FIG. 24 is a block diagram of a transmission system in which two types of networks are connected by a digital one link via a relay node. In this figure, components having the same functions as those in FIG. 23 are denoted by the same reference numerals, and description thereof will be omitted.

The silence-compressed voice code transmitted from the transmission end 60 to the transmission path A is supplemented with a silence code by the relay node 80 and transmitted to the reception end 70 via the transmission path B. The audio signal input to the transmitting end 60 is encoded by an encoder 601.
Is encoded with high efficiency. The sound detector 603 detects sound / no sound (the presence or absence of a talk spurt) based on the sound signal, and controls the changeover switch 604. Changeover switch 6
04 is an encoder 601 only when there is a talk spurt.
Is transmitted to the transmission path A. If there is no talk spurt, the speech code is discarded, and nothing is output to the transmission path A. As a result, a speech code that has been subjected to silence compression is transmitted to the transmission path A.

[0021] The relay node 80 receives the silence-compressed voice code transmitted from the transmitting end 60 via the transmission path A. The presence / absence information extraction unit 807 constantly monitors the reception status of the voice code from the transmission path A, and controls the changeover switch 804 according to the result. That is, when a voice code is input from the transmission path A, the switch 804 is moved to the 804b side, and the received voice code is relayed to the transmission path B as it is.

On the other hand, when the reception of the signal from the transmission line A is not detected (that is, in the section determined to be a silent section),
The silence information extraction unit 807 switches the changeover switch 804 to the pseudo background noise generation unit 806 side, and supplements the signal in the silence section with the pseudo background noise signal. Note that the delay unit 209 sets the transmission path A for the processing time of the presence / non-speech information
, And the changeover switch 804 is operated in synchronization with the input of the voice code.

At the receiving end 70, the decoder 702
, And extracts the audio signal. Since a pseudo signal is inserted into the signal in the silent section at the relay node 80, it looks as if a continuous audio signal is being input to the receiving end 70.

As described above, the relay node 80 merely executes only switching depending on the presence or absence of voice. For this reason, the speech code transmitted to the receiving end 70 is a silence section in which interpolation processing is performed at the relay node, but a voiced section serving as main information is nothing but the one transmitted from the transmitting end 60. And transmission lines A and B
Since the signal passing between them is relayed to the other network without being processed as a digital signal, this inter-network transmission system is generally called a digital one link.

[0025]

However, the above-described tandem connection and the connection of the transmission lines A and B by one digital link have the following problems. That is,
In the tandem connection, the speech code compressed at the transmitting end 60 is converted into a speech signal at the relay node 80, and then a noise signal is inserted. Thereafter, the noise signal is re-encoded and transmitted to the transmission path B. Therefore, the internal state of the encoder 801 of the relay node 80 always coincides with the internal state of the decoder 702 of the receiving end 70, and the above-described generation of abnormal noise is prevented.

However, since the relay node 80 performs a process of decoding and re-encoding the voice code, the voice signal input to the transmitting end is output twice from the receiving end until output from the receiving end. It will undergo encoding / decoding processing. Therefore, there is a problem that quantization errors are accumulated and the quality of the audio signal output from the receiving end 70 is deteriorated. This degradation in voice quality is due to the high bit rate (l6kbit
/ s or more), it is almost unnoticeable, but the higher the compression ratio, the more pronounced the deterioration tendency.
In particular, voice transmission systems have a low bit rate,
This deterioration in voice quality cannot be ignored.

On the other hand, the situation is completely opposite in the connection by one digital link. In this case, since the voice code in the sound part transmitted to the receiving end 70 is the same as the voice code generated at the transmitting end 60, the quality deterioration of the voice signal due to accumulation of the quantization error is prevented. But,
In general, the internal state of the encoder 601 at the transmitting end 60 and the internal state of the decoder 702 at the receiving end 70 do not coincide with each other at the timing of transition from the silent state to the voiced state.

That is, although the speech code itself is the same, the internal state is different in the encoding / decoding processing,
The audio signal decoded by the decoder 702 is different from the signal intended by the encoder 601, and there is a problem that the above-described abnormal sound may occur. The occurrence of the abnormal noise not only gives the recipient discomfort, but also usually occurs at the beginning of the talk spurt, so that there is a problem that the degree of understanding of the contents of the call is significantly reduced.

Due to the above problems, it has been difficult to connect a silence compression transmission network to an existing transmission network that does not perform silence compression without making any improvements to the transmission network.

The present invention solves such a problem, and enables high-quality voice transmission in a voice coded transmission system in which a transmission network that performs silence compression is connected to an existing transmission network that does not perform silence compression. It is an object of the present invention to provide a voice coded transmission system that performs

[0031]

According to a first aspect of the present invention, there is provided a speech coded transmission system which encodes a voiced section of an original speech signal and performs a first transmission of a first speech code generated by silence compression. A second speech code generated by interpolating background noise into a silent section of an original speech signal based on a transmission end to be output to a channel and a first speech code received from a first transmission channel, to a second speech code. In a speech coded transmission system including a relay node that outputs to a transmission path and a reception end that decodes a second speech code received from the second transmission path and reproduces an original speech signal, the transmission end has a background Transmitting side background noise generating means for generating a first pseudo signal imitating noise;
Transmitting side background noise inserting means for inserting a pseudo signal of the following, transmitting side encoding means for encoding a speech signal obtained by the transmitting side background noise inserting means, and speech code encoded by the transmitting side encoding means. A transmitting-side voice code generating means for compressing a silent section to generate a first voice code, wherein the relay node decodes the first voice code;
Relay-side background noise generation means for generating a second pseudo signal imitating background noise, and relay-side background noise insertion for inserting the second pseudo signal into a silent section of the audio signal decoded by the relay-side decoding means Means, a relay coding means for coding the voice signal obtained by the relay background noise insertion means, and a voice code coded by the relay coding means inserted into a silent section of the first voice code. And a relay-side voice code generation means for generating a second voice code.

According to a second aspect of the present invention, there is provided a transmission terminal for encoding a voiced section of an original voice signal and outputting a first voice code generated by silence-compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a reception end that decodes a second speech code received from a second transmission path and reproduces an original speech signal, the transmission end includes a first simulated background noise. Transmission-side background noise generation means for generating a pseudo-signal of the following, transmission-side background noise insertion means for inserting a first pseudo-signal in a silent section of the original audio signal, and speech signal obtained by the transmission-side background noise insertion means. A transmitting side encoding means for encoding, and a transmitting side code And a transmitting-side voice code generating means for compressing a silent section of the voice code encoded by the converting means to generate a first voice code, wherein the relay node decodes the first voice code. Decoding means; relay-side background noise generation means for generating a second pseudo signal imitating background noise;
A second encoding unit that encodes the second pseudo signal and a pseudo code that is encoded by the second encoding unit is inserted into a silent section of the first audio code to generate a second audio code; Relay-side speech code generation means, and the relay-side encoding means inputs an internal parameter of the relay-side decoding means, and holds an internal state equivalent to that of the relay-side decoding means using the internal parameters. It is characterized by being.

According to a third aspect of the present invention, there is provided a speech encoding transmission system which encodes a voiced section of an original speech signal and outputs a first speech code generated by silence compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a reception end that decodes a second speech code received from a second transmission path and reproduces an original speech signal, the transmission end includes a first simulated background noise. Transmitting side background noise generating means for generating a pseudo signal of the following, transmitting side background noise inserting means for inserting a first pseudo signal into a silent section of the original audio signal, and transmitting a sound signal obtained by the transmitting side background noise transmitting means. A transmitting side encoding means for encoding, and a transmitting side code And a transmission-side speech code generation means for compressing a silent section of the speech code encoded by the conversion means to generate a first speech code, wherein the relay node converts a second pseudo signal imitating background noise Relay side background noise generation means for generating, a first speech code, and an interpolation signal generation means for coding a second pseudo signal, and a pseudo code coded by the interpolation signal generation means in the first And a relay-side speech code generation means for generating a second speech code by inserting the speech signal into a silent section of the speech code, and the interpolation signal generation means holding an internal state common to the decoding process and the encoding process. It is characterized by doing.

According to a fourth aspect of the present invention, there is provided a speech encoding transmission system which encodes a voiced section of an original speech signal and outputs a first speech code generated by silence compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a reception end that decodes a second speech code received from a second transmission path and reproduces an original speech signal, the transmission end includes a first simulated background noise. Transmitting side background noise generating means for generating a pseudo code, transmitting side decoding means for decoding the first pseudo code, and converting the pseudo signal decoded by the transmitting side decoding means into a silent section of the original audio signal. Transmitting side background noise inserting means to be inserted,
A transmitting-side encoding unit that encodes the audio signal obtained by the transmitting-side background noise inserting unit; and a silence section of the audio code encoded by the transmitting-side encoding unit, thereby generating a first audio code. The relay node includes: a relay-side background noise generator that generates a second pseudo-code imitating background noise; and a second pseudo-code in a silent section of the first voice code. And a relay-side speech code generation means for generating a second speech code, wherein the transmission-side decoding means inputs the speech code coded by the transmission-side coding means, and Is characterized in that the speech code is decoded, and the second pseudo code is decoded in the silent section.

According to a fifth aspect of the present invention, there is provided a speech encoding transmission system which encodes a voiced section of an original speech signal and outputs a first speech code generated by silence compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a reception end that decodes a second speech code received from a second transmission path and reproduces an original speech signal, the transmission end includes a first simulated background noise. Transmitting side background noise generating means for generating a pseudo code, transmitting side decoding means for decoding a first pseudo code, transmitting side coding means for coding an original audio signal, and transmitting side coding means. By compressing the silent section of the encoded speech code, the first speech And a relay-side background noise generating means for generating a second pseudo-code imitating background noise, and a second side-noise section for generating a second pseudo-code in the silent section of the first voice code. And a relay-side speech code generation unit that generates a second speech code by inserting a pseudo code of
The transmitting-side encoding means inputs the internal parameters of the transmitting-side decoding means, and uses the internal parameters to hold an internal state equivalent to that of the transmitting-side decoding means.

According to a sixth aspect of the present invention, there is provided a speech coded transmission system which encodes a voiced section of an original speech signal, and outputs a first speech code generated by silence compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a reception end that decodes a second speech code received from a second transmission path and reproduces an original speech signal, the transmission end includes a first simulated background noise. Transmission-side background noise generation means for generating a pseudo-signal, transmission-side encoding means for encoding an original audio signal while maintaining an internal state based on the first pseudo-signal, and transmission-side encoding means The first voice code is compressed by compressing the silent section of the voice code A transmission-side speech code generation means for generating the pseudo-code, and a relay node for generating a second pseudo-code imitating background noise; and a second pseudo-code in a silent section of the first speech code. And a relay-side speech code generation means for inserting a code to generate a second speech code.

According to a seventh aspect of the present invention, there is provided a speech encoding transmission system which encodes a voiced section of an original speech signal and outputs a first speech code generated by silence-compression to a first transmission path. And relaying a second speech code generated by interpolating background noise in a silent section of the original speech signal based on the first speech code received from the first transmission path to the second transmission path. In a speech coded transmission system including a node and a receiving end that decodes a second speech code received from a second transmission path to reproduce an original speech signal, a transmission end encodes the original speech signal. A transmitting-side encoding unit, and a transmitting-side audio code generating unit that compresses a silent section of the audio code encoded by the transmitting-side encoding unit to generate a first audio code;
The relay node decodes the first audio code to generate an audio signal, a relay-side background noise generating unit that generates a pseudo signal simulating background noise, and a relay-side decoding unit that generates the audio signal. Relay-side background noise insertion means for inserting a pseudo signal in a silent section of the speech signal, relay-side encoding means for encoding the speech signal obtained by the relay-side background noise insertion means, and relay-side encoding means. The encoded speech code,
And a relay-side speech code generation unit that generates a second speech code by inserting the first speech code into a silent section and a transition section immediately after switching from silence to speech.

[0038] In claim 8, the relay node further includes a timer for counting the transition section, and the relay-side speech code generation means determines the end timing of the transition section based on the count value of the timer. Features.

In the ninth aspect, the relay node further includes comparison means for comparing the internal state of the relay side decoding means with the internal state of the relay side encoding means. The end timing of the transition section is determined based on the comparison result.

According to a tenth aspect of the present invention, the relay node includes: a timer for counting the transition section; a comparing means for comparing the internal state of the relay side decoding means with the internal state of the relay side encoding means; A determination unit that determines the end timing of the transition section based on the comparison result of the comparison unit; and the relay-side speech code generation unit determines the end timing of the transition section based on the determination result of the determination unit. It is characterized by the following.

In the eleventh aspect, the transmitting-side background noise generating means and the relay-side background noise generating means generate a digital signal simulating background noise in a silent section as a pseudo voice.

In the twelfth aspect, the transmitting-side background noise generating means and the relay-side background noise generating means are memories storing a predetermined digital signal.

According to a thirteenth aspect, the transmitting background noise generating means and the relay background noise generating means are random signal generators.

[0044] According to a fourteenth aspect of the present invention, there is further provided a synchronizing signal transmitting means for transmitting a synchronizing signal indicating the start of a silent section of the original audio signal from the transmitting end to the relay node. The background noise generating means and the relay-side background noise generating means are synchronized.

In the fifteenth aspect, the transmitting end further includes an identification pattern generating means for generating an identification pattern to be superimposed on a silent section of the original audio signal, and the relay node recognizes the identification pattern and generates the background noise on the transmission side. The means and the relay-side background noise generating means are synchronized.

The speech coded transmission system according to claim 16 is
A transmitting end that outputs a first speech code generated by performing a speech process on the voiced section of the original speech signal and performing silence compression on a first transmission path, and a first speech received from the first transmission path A relay node that outputs a second speech code generated by interpolating background noise into a silent section of the original speech signal to a second transmission path based on the code, and a second node received from the second transmission path. In a speech coded transmission system including a reception end that decodes a speech code to reproduce an original speech signal, a transmission end includes: a transmission-side parameter generation unit that generates a first parameter signal for generating pseudo background noise; A transmitting-side encoding unit that encodes the original audio signal while retaining an internal state based on the first parameter signal; and a silent section of the audio code encoded by the transmitting-side encoding unit, which is compressed by the first encoding unit. A transmitter-side speech code generator that generates speech codes Relay node generating means for generating a second parameter signal for generating pseudo background noise, relay side encoding means for generating a pseudo code based on the second parameter signal, And a relay-side speech code generation unit that generates a second speech code by inserting a pseudo code into a silent section of one speech code.

According to a seventeenth aspect, the transmission-side parameter generation means and the relay-side parameter generation means are memories storing predetermined digital signals.

[0048] In the eighteenth aspect, the transmission-side parameter generation means and the relay-side parameter generation means are random code generators.

In claim 19, the transmitting end further comprises a synchronizing signal transmitting means for transmitting a synchronizing signal indicating the start of a silent section of the original audio signal, and the relay node receives the synchronizing signal and generates a transmitting-side parameter. The means and the relay-side parameter generating means are synchronized.

In the twentieth aspect, the transmitting end further includes an identification pattern generating means for generating an identification pattern to be superimposed on a silent section of the original audio signal, and the relay node recognizes the identification pattern and generates a transmission-side parameter generation means. And the relay-side parameter generating means are synchronized.

[0051]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of a speech coded transmission system according to the present invention will be described below with reference to the accompanying drawings.

Embodiment 1 FIG. 1 is a block diagram showing a configuration of the speech coded transmission system according to Embodiment 1. In FIG. 1, reference numeral 10 denotes a transmitting end which encodes a sound section of an original audio signal and outputs a first audio code generated by silence compression to a transmission path A (first transmission path); A second speech code generated by interpolating a speech code corresponding to a silent section of the original speech signal based on the first speech code received from the path A is output to a transmission path B (second transmission path). The relay node 30 is a receiving end that decodes the second audio code received from the transmission path B and reproduces the original audio signal.

Reference numeral 101 denotes an encoder (transmitting-side coding means) for performing high-efficiency coding of the input speech signal based on a predetermined algorithm, and 103 denotes a control signal 1A for controlling the operation mode of the transmitting end 1. Voice detector that outputs
4 compresses the audio code encoded by the encoder 101,
A changeover switch (transmission-side speech code generation means) for generating a first speech code, and a changeover switch (transmission-side background noise insertion means) 105 for inserting a background noise signal (first pseudo signal) into a silent section of the original speech signal ) And 106 are pseudo background noise generation units (transmission side background noise generation means) for generating this background noise signal in a pseudo manner.

Further, reference numeral 201 denotes an encoder (relay-side encoding means) for performing high-efficiency encoding of the input audio signal based on a predetermined algorithm, and 202 decodes the first audio code sent from the transmitting end 10. A switch (relay-side decoding means) for inserting a background noise signal (second pseudo signal) into a silent section of the audio signal decoded by the decoder 202; ) And 206 are pseudo background noise generation units (relay side background noise generation means) for generating this background noise signal in a pseudo manner.

Further, reference numeral 207 denotes a presence / absence information extraction unit which constantly monitors the reception status of the voice code from the transmission path A and outputs a control signal in accordance with the result, and 209 denotes a delay unit,
Reference numeral 14 denotes a changeover switch (relay-side speech code generation means) that inserts the speech code encoded by the encoder 201 into a silent section of the first speech code to generate a second speech code. Is a switch for interpolating the silent section of the first speech code using the speech code encoded in step (1), and a decoder 302 for decoding the second speech code.

The pseudo background noise generators 106 and 206
Preferably, a digital signal simulating background noise in a silent section is generated as a pseudo sound.

Next, the operation of the voice transmission system will be described. The transmitting end 10 generates a speech code obtained by encoding the original speech signal with high efficiency. This speech code is transmitted to the transmission path A in a sound section and is not transmitted to the transmission path A in a silent section according to the determination result of the speech detector 103. Thus, the transmitting end 10 performs so-called silence compression.
The transmission path A is a transmission path for a speech code that has undergone silence compression. On the other hand, the transmission path B to which the receiving end 30 is connected is a transmission path through which audio codes that are not subjected to silent compression are transmitted. And
The relay node 20 connects these two transmission paths, receives the speech code from the transmission end 10 from the transmission path A, inserts a pseudo speech code in a silent section, and outputs the pseudo speech code to the transmission path B. The receiving end 30 decodes the voice code and outputs a voice signal.

At the transmitting end 10, the voice detector 103
Performs a distinction between a sound section and a silent section based on the input original audio signal, and outputs a control signal 1A for controlling the operation mode of the transmitting end 10 based on the result. If the sound detector 103 determines that the sound is present, the switch 105 is set to 105B and the switch 104 is set to 104.
Push to B side. As a result, the original audio signal input to the transmitting end 10 is input to the encoder 101, is subjected to high-efficiency encoding, and is output to the transmission line A via the switch 104.

On the other hand, if the sound detector 103 determines that there is "silence", the changeover switch 105 is set to 105A and the changeover switch 104 is set to 104A. Accordingly, the pseudo background noise signal output from pseudo background noise generation section 106 is input to encoder 101. However, the signal in which the pseudo background noise signal output from the encoder 101 is encoded is cut off by the changeover switch 104, so that nothing is output to the transmission path A.

Here, the pseudo background noise generator 106 will be described. At the timing when the determination by the voice detector 103 transitions from “voiced” to “silent”, the voice detector 103
Outputs an initialization signal 1B for controlling the initialization of the pseudo background noise generation unit 106. Pseudo background noise generator 10
6, the initialization signal 1B is input, and the pseudo background noise generator 106 resets the internal state at the input timing of the initialization signal 1B. This operation is essential to realize an operation synchronized with the pseudo background noise generation unit 206 in the relay node 20 described below.

In the relay node 20, the first end transmitted from the transmitting end 10 after being silently compressed and transmitted via the transmission line A
Receiving the voice code of. The existence / non-speech information extraction unit 207 includes:
The reception status of the first speech code from the transmission path A is constantly monitored, and the changeover switches 204, 205, and 2
14 to output a control signal 2A. That is,
When a speech code is input from the transmission line A, it is determined that the section is a sound section, and the changeover switch 204 is set to the 204B side, the changeover switch 205 is set to the 205B side, and the changeover switch 214 is set to 2
Fold each to 14B side.

The first speech code received by the relay node 20 is relayed to the transmission line B as it is via the delay unit 209 and the switch 214. The first speech code is also provided to the decoder 202 via the switch 205.
The decoder 202 performs a decoding process based on the input first audio code, and decodes the audio signal. The decoded audio signal is supplied to the encoder 201 via the changeover switch 204. The encoder 201 performs an encoding process based on the input audio signal.

Here, between the encoder 101 of the transmitting end 10 and the encoder 201 of the relay node 20, there are differences between the input signals, that is, the original audio signal and the decoded audio signal decoded by the decoder 202, respectively. However, the nature of the speech is very similar. Therefore, the internal state of the encoder 101 of the transmitting end 10 corresponding to the numerical value of the property of the audio signal and the internal state of the encoder 201 of the relay node 20 are as follows.
It is very similar.

On the other hand, when the presence / absence information extraction unit 207 does not detect the reception of the signal from the transmission line A, that is, in the section where “silence” is determined, the changeover switch 204 is switched to the 204A side. Switch 205 to 2
The changeover switch 214 is tilted to the 05A side and to the 214A side.

The output signal of the pseudo background noise generator 206 is
The signal is input to the encoder 201 via the switch 204.
The encoder 201 performs an encoding process based on the input pseudo background noise signal, and outputs a pseudo background noise code. The pseudo background noise code output from the encoder 201 is transmitted to the transmission path B via the switch 214.

By this series of operations, a signal in a silent section can be supplemented with a pseudo background noise signal. Here, the delay unit 209 delays the audio code received from the transmission path A and transmitted to the transmission path B by the time required for the processing in the presence / non-speech information extraction unit 207 and the processing in the encoder 201. By this delay operation, synchronization between the voice code of the voiced section output from the delay unit 209 and the pseudo background noise code output from the encoder 201 is realized.

Further, at the timing when the determination by the presence / absence information extraction section 207 shifts from “speech” to “silence”, the presence / absence information extraction section 207 controls the initialization of the pseudo background noise generation section 206. Output for the initialization signal 2B. The initialization signal 2B is input to the pseudo background noise generation unit 206, and the pseudo background noise generation unit 206 resets the internal state at the input timing of the initialization signal 2B.

The pseudo-background noise generation unit 206 realizes the same operation as the pseudo-background noise generation unit 106 of the transmitting end 10, and also performs the initialization operation described above. It is the same as the unit 106. Since the initialization operation is performed in synchronization with the presence / absence of a speech signal input from the transmission path A, and eventually with the determination signal of the speech detector 103 of the transmission end 10, the encoder 101 of the transmission end 10 and the relay node 2
The same pseudo background noise signal is supplied to the 0 encoder 201 at the same timing.

Further, in the sound interval, the internal state of the encoder 101 of the transmitting end 10 and the encoder 20 of the relay node 20 are changed.
Since the internal state substantially matches the internal state of 1, the pseudo background noise code output from these encoders 101 and 201 in the silent section can also take an extremely approximate value.

The pseudo background noise code output from the encoder 201 is transmitted to the decoder 2 via the switch 205.
02 is also supplied. The decoder 202 performs a decoding process based on the input pseudo background noise code. With this operation, the internal state of the encoder 101 of the transmitting end 10 and the internal state of the decoder 202 of the relay node 20 are always kept the same. For this reason, it is determined again as "voiced" and the decoder 20
Even if the voice code from the transmission path A is input to 2, the decoding of the voice can be normally realized because the internal states match in transmission and reception.

The receiving end 30 receives the second speech code which is continuously transmitted from the relay node 20 via the transmission path B and whose silent section is interpolated with the pseudo background noise code. The decoder 302 performs a decoding process based on the input second audio code, decodes the audio signal into an audio signal, and outputs the audio signal from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 101 of the transmitting end 10 during the voiced interval, and therefore, the encoder 101 of the transmitting end 10 is output during the voiced interval. And the internal state of the decoder 302 of the receiving end 30 match. In the silent period, the receiving end 30
Is the output of the encoder 201 of the relay node 20, the second speech code
01 and the internal state of the decoder 302 of the receiving end 30 match. As described above, even in the silent period, the internal state of the encoder 101 of the transmitting end 10 and the relay node 2
Since the internal state of the encoder 201 of 0 is matched,
As apparent from the receiving end 30, it is possible to obtain an effect as if the voice code not subjected to silence compression was directly received from the transmitting end 10.

As described above, the encoder 10 of the transmitting end 10
Since the internal state of the receiver 1 and the internal state of the decoder 302 of the receiving end 30 are always matched, it is possible to avoid the possibility that an unpleasant sound is decoded due to the mismatch of the internal states.

Embodiment 2 Next, a second embodiment will be described with reference to FIG. In the present embodiment, a relay node 21 is obtained by improving the relay node 20 in the speech coded transmission system of the first embodiment. The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

The relay node 21 receives the first speech code transmitted from the transmitting end 10 after being silently compressed and transmitted via the transmission path A. The presence / absence information extraction unit 207 constantly monitors the reception status of the voice code from the transmission line A, and controls the changeover switch 214 according to the result.
Is output.

That is, when a speech code is input from the transmission line A, it is determined that the section is a sound section, and the changeover switch 214 is
Push down to 14B side. The first speech code received by the relay node 21 is relayed to the transmission line B as it is via the switch 214. Also, the first speech code is simultaneously transmitted to the decoder 2
02 is also supplied. The decoder 202 performs a decoding process based on the input first speech code.

Here, in the encoder 101 of the transmitting end 10 and the decoder 202 of the relay node 21, the adaptive processing based on the same speech code is performed for both the sound portions, so that the transmitting end encoder 101 Internal state of the relay encoder 20
The internal state of 1 is kept at the same value.

Next, at the timing when the determination by the presence / absence information extraction section 207 shifts from “speech” to “silence”, the presence / absence information extraction section 207 controls the initialization of the pseudo background noise generation section 206. The initialization signal 2B is output. The pseudo background noise generator 206 resets the internal state at the timing of receiving the initialization signal 2B, and starts transmitting the pseudo background noise signal.

The pseudo-background noise generation section 206 realizes the same operation as the pseudo-background noise generation section 106 of the transmitting end 10, and also performs the initialization operation on the transmitting end 1
0 is the same as the pseudo background noise generation unit 106. Since the initialization operation is performed in synchronization with the presence / absence of an audio signal input from the transmission line A, and eventually with the determination signal of the audio detector 103 of the transmitting end 10, the encoder 101 of the transmitting end 10 and the relay node 2
The same pseudo background noise signal can be supplied to one encoder 201 at the same timing.

At the same time, the initialization signal 2B is also supplied to the encoder 201 and the decoder 202. The decoder 202 that has received the initialization signal 2B sends the internal parameters to the encoder 201 and stops operating. The encoder 201 receives internal parameters from the decoder 202 by the initialization signal 2B,
Based on this, the coding process of the pseudo background noise signal supplied from the pseudo background noise generation unit 206 is started.

Here, in the sound period, the internal state matched by the encoder 101 of the transmitting end 10 and the decoder 202 of the relay node 21 is inherited by the encoder 201 of the relay node 21 as it is. Although there is no direct exchange of information with each other, the pseudo background noise codes output from both the encoder 101 of the transmitting end 10 and the encoder 201 of the relay node 21 in the silent period match.

In the section where the presence / absence information extracting section 207 determines that the signal from the transmission path A is "silence", the switch 214 is moved to the side of 214A. Upon receiving the output signal of the pseudo background noise generation unit 206, the pseudo background noise code output from the encoder 201 is transmitted to the transmission path A via the switch 214. With this series of actions,
A signal in a silent section can be supplemented with a pseudo background noise signal.

Lastly, when the presence / absence information extraction unit 207 determines that the presence / absence information transitions from “silence” to “sound”, the presence / absence state is determined.
The silence information extraction unit 207 outputs a control signal 2C.
Encoder 201 having received control signal 2C sends the internal parameters to decoder 202, and stops operation. Decoder 2
02 receives an internal parameter from the encoder 201 by the control signal 2C, and starts decoding of the first speech code received from the transmission path A based on the internal parameter. By a series of these operations, the encoder 101 of the transmitting end 10 in the silent period.
Despite the fact that no information is exchanged between the relay node 21 and the decoder 202, the internal state is kept the same, so that the decoding process can be restarted normally.

The receiving end 30 receives the second speech code which is continuously transmitted from the relay node 21 via the transmission path B and whose silent section is interpolated with the pseudo background noise code. The decoder 302 performs a decoding process based on the input second audio code, decodes the audio signal into an audio signal, and outputs the audio signal from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 101 of the transmitting end 10 in the voiced interval, and therefore, in the voiced interval, the encoder 101 of the transmitting end 10 is output. And the internal state of the decoder 302 of the receiving end 30 match. In the silent period, the receiving end 30
Is the output of the encoder 201 of the relay node 21, the second speech code is input to the encoder 20 of the relay node 21.
1 and the internal state of the decoder 302 of the receiving end 30 match.

As described above, since the internal state of the encoder 101 of the transmitting end 10 and the internal state of the encoder 201 of the relay node 21 are matched even in the silent period, the apparent state from the receiving end 30 is In addition, it is possible to obtain an effect as if the voice code not subjected to the silent compression was directly received from the transmitting end 10.

As described above, the encoder 10 of the transmitting end 10
Since the internal state of the receiver 1 and the internal state of the decoder 302 of the receiving end 30 are always matched, it is possible to avoid the possibility of decoding an unpleasant sound due to the mismatch of the internal states.
Furthermore, in the relay node 21, the encoder 201 operates only in the silent section and the decoder 202 operates only in the sound section, so that they do not execute processing at the same time, and the processor load is reduced. Reduction in cost and power consumption can be realized.

Embodiment 3 Next, a third embodiment will be described with reference to FIGS. This embodiment is different from the speech coded transmission system of the first embodiment in that
The relay node 20 is modified to provide a relay node 22. Here, reference numeral 208 denotes an interpolation signal generator (interpolation signal generation means) that encodes the first speech code sent from the transmission end 10 and also encodes a background noise signal (second pseudo signal). The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

The relay node 22 receives the first voice code transmitted from the transmitting end 10 after being silently compressed and transmitted via the transmission line A. The presence / absence information extraction unit 207 constantly monitors the reception status of the voice code from the transmission line A, and controls the changeover switch 214 according to the result.
Is output. Also, at the timing when the determination by the presence / absence information extraction unit 207 transitions from “speech” to “silence”, an initialization signal for controlling the initialization of the pseudo background noise generation unit 206 from the presence / absence information extraction unit 207. 2B is output.

The pseudo background noise generation section 206 resets the internal state at the timing of receiving the initialization signal 2B, and starts transmitting the pseudo background noise signal. The pseudo-background noise generation unit 206 implements the same operation as the pseudo-background noise generation unit 106 of the transmitting end 10, and also performs the initialization operation of the pseudo-background noise generation unit 1 of the transmission end 10.
Same as 06. Since the initialization operation is performed in synchronization with the presence / absence of an audio signal input from the transmission path A and, consequently, the determination signal of the transmission end audio detector 103, the encoder 10 of the transmission end 10
1, and the same pseudo background noise signal are supplied to the interpolation signal generator 208 at the same timing. The operation up to this point is almost the same as in the first embodiment.

Next, a specific configuration of the interpolation signal generator 208 will be described with reference to FIG. The interpolation signal generator 208 shown in FIG.
It applies Recommendation G.728. As can be seen from a comparison between FIG. 4 and FIG. 21, this is obtained by adding changeover switches 421, 422, and 423 to an encoder based on ITU recommendation G.728.

[0092] The ITU recommendation G.728 LD-CELP system and G.729 CS
-In the speech coding method based on "analysis by synthesis" represented by the ACELP method, in order to select the optimal speech code, decoding processing is performed on all the assumed speech code candidates and these are input. The synthesized speech signal is compared with the speech signal one by one, the synthesized signal closest to the input speech signal is selected, and the speech code on which the synthesized signal is based is transmitted to the decoder as the optimum code. An encoder of this system has a function equivalent to a decoder usually called a local decoder in order to realize this decoding process. The function block 420 shown in FIG. 4 is a part corresponding to a local decoder.

In the section where the presence / absence information extracting section 207 determines that the signal from the transmission line A is “sound”, the control signal 2A causes the changeover switch 421 to be set to 421.
The changeover switch 422 is set to the B side, the changeover switch 423 is set to the 422B side, and the changeover switch 423 is set to the 423B side. At this time, the local decoder 420 performs a normal decoding operation by using the signal from the transmission path A as an input.

The purpose of this operation is that the backward type gain adaptation section 410 and the backward type synthesis filter adaptation section 4
11 to keep the internal states of the gain multiplier 404 and the synthesis filter 405 consistent with the internal states of the transmitting end encoder 101, not to obtain a decoded speech signal.

Here, the output path of the synthesis filter 405,
That is, since the output path of the decoded audio signal is blocked by the switch 423, the output of the local decoder 420 is not referred to from other functional blocks. In the interpolation signal generator 208, functional blocks other than the local decoder 420, that is, the PCM decompression unit 401,
The vector buffer unit 402, the adder 406, the auditory weighting filter 407, and the least square error search unit 408
Since there is no signal to be input, the stop state is maintained.

In the section where the presence / absence information extracting section 207 determines that the signal from the transmission path A is a silent section, the control signal 2A causes the changeover switch 421 to be set to 421.
The changeover switch 422 is set to the 422A side, and the changeover switch 423 is set to the 423A side. The structure of the interpolation signal generator 208 at this time is exactly the same as the structure of the encoder. Its operation is exactly the same as that of the encoder.
That is, with the pseudo background noise signal output from the pseudo background noise generation unit 206 as an input, the most similar synthesized signal is selected from the synthesized speech signal candidates synthesized by the local decoder, and the background from which the synthesized signal is based is selected. The noise code is output from the interpolation signal generator 208 as an optimum code.

Here, in the sound period, the gain multiplier 40
4 and the internal state of the synthesis filter 405 are matched with the internal state of the transmitting end encoder 101, and the signal input from the pseudo background noise generating section 206 even in the silent section is the pseudo background noise of the transmitting end 10. Since the pseudo background noise code output from the interpolation signal generator 208 is exactly the same as the generation unit 106, the pseudo background noise code output from the transmission end encoder 101 (however, not output to the transmission path A by the changeover switch 104) It is exactly the same as the noise code. By these series of operations, the transmitting end encoder 101
And the interpolation signal generator 208 always keeps the same internal state even though there is no direct signal exchange.

The receiving end 30 receives the second speech code which is continuously transmitted from the relay node 22 via the transmission path B and whose silent section is interpolated with the pseudo background noise code. The decoder 302 performs a decoding process based on the input second audio code, decodes the audio signal into an audio signal, and outputs the audio signal from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 101 of the transmitting end 10 during the voiced interval, and therefore, the encoder 101 of the transmitting end 10 is output during the voiced interval. And the internal state of the decoder 302 of the receiving end 30 match. In the silent period, the receiving end 3
0 is input to the interpolation signal generator 208.
Therefore, the internal state of the interpolation signal generator 208 and the internal state of the receiving end decoder 302 match.

As described above, even in the silent period, the internal state of the transmitting end encoder 101 and the interpolation signal generator 2
Since the internal state of 08 is matched, it is possible to obtain an effect as if the speech code from the receiving end 30 was directly received from the transmitting end 10 as if the speech code had not been subjected to silent compression.

As described above, since the internal state of the transmitting end encoder 101 and the internal state of the receiving end decoder 302 are always matched, unpleasant speech caused by the internal state mismatch is decoded. The risk of doing so is avoided. Furthermore, the number of functional blocks to be added in the relay node 22 is small, and high-quality voice transmission can be realized economically.

Embodiment 4 Next, a fourth embodiment will be described with reference to FIG. In the present embodiment, the transmission end 10 and the relay node 20 in the speech coded transmission system of the first embodiment are modified to provide a transmission end 11 and a relay node 23.

Here, reference numeral 102 denotes a decoder (transmitting side decoding means) for decoding the speech code which has been highly efficient encoded by the encoder 101 into an audio signal, and 112 designates a pseudo background noise signal which is encoded by a high efficiency encoding method. Pseudo-noise generator (transmission-side background noise generating means) for generating a converted code (first pseudo signal), 114 a switch for switching the input to the decoder 102, 212 a high-efficiency code for the pseudo-background noise signal Background noise generation unit (relay-side background noise generation means) that generates a code (second pseudo code) encoded by the coding scheme. The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

This voice transmission system is similar to relay node 2 in voice coded transmission system according to the first embodiment.
The purpose of the present invention is to construct a speech coded transmission system that enables high-quality speech transmission that solves the above-mentioned problems by applying the improvement implemented to the transmission end 11 to the transmission end 11.

Next, the operation of the speech coded transmission system according to the present embodiment will be described. At the transmitting end 11, the audio detector 103 discriminates between a sound section and a silent section based on the input original audio signal, and outputs a control signal 1A for controlling the operation mode of the transmitting end 11 based on the result. I do. If the sound detector 103 determines that the sound is “voiced”, the changeover switch 114 is set to 114B, the changeover switch 105 is set to 105B, and the changeover switch 104 is set to 104B.
Defeat each to the side. As a result, the audio signal input to the transmitting end 11 is input to the encoder 101 via the changeover switch 105, is subjected to high-efficiency encoding, and is output to the transmission line A via the changeover switch 104. .

At the same time, the speech code obtained by encoding the input speech output from the encoder 101 with high efficiency is also supplied to the decoder 102 via the changeover switch 114. The decoder 102 performs a normal decoding operation using the audio signal from the encoder 101 as an input. The purpose of this operation is to keep the internal state of the decoder 102 consistent with the internal state of the transmitting end encoder 101, not to obtain a decoded speech signal. Then, the changeover switch 115 is switched so as to prevent the obtained decoded audio signal from being output to another functional block.

At the timing when the sound detector 103 makes a transition from “voiced” to “silent”, the sound detector 103
Outputs an initialization signal 1B that controls the initialization of the pseudo background noise generation unit 112. The pseudo background noise generation unit 112
The internal state is reset at the timing when the initialization signal 1B is received. This operation is performed for exactly the same purpose as that described in the first embodiment.

When the sound detector 103 determines that the sound is "silence", the changeover switch 114 is set to the 14A side, the changeover switch 105 is set to the 105A side, and the changeover switch 104 is set to the 104A side.
Defeat each to the B side. As a result, the pseudo background noise code output from the pseudo background noise generation unit 112 is input to the decoder 102 via the changeover switch 114, and after being subjected to decoding processing, the obtained decoded pseudo background noise signal is changed to the changeover switch. The signal is output to the encoder 101 via the input terminal 105. The encoder 101 performs a normal encoding operation using the pseudo background noise signal decoded by the decoder 102 as an input. The purpose of this operation is to keep the internal state of the encoder 101 always consistent with the internal state of the decoder 102, not to obtain an encoded audio signal. Then, to prevent the obtained voice code from being output to the transmission path A,
The changeover switch 104 is switched. By these series of operations, the internal state of the encoder 101 and the internal state of the decoder 102 are always kept in the same state.

Here, in the encoder 101 and the decoder 102, although the input signal has a difference between the original audio signal and the audio code once encoded by the encoder 101, the characteristics of the audio are extremely high. It is similar.
Therefore, the internal state of the encoder 101, which corresponds to a digitized property of the audio signal, and the internal state of the decoder 102 are maintained at very similar values.

The relay node 23 receives the silence-compressed first speech code transmitted from the transmitting end 11 via the transmission path A. The presence / absence information extraction unit 207 constantly monitors the reception status of the first speech code from the transmission line A, and outputs a control signal 2A for controlling the changeover switch 214 according to the result. That is, when the first speech code is input from the transmission line A, it is determined that a sound section is present, and the switch 214 is moved to the side 214B. The first speech code received by the relay node 23 is transmitted to the transmission path B via the switch 214 as it is.
Is relayed to

On the other hand, when the presence / absence information extraction unit 207 does not detect the reception of the signal from the transmission line A, that is, in a section where it is determined that there is “silence”, the control signal 2A causes the changeover switch 214 to be turned on. Push down to 214A side. The encoded data of the pseudo background noise signal is supplemented and output to the transmission path B. Of course, as in the first embodiment,
At the timing when the determination by the presence / absence information extraction unit 207 transitions from “speech” to “silence”, the presence / absence information extraction unit 20
7 outputs an initialization signal 2B for controlling the initialization of the pseudo background noise generation unit 212. Pseudo background noise generator 212
Resets the state at the timing when the initialization signal 2B is received.

The receiving end 30 receives the second speech code which is continuously transmitted from the relay node 23 via the transmission path B and whose silent section is interpolated with the pseudo background noise code. The decoder 302 performs a decoding process based on the input second audio code, decodes the audio signal into an audio signal, and outputs the audio signal from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 101 of the transmitting end 11 in the voiced interval, and therefore, the encoder 101 of the transmitting end 11 is output in the voiced interval. And the internal state of the decoder 302 at the receiving end 30 match. In the silent period, the receiving end 30
Is the output of the pseudo background noise generator 212 of the relay node 23. As described above, the output of the pseudo-background noise generation unit 212 of the relay node 23 matches the output of the pseudo-background noise generation unit 112 of the transmission end 11, so that the output from the reception end 30 looks as if it were transmitted. An effect can be obtained as if the pseudo background noise code not subjected to silence compression was directly received from the end 11.

Further, the second speech code which transits from “silence” to “voiced” and is input to the decoder 302 is encoded by the encoder 101.
The output signal of the encoder 10 of the transmitting end 11
1 and the internal state of the decoder 302 at the receiving end 30 match very well, so that it is possible to prevent the signal output from the decoder 302 from generating abnormal noise due to the difference in the internal state. .

As described above, the encoder 10 of the transmitting end 11
Since the internal state of the receiver 1 and the internal state of the decoder 302 of the receiving end 30 are always matched, it is possible to avoid the possibility of decoding an unpleasant sound due to the mismatch of the internal states.
Also, the number of functional blocks to be added in the relay node 23 is small, and high-quality voice transmission can be realized economically.

Embodiment 5 FIG. Next, a fifth embodiment will be described with reference to FIG. The present embodiment is different from the transmitting end 11 shown in the fourth embodiment in the second embodiment.
In addition to the improvement in the relay node 21 shown in FIG.
It is set to 2. Note that the same or equivalent components as those of the fourth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

At the transmitting end 12, the voice detector 103
Performs a distinction between a sound section and a silent section based on the input original audio signal, and outputs a control signal 1A for controlling the operation mode of the transmitting end 12 based on the result. Voice detector 10
If it is determined as “voiced” in step 3, the changeover switch 104
To the 104B side. As a result, the audio signal input to the transmission end 12 is input to the encoder 101, is subjected to high-efficiency encoding, and is output to the transmission path A via the changeover switch 104.

Next, the voice detector 103 determines that there is "voice".
From the voice detector 10 to the “silence”
3 outputs an initialization signal 1B for controlling the initialization of the pseudo background noise generation unit 112. Pseudo background noise generator 112
Resets the internal state at the timing of receiving the initialization signal 1B, and sets the pseudo background noise code (first pseudo code).
Starts sending.

The pseudo-background noise generation unit 112 realizes the same operation as the pseudo-background noise generation unit 212 of the relay node 23. The pseudo-background noise generation unit of the relay node 23 also performs the initialization operation. Same as 212.
Since the initialization operation is performed in synchronization with the presence / absence of an audio signal input from the transmission path A, and eventually with the determination signal of the audio detector 103 of the transmitting end 12, the switch 102 of the decoder 102 of the transmitting end 12 and the relay node 23 214, the same pseudo background noise code can be supplied at the same timing.

At the same time, the initialization signal 1B is also supplied to the encoder 101 and the decoder 102. Upon receiving the initialization signal 1B, the encoder 101 sends the internal parameters to the decoder 102 and stops operating. The decoder 102 receives internal parameters from the encoder 101 by the initialization signal 2B,
Based on this, the decoding process of the pseudo background noise code supplied from the pseudo background noise generation unit 112 is started.

Finally, at the timing when the determination by the voice detector 103 transits from “silence” to “voiced”, the voice detector 103 outputs a control signal 1C. Upon receiving the control signal 1C, the decoder 102 converts the internal parameters into the encoder 10
1 and stop the operation. The encoder 101 receives the internal parameters from the decoder 102 by the control signal 1C, and starts encoding the original audio signal based on the internal parameters. By a series of these operations, the internal state of the encoder 101 and the internal state of the decoder 102 are always kept in the same state, so that the encoding process can be restarted normally.

In relay node 23, the fourth embodiment
The same operation as the operation indicated by is performed. Further, the receiving end 30
In the second example, the continuous transmission is performed from the relay node 23 via the transmission path B, and the silent section is interpolated with the pseudo background noise code.
Receiving the voice code of. The decoder 302 receives the input second
The decoding process is performed based on the audio code of the above, decoded into an audio signal, and output from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 101 of the transmitting end 12 in the voiced interval, and therefore, the encoder 101 of the transmitting end 12 is output in the voiced interval. And the internal state of the decoder 302 of the receiving end 30 match. In the silent period, the receiving end 30
Is the output of the pseudo background noise generator 212 of the relay node 23. As described above, the output of the pseudo-background noise generation unit 212 of the relay node 23 matches the output of the pseudo-background noise generation unit 112 of the transmission end 12, so that the output from the reception end 30 looks as if it were transmitted. It is possible to obtain an effect as if the voice code not subjected to silence compression was directly received from the end 12.

As described above, the encoder 10 of the transmitting end 12
Since the internal state of the receiver 1 and the internal state of the decoder 302 of the receiving end 30 are always matched, it is possible to avoid the possibility of decoding an unpleasant sound due to the mismatch of the internal states.
Furthermore, at the transmitting end 12, the encoder 101 operates only in a silent period and the decoder 102 operates only in a sound period. Reduction in cost and power consumption can be realized. Also, the number of functional blocks to be added in the relay node 23 is small, and high-quality voice transmission can be realized economically.

Embodiment 6 FIG. Next, a sixth embodiment will be described with reference to FIGS. In the present embodiment, the transmitting end 13 is obtained by adding the improvement in the relay node 22 shown in the third embodiment to the transmitting end 11 shown in the fourth embodiment. Here, reference numeral 108 denotes an encoder with an internal state holding function (transmission side encoding means) having the same function as the interpolation signal generator 208 in the third embodiment. Note that the same or equivalent components as those of the fourth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

First, the encoder with internal state holding function 108
Will be described with reference to FIG. This shows an example in which ITU recommendation G.728 is applied to a high-efficiency speech coding method as an example to show a specific configuration of the encoder 108 with an internal state holding function. As can be seen by comparing FIGS. 8 and 4, this is because the interpolation signal generator 208 in the third embodiment is used.
There is no difference from the structure. Therefore, description of each functional block is omitted.

Next, the operation of the voice transmission system according to the sixth embodiment will be described with reference to FIG. Voice detector 10
3, in the section where the signal from the transmission path A is determined to be “voiced”, the control signal 1A causes the changeover switch 421 to move to the side 421B and the changeover switch 422 to move to the side 422B.
Switch 423 to the side 423B. At this time, the structure and operation of encoder with internal state holding function 108 are the same as the structure and operation of encoder 101 in the first embodiment. That is, with the input original audio signal as an input, a synthesized signal that is the closest to the candidate of the synthesized audio signal synthesized by the local decoder is selected, and the background noise code that is the source of the synthesized signal is used as the optimum code.
Output from the encoder with internal state holding function. This output signal is transmitted to the transmission path A.

In the section in which the signal from the transmission line A is determined to be "silence" in the sound detector 103, the control signal 1A causes the changeover switch 421 to be set to 421A, the changeover switch 422 to be set to 422A, and the changeover switch 422 to be set. 423
To the 423A side. Local decoder section 4
20 performs a normal decoding operation using the signal from the pseudo background noise generation unit 112 as an input. The purpose of this operation is to operate the backward-type gain adaptation unit 410 and the backward-type synthesis filter adaptation unit 411 so that the gain multiplier 404,
The internal state of the synthesis filter 405 is determined by the receiving end decoder 302.
To keep it consistent with the internal state of
It is not to obtain a decoded audio signal.

Since the output of the synthesis filter 405, that is, the output path of the decoded audio signal is cut off by the switch 423, the output of the local decoder 420 is not referred to from other functional blocks. . Also, in the encoder 108 with the internal state holding function, functional blocks other than the local decoder 420, that is, the PCM decompression unit 401, the vector buffer unit 402, and the adder 40
6. The auditory weighting filter 407 and the least square error search unit 408 are stopped because there is no signal to be input.

In relay node 23, the fourth embodiment
The same operation as the operation indicated by is performed. Further, the receiving end 30
In the second example, the continuous transmission is performed from the relay node 23 via the transmission path B, and the silent section is interpolated with the pseudo background noise code.
Receiving the voice code of. The decoder 302 receives the input second
The decoding process is performed based on the audio code of the above, decoded into an audio signal, and output from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 108 with the internal state holding function of the transmitting end 13 in the sounding interval, and therefore, the transmitting end in the sounding interval. 13 and the internal state of the decoder 302 at the receiving end 30 coincide with each other. The second speech code input to the receiving end 30 in the silent section is the output of the pseudo background noise generation unit 212 of the relay node 23. Since this is the same as the output of the pseudo background noise generation unit 112 of the transmitting end 13, the internal state of the encoder 108 with the internal state holding function of the transmitting end 13 matches the internal state of the decoder 302 of the receiving end 30. . As described above, apparently from the receiving end 30, it is as if the transmitting end 13.
As a result, it is possible to obtain an effect as if the voice code not subjected to the silence compression was directly received.

As described above, since the internal state of the encoder 108 with the internal state holding function of the transmitting end 13 always coincides with the internal state of the decoder 302 of the receiving end 30, the internal state is not correct. The risk of decoding harsh speech due to the matching is avoided. Further, only a small number of functional blocks need to be added in the relay node 23, and high-quality voice transmission can be realized economically.

Embodiment 7 FIG. Next, a seventh embodiment will be described with reference to FIGS. This embodiment is different from the transmitting end 13 shown in the sixth embodiment in that the relay node 22 shown in the third embodiment is improved.
This is the transmitting end 14. Further, a relay node 24 is obtained by improving the relay node 23 shown in the sixth embodiment.

Here, reference numeral 110 denotes a coder with an internal state holding function (transmission side coding means) obtained by improving the encoder with an internal state holding function 108 of the transmitting end 13 in the sixth embodiment, and 111 denotes a pseudo background noise. A generation parameter generation unit (transmission-side parameter generation unit), 211 is a pseudo-background noise generation parameter generation unit (relay-side parameter generation unit), and 210 is a simple encoder having some functions of the encoder in the sixth embodiment. is there. Note that the same or equivalent components as those of the sixth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

First, the encoder 110 with an internal state holding function
Will be described with reference to FIG. This shows a specific configuration of the encoder 110 with an internal state holding function. As an example, the ITU recommendation G.728 is applied to a high-efficiency speech coding scheme.
Is applied. As can be seen by comparing FIGS. 10 and 8, this is almost the same as the structure of the encoder with internal state holding function 108 in the sixth embodiment except that the position of the changeover switch 421 is different. Therefore, description of each functional block is omitted.

[0136] In addition, the encoding method is based on ITU recommendation G.728 LD-CELP.
When the method is used, an excitation signal for driving a synthesis filter is used as an example of a signal output from the pseudo background noise generation parameter generation unit 111. The decoder based on the LD-CELP speech coding system has a structure in which a human generation mechanism is modeled. That is, an excitation signal corresponding to a human vocal tract sound source is used as a drive source, and a sound is synthesized by a synthetic filter modeling human vocal tract information. As a speech code, a waveform codebook obtained by vector-quantizing an excitation signal is used. The pseudo-background noise generation parameter generation unit 111 is a functional block that generates an excitation signal before being quantized by some means.

Next, the operation of the voice transmission system according to the seventh embodiment will be described with reference to FIG. Voice detector 10
In the section where the voice signal is determined to be "voiced" in 3, the control signal 1A causes the changeover switch 421 to be switched to 4
The changeover switch 422 is set to the 422B side, and the changeover switch 423 is set to the 423B side. At this time, the structure and operation of the encoder 110 with the internal state holding function are the same as the structure and operation of the normal encoder 101 in the first embodiment. That is, with the input original audio signal as an input, a synthesized signal that is the closest from the synthesized audio signal candidates synthesized by the local decoder is selected, and the background noise code that is the source of the synthesized signal is set as the optimum code, Output from encoder 110 with state holding function. This output signal is transmitted to the transmission path A.

In the section where the voice signal is determined to be "silence" by the voice detector 103, the control signal 1A causes the changeover switch 421 to be set to 421A, the changeover switch 422 to be set to 422A, and the changeover switch 423 to be set to 423.
Defeat each to A side. The local decoder unit 420 receives the excitation signal (first parameter signal) output from the pseudo background noise generation parameter generation unit 111, and performs a decoding operation using this signal. This decoding processing is the same as that obtained by skipping the inverse quantization means 403 from the ordinary decoding processing.

The purpose of this operation is to adjust the backward gain adaptation section 410 and the backward synthesis filter adaptation section 4.
11 to operate, the gain multiplier 404, the synthesis filter 4
05, the internal state of the receiving end decoder 302,
This is in order to always keep the same state, not to obtain a decoded audio signal. Then, the synthesis filter 405
, The output path of the decoded audio signal is blocked by the switch 423.
The output of 20 is not referred to from other function blocks. In the encoder 110 with the internal state holding function, the function blocks other than the local decoder 420, that is, the PCM decompression unit 401, the vector buffer unit 402, the adder 406, the auditory weighting filter 407, and the least square error search unit 408, There is no signal to be stopped.

The relay node 24 includes the same pseudo-background noise generation parameter generation unit 211 as the transmission end 14, and operates in synchronization with the pseudo-background noise generation parameter generation unit 111 of the transmission end 14. This method is realized by using a method similar to that of pseudo background noise generating sections 106 and 206 in the first embodiment. Relay node 2
In order to interpolate the speech code of the silent section missing in FIG. 4, the pseudo encoder 210 having the function of encoding the excitation signal (second parameter signal) output from the pseudo background noise generation parameter generation unit 211 generates a pseudo code. The background noise code is interpolated.

Next, the structure of the simple encoder 210 is shown in FIG. In the vector buffer 431, the excitation signal that is temporally continuous is accumulated for five samples, vectorized, and output to the adder 434. On the other hand, the inverse quantizer 432 performs inverse quantization on all the code candidates stored in the waveform codebook to generate excitation signal candidates. All of these excitation signal candidates are input to the adder 434, a square error with respect to the input excitation signal is obtained, and a code that generates the excitation signal candidate closest to the input is output as an optimum code. This makes it possible to generate a pseudo background noise code equivalent to a code based on the LD-CELP scheme.

Further, at the receiving end 30, the relay node 24
, And receives a second speech code in which a silent section is interpolated by a pseudo background noise code. The decoder 302 performs a decoding process based on the input second audio code, decodes the audio signal into an audio signal, and outputs the audio signal from the receiving end 30.

Here, the second speech code input to the receiving end 30 is the output of the encoder 110 with the internal state holding function of the transmitting end 14 in the sounding interval, and therefore, the transmitting end in the sounding interval. 14 and the internal state of the decoder 302 at the receiving end 30 are the same. The second speech code input to the receiving end 30 in the silent section is the output of the pseudo background noise generation parameter generation unit 211 of the relay node 24. Since this is the same as the output of the pseudo-background noise generation parameter generation unit 111 of the transmitting end 14, the internal state of the encoder 110 with the internal state holding function of the transmitting end 14 and the decoder 30 of the receiving end 30
The internal states of 2 match. As described above, it is possible to obtain an effect as if the voice code that is not subjected to the silence compression is directly received from the transmission terminal 14 as apparent from the reception terminal 30.

As described above, since the internal state of the encoder 110 with the internal state holding function at the transmitting end 14 and the internal state of the decoder 302 at the receiving end 30 are always matched, the internal state is not correct. The risk of decoding harsh speech due to the matching is avoided. Further, only a small number of functional blocks need to be added in the relay node 24, and high-quality voice transmission can be realized economically.

Embodiment 8 FIG. Hereinafter, the eighth embodiment will be described with reference to FIGS. In the present embodiment, the transmitting end 10 and the relay node 20 in the speech coded transmission system of the first embodiment are improved to provide a transmitting end 15 and a relay node 25. Here, 213 is a timer that counts a transition section immediately after the first audio signal output from the transmitting end 15 is switched from “silence” to “voiced”. The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

Next, the operation of the eighth embodiment will be described. At the transmitting end 15, the audio detector 103 discriminates between a sound section and a silent section based on the input original audio signal, and outputs a control signal 1A for controlling the operation mode of the transmitting end 15 based on the result. I do. When the sound detector 103 determines that the sound is present, the switch 104 is moved to the side 104B. As a result, the original audio signal input to the transmission terminal 15 is input to the encoder 101, is subjected to high-efficiency encoding, and is output to the transmission line A via the changeover switch 104.

On the other hand, if the sound detector 103 determines that there is no sound, the switch 104 is moved to the side 104A. At this time, the voice code output from the encoder 101 is cut off by the changeover switch 104, and nothing is output to the transmission path A. Also, at the timing when the sound detector 103 detects that the state has transitioned from the “silence” state to the “sound” state,
An initialization signal 1B for initializing the encoder 101 is output to the encoder. The encoder 101 performs an internal state initialization operation at the timing when the encoder 101 receives the initialization signal 1B.

In the present embodiment, the operation of the relay node 25 has three modes depending on the state of the control signal output from the presence / absence information extraction unit 207. This operation mode will be described with reference to FIG. FIG.
7 is a waveform diagram of an original audio signal input to the encoder 101 of the transmitting end 15. FIG. The vertical axis represents the signal level, and the horizontal axis represents time. The voiced / silent information extraction unit 207 constantly monitors the reception status of the voice code from the transmission line A that reflects the operation of the voice detector 103 of the transmitting end 15 and, as a result, classifies the voice code into three sections corresponding to the operation mode. Then, the operation of the relay node 25 is controlled.

First, a period during which no talk spurt is detected from a highly efficient speech code manually input to the relay node 25
(Silent section) is set to mode 1. Second, the relay node 25
Mode 2 is a period from several tens msec to several hundred msec (referred to as a transition period or a transition period) from the start of the detection of the talk spurt from the high-efficiency speech code input to (1). Third
In the meantime, after the mode 2, a period during which a talk spurt is continuously detected is referred to as a mode 3. The sound / non-sound information extraction unit 207 and the timer 213 output the control signals 2A and 2F reflecting the operation mode determination result described above, and supply the control signals to the functional blocks of the relay node 25.

Here, the continuation time (transition period) of mode 2
The value of several tens to 100 msec was presented as
The basis for this value is based on the following rules of thumb. First, as a precondition, ITU recommendation G.728 (LD-CE
It is assumed that the internal state of the encoder 201 of the relay node 25 is completely different from the internal state of the decoder 302 of the receiving end 30 using the LP method. Based on this precondition, encoder 2
01 and the decoder 302 perform encoding / decoding via a transmission path. Since the stability of all filters used in the LD-CELP method is guaranteed, the internal state of transmission and reception gradually converges in the direction in which transmission and reception become the same.

While the encoding / decoding is continued as it is, the internal states are sufficiently coincident to such an extent that there is no possibility that abnormal noise will occur. The time required from the mode transition until the internal states sufficiently match is several tens msec to 100 msec. Needless to say, this value is expected to change depending on the high-efficiency coding method used, and it is important to set the transition period according to each coding method.

FIG. 14 is a mode transition diagram summarizing the transition of each mode described above. Transitions between the three modes are allowed only in the directions indicated by the arrows, and the other transitions are prohibited transitions or physically impossible transitions.

Next, the operation of the relay node 25 for each mode will be described. First, in the mode 1, that is, in the section where the sound / silence information extraction unit 207 determines “silence”, the changeover switch 205 is set to 205A and the changeover switch 214 is set to 214A. At this time, the output signal on the side of the pseudo background noise generation unit 206 is
Is input to the encoder 201 via the. The encoder 201 performs an encoding process based on the input pseudo background noise signal, and outputs a pseudo background noise code. The pseudo background noise code output from the encoder 201 is transmitted to the transmission path A via the changeover switch 214. With this series of actions,
A signal in a silent section can be supplemented with a pseudo background noise signal. At this time, since the decoder 202 loses the input signal, the operation is stopped.

When a transition from the mode 1 to the mode 2 is made, that is, when the speech code starts to be detected in the speech / silence information extraction section 207, the reset signal 2 is sent to the decoder 202.
Issue B. Since the operation of the sound / silence information extraction unit 207 faithfully reflects the operation of the voice detector 103,
This is performed in synchronization with the initialization of the encoder 101 of the transmitting end 15.

In the mode 2, the changeover switch 205
To 205B and the changeover switch 214 remains connected to 214A. The speech code received by the relay node 25 is supplied to the decoder 202. The encoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is supplied to the switch 205
Is supplied to the encoder 201 via Encoder 201
Performs an encoding process based on the input audio signal.
This voice code is output to the transmission line B via the changeover switch 214. That is, the relay node 25 has the same form as the tandem connection.

In the mode 3, the switch 214 which has received the control signal 2F from the timer
Fall to the side. The changeover switch 205 is kept connected to the side 205B. The voice code received by the relay node 25 is transmitted to the transmission
Is relayed to The speech code is also supplied to the decoder 202. The decoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is sent to the encoder 2 via the changeover switch 205.
01 is supplied. The encoder 201 performs an encoding process based on the input audio signal.

Here, in the encoder 101 of the transmitting end 15 and the encoder 201 of the relay node 25, there are differences in the input signals between the original audio signal and the decoded audio signal decoded by the decoder 202, respectively. However, the nature of the speech is very similar. Therefore, the internal state of the transmitting end encoder 101 and the internal state of the relay encoder 201, which correspond to a numerical value of the property of the audio signal, are held at very similar values.

In the present embodiment, the case where the mode is suddenly shifted from mode 1 to mode 3 will be considered. For the time being, it is guaranteed that the internal state of the transmitting end encoder 101 and the relay decoder 202, and the internal state of the relay encoder 201 and the internal state of the receiving end decoder 302 are identical. However, the internal states of the transmitting end encoder 101 and the receiving end decoder 302 are not guaranteed to match at all. Therefore, when the transition from the mode 1 to the mode 3 jumps one step, it becomes exactly the same as the digital one-link form shown in the conventional example, and an abnormal sound is generated due to the mismatch of the internal state. Here, by providing mode 2 as a transition section during the transition from mode 1 to mode 3, the internal state of the transmitting end encoder 101 and the internal state of the receiving end decoder 302 sufficiently match. Even when the mode is shifted to the mode 3, it is possible to avoid the generation of abnormal noise.

In the speech coded transmission system according to this embodiment, the period during which a tandem connection, which is known to cause voice quality degradation, is allowed to take a short period of time during a transition from a silence state to a speech state. By restricting and connecting most talk spurts by one-link encoding / decoding, it is possible to avoid quality degradation and to make full use of the performance of the high-efficiency speech coding scheme. Further, the processing load of the processor and the hardware scale at the relay node can be reduced. In the present embodiment, the system in which the ITU recommendation G.728 LD-CELP system is applied to the high-efficiency coding system has been described, but it is shown that the application example of the present invention is limited to this coding system. However, it is easy to guess that the present invention can be applied to any audio coding method using the past coding / decoding results.

Embodiment 9 FIG. Hereinafter, the ninth embodiment will be described with reference to FIG. In the present embodiment, with respect to the relay node 25 shown in the eighth embodiment,
The timer 213 is replaced with a comparator (comparing means) 215 to provide a relay node 26. Here, the comparator 215 numerically compares the internal state of the decoder 202 and the internal state of the encoder 201. The same or equivalent components as those in the eighth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

Next, the operation of the relay node 26 will be described.
First, in the mode 1, that is, in the section in which the sound / silence information extraction unit 207 determines “silence”, the changeover switch 205 is set to the 205A side, and the changeover switch 214 is set to 214.
Defeat each to A side. At this time, the pseudo background noise generator 2
The output signal on the 06 side is input to the encoder 201 via the changeover switch 205. The encoder 201 performs an encoding process based on the input pseudo background noise signal, and outputs a pseudo background noise code. The pseudo background noise code output from the encoder 201 is transmitted to the transmission path A via the changeover switch 214. Through this series of operations, a signal in a silent section can be supplemented with a pseudo background noise signal. At this time, since the decoder 202 loses the input signal, the operation is stopped.

When a transition is made from mode 1 to mode 2, that is, when the speech / non-speech information extraction unit 207 starts detecting a speech code, the decoder 202 sends a reset signal 2 to the decoder 202.
Issue B. Since the operation of the sound / silence information extraction unit 207 faithfully reflects the operation of the voice detector 103,
This is performed in synchronization with the initialization of the encoder 101 of the transmitting end 15.

In the mode 2, the changeover switch 205
To 205B and the changeover switch 214 remains connected to 214A. The speech code received by the relay node 25 is supplied to the decoder 202. The encoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is supplied to the switch 205
Is supplied to the encoder 201 via Encoder 201
Performs an encoding process based on the input audio signal.
This voice code is output to the transmission line B via the changeover switch 214. That is, the relay node 25 has the same form as the tandem connection.

In the mode 3, the internal parameters of the encoder 201 and the internal parameters of the decoder 202 are given to the comparator 215. The comparator 215 determines whether these internal parameters match within a predetermined level,
When they match within this level, the control signal 2D is output to the changeover switch 214. And this control signal 2F
Is received, the switch 214 falls to the 214B side.
The changeover switch 205 is kept connected to the side 205B.

The voice code received by the relay node 25 is directly relayed to the transmission line B via the changeover switch 214. The speech code is also supplied to the decoder 202. The decoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is supplied to the encoder 201 via the switch 205. The encoder 201 performs an encoding process based on the input audio signal.

Here, in the encoder 101 of the transmitting end 15 and the encoder 201 of the relay node 25, there are differences in the input signals between the original audio signal and the decoded audio signal decoded by the decoder 202, respectively. However, the nature of the speech is very similar. Therefore, the internal state of the transmitting end encoder 101 and the internal state of the relay encoder 201, which correspond to a numerical value of the property of the audio signal, are held at very similar values.

As described above, the determination of the transition timing from mode 2 to mode 3 is controlled by the determination result of the comparator 215, so that the internal state of the encoder 101 of the transmitting end 15 and the decoder of the receiving end 30 are controlled. It is possible to more reliably match the internal state of the motor 302 and the internal state of the motor 302, and there is an effect of further reducing the possibility of generating abnormal noise.

Embodiment 10 FIG. Hereinafter, the tenth embodiment will be described with reference to FIG. In the present embodiment, a comparator (comparison means) 215 and a judgment circuit (judgment means) are provided for the relay node 25 shown in the eighth embodiment.
216 is added to form the relay node 27. Here, the comparator 215 numerically compares the internal state of the decoder 202 and the internal state of the encoder 201, and the determination circuit 216 compares the control signal 2F output from the timer 213 with the comparator 215. Control signal 2 output from
The transition timing to mode 3 is determined based on D. The same or equivalent components as those in the eighth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

Next, the operation of the relay node 27 will be described.
First, in the mode 1, that is, in the section in which the sound / silence information extraction unit 207 determines “silence”, the changeover switch 205 is set to the 205A side, and the changeover switch 214 is set to 214.
Defeat each to A side. At this time, the pseudo background noise generator 2
The output signal on the 06 side is input to the encoder 201 via the changeover switch 205. The encoder 201 performs an encoding process based on the input pseudo background noise signal, and outputs a pseudo background noise code. The pseudo background noise code output from the encoder 201 is transmitted to the transmission path A via the changeover switch 214. Through this series of operations, a signal in a silent section can be supplemented with a pseudo background noise signal. At this time, since the decoder 202 loses the input signal, the operation is stopped.

When a transition is made from mode 1 to mode 2, that is, when the speech code starts to be detected in the speech / silence information extraction section 207, the reset signal 2 is sent to the decoder 202.
Issue B. Since the operation of the sound / silence information extraction unit 207 faithfully reflects the operation of the voice detector 103,
This is performed in synchronization with the initialization of the encoder 101 of the transmitting end 15.

In the mode 2, the changeover switch 205
To 205B and the changeover switch 214 remains connected to 214A. The speech code received by the relay node 25 is supplied to the decoder 202. The encoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is supplied to the switch 205
Is supplied to the encoder 201 via Encoder 201
Performs an encoding process based on the input audio signal.
This voice code is output to the transmission line B via the changeover switch 214. That is, the relay node 25 has the same form as the tandem connection.

In the mode 3, the internal parameters of the encoder 201 and the internal parameters of the decoder 202 are given to the comparator 215. The comparator 215 determines whether these internal parameters match within a predetermined level,
If they match within this level, the control signal 2D is output to the determination circuit 216. Further, the timer 213 counts the transition timing to the mode 3 and outputs the control signal 2F to the determination circuit 216 with a predetermined count value.

In the determination circuit 216, these control signals 2
D and 2F are received, and both control signals 2D and 2F are received.
When F is input, the control signal 2E is switched to the changeover switch 2
14 is output. Then, the changeover switch 214 that has received the control signal 2E falls down to the 214B side. The changeover switch 205 is kept connected to the side 205B.

The voice code received by the relay node 25 is relayed as it is to the transmission line B via the changeover switch 214. The speech code is also supplied to the decoder 202. The decoder 202 performs a decoding process based on the input audio code, and decodes the audio signal. The decoded audio signal is supplied to the encoder 201 via the switch 205. The encoder 201 performs an encoding process based on the input audio signal.

Here, in the encoder 101 of the transmitting end 15 and the encoder 201 of the relay node 25, there are differences in the input signals between the original audio signal and the decoded audio signal decoded by the decoder 202, respectively. However, the nature of the speech is very similar. Therefore, the internal state of the transmitting end encoder 101 and the internal state of the relay encoder 201, which correspond to a numerical value of the property of the audio signal, are held at very similar values.

As described above, the determination of the transition timing from mode 2 to mode 3 is controlled by the determination result of this determination unit, so that the internal state of encoder 101 at transmitting end 15 and the decoding at receiving end 30 are determined. The internal state of the container 302 can be more reliably matched. As a result, it is possible to prevent the sound quality from deteriorating due to the tandem time being extended for a long time, while reducing the possibility of generating abnormal noise.

Embodiment 11 FIG. FIG. 17 shows Embodiment 1.
Background noise generators 106 and 2 in ６，6, 8 and 10
06, details of an example of the structure of the pseudo-background noise generation parameter generators 111 and 211 in the seventh embodiment. The interpolation signal is stored in a cyclic memory shown in FIG. The pseudo background noise signal sequence is sequentially stored at a predetermined address in the memory space on a sample basis (or on a frame basis for the pseudo background noise code). When the pseudo background noise signal is output, the sample indicated by the address pointer is output, and the pointer is advanced by one address.

By making the memory cyclic, it is possible to cope with silent sections of any length. However, if the memory size is reduced, the signal component of a specific frequency is added to the pseudo background noise signal due to the periodicity of the cyclic memory. The size of the memory should be large enough, because it has a bad effect of riding. Here, these pseudo background noise generation units 106,
206 or pseudo background noise generation parameter generation unit 11
When 1, 211 receive the initialization signal 2B, the position of the pointer is reset to address 0. With this operation, even if a disconnection state due to silence compression occurs between the transmitting terminals 10 to 15 and the pseudo background noise generation units 106 and 206 of the relay nodes 20 to 27, synchronized operations can be realized at both ends.

Embodiment 12 FIG. FIG. 18 is Embodiment 1
Pseudo background noise generators 106, 20 in 6, 8, 10
6, an example of the structure of the pseudo background noise generation parameter generation units 111 and 211 according to the seventh embodiment is described in detail. Reference numeral 441 denotes a white noise generation unit (random signal generator), and 442 denotes a filter. White noise generator 441
Is the background noise component that occurs in normal voice communication,
Focusing on strong whiteness (randomness), a background noise signal to be interpolated in a silent section is pseudo-realized by an output signal of a random number generator. The method of generating random numbers is
For example, there are a method using a PN pattern and a method for taking a remainder.

Here, these pseudo background noise generators 10
6, 206 or when the pseudo-background noise generation parameter generation units 111 and 211 receive the initialization signal 2B, the transmission ends 10 to 15 and the relay node 20 are set as seeds, which are initial values of the random number generator. To 27 are given preset values. By this operation, even if a disconnection state due to silence compression occurs between the transmission terminals 10 to 15 and the pseudo background noise signal generation units 106 and 206 of the relay nodes 20 to 27, synchronized operations can be realized at both ends. . Further, by inserting a filter 442 (for example, a filter having a 1 / f characteristic) having a characteristic close to the frequency characteristic of natural background noise at the output terminal of the random number generator, a pseudo background noise signal that is less uncomfortable Can be obtained.

Embodiment 13 FIG. Next, a thirteenth embodiment will be described with reference to FIG. In the present embodiment, the voice detector 10 of the transmitting end 10 shown in the first embodiment is used.
Channel 3 (synchronous signal sending means) 103 between the communication node 3 and the sound / silence information extraction unit 207 of the relay node 20
A (for example, an ISDN D channel). The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

As described in the first embodiment, there is a method of checking the presence or absence of a speech code as a criterion for determining a sound section and a silent section. For example, when the transmission path A is an ATM network, it is one way to determine that a transition has been made to a silent section if no cell is transmitted from the transmitting end 10.

However, when cell discarding, which is a deterioration factor peculiar to the network, occurs in the ATM network, the relay node 20 may erroneously recognize a silent section. In order to avoid this, the voice detector 1 of the transmitting end 10
A signaling channel 103 </ b> A is connected between the relay node 03 and the sound / silence information extraction unit 207 of the relay node 20. Then, by using the signaling channel 103A, the information notifying that the transition from the voiced section to the silent section is separately transmitted from the transmitting end 10 to the relay node 20, it is possible to prevent erroneous recognition of the silent transition. High quality voice transmission can be realized.

In the second to ninth embodiments as well, the voice detector 103 of each of the transmitting terminals 10 to 15 and the relay nodes 21 to 2
It is needless to say that the present method in which the signaling channel 103A is connected to the voiced / silent information extracting unit 207 of No. 7 and the information is separately notified from the transmitting ends 10 to 15 to the relay nodes 21 to 27 can be applied.

Embodiment 14 FIG. Next, a fourteenth embodiment will be described with reference to FIG. In the present embodiment, a transmitting end 16 is obtained by adding an identification pattern generation unit (identification pattern generating means) 217 to the transmitting end 10 shown in the first embodiment. Here, the identification pattern generation unit 217 generates a signaling transmission code pattern indicating the end of a sound section.
The same or equivalent components as those of the first embodiment are denoted by the same reference numerals, and the description thereof will be omitted.

In the present embodiment, an improvement is made to a signaling transmission method for notifying from the transmitting end 10 to the relay node 20 that a transition has been made from a voiced section to a silent section. In the present embodiment, this signaling information is transmitted on the in-channel. That is, at the transmitting end 10,
At the timing when the voice detector 103 detects that the transition from the voiced section to the silent section has been performed, after the transmission of the final voice code,
The changeover switch 104 is switched to the 104C side. By this switching, the signaling transmission code pattern generated by the identification pattern generation unit 217 is transmitted to the relay node 20. After the transmission of the signaling transmission code pattern, the changeover switch 104 is switched to the 104A side to stop the transmission of the signaling transmission code pattern.

Voice / silence information extraction unit 2 of relay node 20
07 always monitors the signal input from the transmission line A, and outputs a control signal 2A indicating that the transition from "voiced" to "silent" has been made at the timing when the signaling transmission code pattern is recognized. By using this method, erroneous recognition of silence transition can be prevented, and high-quality voice transmission can be realized. In addition, there is an advantage that it can be realized only by adding a functional block without changing a signal system. .

[0188] Also in Embodiments 2 to 9, an identification pattern generation unit 217 is added to transmitting ends 10 to 15, and
It goes without saying that the present method of transmitting signaling information in-channel can be applied.

[0189]

According to the speech coded transmission system according to the present invention, it is possible to avoid the degradation of speech quality due to tandem connection, and it is possible to prevent the internal state at the time of encoding processing at the transmitting end and the decoding processing at the receiving end. Since the matching with the internal state is always performed, abnormal noise and harshness caused by the mismatch of the internal state can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 1.

FIG. 2 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 2.

FIG. 3 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 3.

FIG. 4 is a block diagram showing a configuration of an interpolation signal generator according to a third embodiment in detail.

FIG. 5 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 4.

FIG. 6 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 5.

FIG. 7 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 6.

FIG. 8 is a block diagram showing in detail a configuration of an encoder with an internal state holding function according to a sixth embodiment.

FIG. 9 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 7.

FIG. 10 is a block diagram showing in detail a configuration of an encoder with an internal state holding function according to a seventh embodiment.

FIG. 11 is a block diagram showing a configuration of a simplified encoder according to Embodiment 7 in detail.

FIG. 12 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 8.

FIG. 13 is a waveform diagram of an original audio signal input to the encoder at the transmitting end according to the eighth embodiment.

FIG. 14 is a state transition diagram showing transition between operation modes according to the eighth embodiment.

FIG. 15 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 9.

FIG. 16 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 10.

FIG. 17 is a diagram showing a configuration of a pseudo background noise signal generation unit according to Embodiment 11.

FIG. 18 is a diagram showing a configuration of a pseudo background noise signal generation unit according to Embodiment 12.

FIG. 19 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 13.

FIG. 20 is a block diagram showing a configuration of a speech coded transmission system according to Embodiment 14.

FIG. 21 is an example of an encoding method, ITU-T Recommendation G.728 LD-
FIG. 2 is a block diagram illustrating a system configuration of a CELP speech coding scheme.

FIG. 22 is a block diagram showing a configuration of a conventional speech coded transmission system.

FIG. 23 is a block diagram showing a configuration of a conventional coded audio transmission system using tandem connection.

FIG. 24 is a block diagram showing a configuration of a conventional voice coded transmission system using digital one-link connection.

[Explanation of symbols]

1 to 16: transmission end, 20 to 27 ... relay node, 30 ... reception end, 101 ... encoder (transmission side encoding means), 102 ...
Decoder (transmission-side decoding means), 103A: signaling channel (synchronous signal transmission means), 104: changeover switch (transmission-side speech code generation means), 105: changeover switch (transmission-side background noise insertion means), 106, 112 ... Pseudo background noise generation unit (transmission side background noise generation means), 108, 1
Reference numeral 10: an encoder with an internal state holding function (transmission side encoding means), 111: a parameter generator for generating pseudo background noise (transmission side parameter generation means), 201 ... encoder (relay side encoding means), 202: decoding Unit (relay-side decoding means), 204: switch (relay-side background noise insertion means), 206, 212 ... pseudo-background noise generation unit (relay-side background noise generation means), 208 ... interpolation signal generator (interpolation signal generation) Means), 210: Simple encoder (relay-side coding means), 211: Pseudo-background noise generation parameter generation unit (relay-side parameter generation means), 213: Timer, 21
4. Changeover switch (relay-side voice code generation means), 215
.., A comparator (comparison means), 216... A judgment circuit (judgment means), 217... An identification pattern generation section (identification pattern generation means), A... Transmission line (first transmission line), B. Transmission path).

 ──────────────────────────────────────────────────の Continued from the front page (72) Inventor Hideaki Ebisawa 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Mitsubishi Electric Corporation

Claims (20)

[Claims] 1. A transmitting end for encoding a voiced section of an original audio signal and outputting a first audio code generated by silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits a first pseudo signal imitating background noise. Transmitting-side background noise generating means for generating, transmitting-side background noise inserting means for inserting the first pseudo signal into a silent section of the original audio signal, and encoding the audio signal obtained by the transmitting-side background noise inserting means. Transmitting side encoding means for converting A transmission-side speech code generation means for compressing a silent section of the speech code encoded by the side encoding means to generate the first speech code, wherein the relay node converts the first speech code Relay-side decoding means for decoding; relay-side background noise generation means for generating a second pseudo signal simulating background noise; and 2, a relay-side background noise insertion unit that inserts the pseudo signal of No. 2, a relay-side encoding unit that encodes the audio signal obtained by the relay-side background noise insertion unit, and a signal encoded by the relay-side encoding unit. A speech coding transmission system, comprising: a speech code generation unit that inserts a speech code into a silent section of the first speech code to generate the second speech code. 2. A transmitting end for encoding a voiced section of an original audio signal and outputting a first audio code generated by silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits a first pseudo signal imitating background noise. Transmitting-side background noise generating means for generating, transmitting-side background noise inserting means for inserting the first pseudo signal into a silent section of the original audio signal, and encoding the audio signal obtained by the transmitting-side background noise inserting means. Transmitting side encoding means for converting A transmission-side speech code generation means for compressing a silent section of the speech code encoded by the side encoding means to generate the first speech code, wherein the relay node converts the first speech code Relay-side decoding means for decoding; relay-side background noise generation means for generating a second pseudo signal imitating background noise; relay-side encoding means for encoding the second pseudo signal; And a relay-side speech code generating means for inserting the pseudo code encoded by the side encoding means into a silent section of the first speech code to generate the second speech code. The speech encoding / transmission system, wherein the means inputs an internal parameter of the relay side decoding means, and holds an internal state equivalent to that of the relay side decoding means using the internal parameter. 3. A transmitting end that encodes a voiced section of an original audio signal and outputs a first audio code generated by performing silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits a first pseudo signal imitating background noise. Transmitting-side background noise generating means for generating; transmitting-side background noise inserting means for inserting the first pseudo signal in a silent section of the original audio signal; and encoding a voice signal obtained by the transmitting-side background noise transmitting means. Transmitting side encoding means for converting A transmitting-side speech code generating means for compressing a silent section of the speech code encoded by the side encoding means to generate the first speech code, wherein the relay node comprises: Relay-side background noise generating means for generating a pseudo-signal of the following; interpolating signal generating means for decoding the first speech code and encoding the second pseudo-signal; Relay-side speech code generation means for inserting the pseudo code thus obtained into a silent section of the first speech code to generate the second speech code, wherein the interpolation signal generation means comprises a decoding process and a code A speech coded transmission system characterized in that it retains an internal state common to the encoding process. 4. A transmitting end for encoding a voiced section of an original audio signal and outputting a first audio code generated by silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits the first pseudo code imitating background noise. Transmitting side background noise generating means for generating, transmitting side decoding means for decoding the first pseudo code, and inserting the pseudo signal decoded by the transmitting side decoding means into a silent section of the original audio signal Transmitting side background noise inserting means, and the transmitting side background A transmission-side encoding unit that encodes the audio signal obtained by the sound insertion unit; and a silence section of the audio code encoded by the transmission-side encoding unit is compressed to generate the first audio code. A transmission side speech code generation unit, wherein the relay node generates a second pseudo code imitating background noise, and the second background noise generation unit; And a relay-side speech code generation unit that inserts a pseudo code to generate the second speech code. The transmission-side decoding unit inputs the speech code encoded by the transmission-side encoding unit. A speech coded transmission system characterized in that the speech code is decoded for a sound section and the second pseudo code is decoded for a silent section. 5. A transmitting end for encoding a voiced section of an original voice signal and outputting a first voice code generated by silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits the first pseudo code imitating background noise. Transmitting side background noise generating means for generating; transmitting side decoding means for decoding the first pseudo code; transmitting side coding means for coding the original audio signal; The silence section of the converted speech code is compressed, A transmitting-side voice code generating means for generating a voice code, the relay node comprising: a relay-side background noise generating means for generating a second pseudo code imitating background noise; and a silent section of the first voice code. And a relay-side speech code generation unit that generates the second speech code by inserting the second pseudo code into the transmission-side decoding unit. The transmission-side coding unit inputs internal parameters of the transmission-side decoding unit. A speech coded transmission system using the internal parameters to maintain an internal state equivalent to that of the transmitting side decoding means. 6. A transmitting end for encoding a voiced section of an original audio signal and outputting a first audio code generated by silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end transmits a first pseudo signal imitating background noise. Transmitting-side background noise generating means for generating; transmitting-side coding means for coding the original audio signal while retaining an internal state based on the first pseudo signal; and coding by the transmitting-side coding means. Compressing the silent section of the speech code to produce the first speech A transmission-side speech code generation means for generating a code; wherein the relay node includes: a relay-side background noise generation means for generating a second pseudo code imitating background noise; and a silent section of the first speech code. A speech coded transmission system comprising: a relay speech code generation unit that generates the second speech code by inserting the second pseudo code. 7. A transmitting end that encodes a voiced section of an original audio signal and outputs a first audio code generated by performing silence compression to a first transmission path; A relay node that outputs a second speech code generated by interpolating background noise to a silent section of the original speech signal to a second transmission path based on the received first speech code; And a receiving end for decoding the second audio code received from the transmission path to reproduce the original audio signal, wherein the transmitting end is a transmitting side code for encoding the original audio signal. Encoding means; and a transmission-side speech code generation means for compressing a silent section of the speech code encoded by the transmission-side encoding means to generate the first speech code, wherein the relay node comprises: Decoding the first audio code to generate an audio signal Relay-side decoding means, relay-side background noise generating means for generating a pseudo signal simulating background noise, and relay-side background for inserting the pseudo signal into a silent section of the audio signal generated by the relay-side decoding means. Noise insertion means, relay-side encoding means for encoding the audio signal obtained by the relay-side background noise insertion means, and the first audio code And a relay-side speech code generation unit that generates the second speech code by inserting the second speech code into a silence section and a transition section immediately after switching from silence to speech. 8. The relay node further includes a timer for counting the transition section, and the relay-side speech code generation means determines an end timing of the transition section based on a count value of the timer. The speech coded transmission system according to claim 7, wherein: 9. The relay node further includes comparison means for comparing an internal state of the relay-side decoding means with an internal state of the relay-side encoding means. 8. The speech coded transmission system according to claim 7, wherein the end timing of the transition section is determined based on a comparison result of the means. 10. The relay node includes: a timer for counting the transition period; a comparing unit for comparing an internal state of the relay side decoding unit with an internal state of the relay side encoding unit; A determination unit that determines an end timing of the transition section based on a value and a comparison result of the comparison unit. The relay-side speech code generation unit determines the end timing of the transition section based on a determination result of the determination unit. The speech encoding transmission system according to claim 7, wherein the end timing is determined. 11. The apparatus according to claim 1, wherein the transmitting-side background noise generating means and the relay-side background noise generating means generate a digital signal simulating background noise in a silent section as a pseudo voice. Item 11. The speech coded transmission system according to any one of Items 10. 12. The apparatus according to claim 1, wherein said transmission side background noise generation means and said relay side background noise generation means are memories storing a predetermined digital signal. A speech coded transmission system according to the preceding claims. 13. The speech code as claimed in claim 1, wherein said transmitting side background noise generating means and said relay side background noise generating means are random signal generators. Transmission system. 14. The relay node further comprising: a synchronization signal transmitting unit configured to transmit a synchronization signal indicating a start of a silent section of the original audio signal from the transmission end to the relay node, wherein the relay node receives the synchronization signal, 2. The transmission-side background noise generator and the relay-side background noise generator are synchronized with each other.
0. The speech coded transmission system according to any one of 0. 15. The transmitting end further comprises an identification pattern generating means for generating an identification pattern to be superimposed on a silent section of the original audio signal, wherein the relay node recognizes the identification pattern and transmits the background of the transmission side. The speech coded transmission system according to any one of claims 1 to 10, wherein the noise generation unit and the relay-side background noise generation unit are synchronized. 16. A first speech code generated by encoding a voiced section of an original speech signal and silence-compressing the speech section.
And a second speech code generated by interpolating background noise into a silent section of the original speech signal based on the first speech code received from the first transmission path. And a receiving node that decodes a second audio code received from the second transmission line to reproduce the original audio signal. A transmission-side parameter generation unit configured to generate a first parameter signal for generating pseudo background noise; and a transmission configured to encode the original audio signal while maintaining an internal state based on the first parameter signal. Side encoding means, and a transmission side speech code generation means for compressing a silent section of the speech code encoded by the transmission side encoding means to generate the first speech code, wherein the relay node comprises: , Pseudo background noise Relay-side parameter generation means for generating a raw second parameter signal; relay-side encoding means for generating a pseudo code based on the second parameter signal; A speech coding transmission system comprising: a relay speech code generation unit that generates a second speech code by inserting a pseudo code. 17. The speech coded transmission system according to claim 16, wherein said transmission-side parameter generation means and said relay-side parameter generation means are memories storing predetermined digital signals. 18. The speech coded transmission system according to claim 16, wherein said transmission side parameter generation means and said relay side parameter generation means are random code generators. 19. The transmitting end further includes a synchronizing signal transmitting means for transmitting a synchronizing signal indicating a start of a silence section of the original audio signal, wherein the relay node receives the synchronizing signal, and 17. The speech coded transmission system according to claim 16, wherein the parameter generation unit and the relay-side parameter generation unit are synchronized. 20. The transmitting end further comprises an identification pattern generating means for generating an identification pattern to be superimposed on a silent section of the original audio signal, wherein the relay node recognizes the identification pattern, and 17. The method according to claim 16, wherein the generation unit and the relay-side parameter generation unit are synchronized.
A speech coded transmission system according to the preceding claims.