JP2002057787A - Voice companding system with echo canceller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a voice companding system with echo canceller that reduces an arithmetic quantity when the input signal is a silence signal. SOLUTION: The voice companding system with echo canceller is provided with a 1st voice silence decision section (VAD) 3 that decides voice/silence of an input voice signal, a decoder 7 that expands a signal received from a line, an echo canceller 4 that references an output signal of the decoder 7 so as to eliminate an echo from the voice signal decided by the VAD 3, a 2nd voice silence decision section (VAD) 5 that decides voice/silence of the signal after the eco is eliminated, and an encoder 6 that receives an output of the VAD 5 and the signal decided to be a silent signal by the VAD 3 to compress the signal. The encoder 6 applies silence compression to the signal decided as a silent signal and applies voiced compression processing to the signal decided to be a voiced signal.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a voice compression / decompression system, and more particularly to a voice compression / decompression system with an echo canceller.

[0002]

2. Description of the Related Art Conventional audio compression / echo canceller /
The decompression system will be described below with reference to FIG. 5 showing an example of the configuration. As shown in FIG. 5, a microphone 2 for inputting voice, an analog / digital (AD) converter 8 for converting an input from the microphone 2 into a digital signal,
An echo canceller 4 for removing an echo by referring to a signal of a line output from the output of the D converter 8 to the speaker 1;
A voice / silence determining unit (VAD) 5 for determining voice / silence of an output of the echo canceller 4, an encoder 6 for compressing output data of the VAD 5, a decoder 7 for expanding a signal input from a line, and a decoder 7, a digital-to-analog (DA) converter 9 for converting the output of
A speaker 1 for outputting the output of the A converter 9 as audio,
It is provided with.

Note that the encoder 6 and the decoder 7 may or may not have the silence compression function, and when using an encoder without the silence compression function, the VAD 5 is omitted.

Next, the operation of a conventional voice compression / decompression system with an echo canceller will be described with reference to FIG.

First, a signal such as voice input from the microphone 1 is converted from an analog signal to a digital signal by the AD converter 9 so that signal processing such as compression can be performed. AD
The signal converted into a digital signal by the converter 9 is subjected to echo removal processing in the echo canceller 4.

[0006] The signal after the echo removal processing in the echo canceller 4 is input to the VAD 5.

[0007] The output signal of the VAD 5 is input to the encoder 6, and a signal determined to be silent is subjected to silent compression, and a signal determined to be sound is subjected to sound compression processing.

The decoder 7 expands a compressed signal received from a line or the like, and outputs it to a DA converter 8. In the case of a decompressor that supports silent compression, the decoder 7 automatically generates a sound / voice depending on the type of the received signal.
Silence can be extended.

[0009] The DA converter 8 converts the received digital signal into an analog signal so as to be able to output it as a sound, and passes it to the speaker 1.

When the encoder 6 is used without the silence compression function, the VAD 5 is omitted, and the decoder 7 may not have the silence compression function or may have the silence compression function.

[0011]

However, the conventional system shown in FIG. 5 has the following problems.

That is, the input audio signal is
Regardless of silence, since the echo cancellation processing is always performed, the echo cancellation processing is performed even when the silent echo removal processing is not required, which takes a long processing time. In order to reduce the processing time, a high-performance DSP (digital signal processor) or the like is required.

The reason is that the input audio data is first subjected to echo removal processing, and there is no means for determining whether or not the audio data requires echo removal processing. That's why.

For example, Japanese Patent Laid-Open Publication No. 4-196624 discloses that a voice signal or a modem signal is transmitted and received, and when a modem signal detector detects that the transmission signal is a modem signal, an echo in the transmission signal is detected. In an echo canceling device that stops the operation of an echo cancellation processing unit that predicts and cancels a received signal, an audio detection circuit that detects that a transmission signal is a non-silent audio signal, and an audio detection signal is extracted from the audio detection circuit An echo cancellation device has been proposed to operate the echo cancellation processing unit only during a period.

Accordingly, the present invention has been made in view of the above problems, and an object of the present invention is to provide a voice compression / expansion system with an echo canceller, which reduces the amount of computation when the input is in a silent state. Is to do.

[0016]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a first sound / non-speech judging unit for judging whether an input audio signal is sound or no sound, A decoder that performs decompression, an echo canceller that performs echo removal processing with reference to an output signal of the decoder from an audio signal determined to be sound by the first sound / non-sound determining unit, The second voiced / silent determination unit that performs voiced / silent determination of the signal after the echo removal processing, the output of the second voiced / silent determination unit, and the first voiced / silent determination unit are determined to be silent. And an encoder that inputs the compressed audio signal and performs compression.

[0017]

Embodiments of the present invention will be described. In the embodiment of the present invention, referring to FIG. 1, a first voiced / silent determining unit 3 for determining voiced / silent of an input audio signal, and a decoder 7 for expanding a signal input from a line
An echo canceller unit 4 for performing echo removal processing by referring to an output signal of the decoder 7 from an audio signal determined to be voiced by the first voiced / silence determination unit 3; A second voiced / silent determination unit 5 that determines the presence / absence of voiced / non-voiced signal, an output of the second voiced / non-voiced determination unit 5, and a voiced / non-voiced determination by the first voiced / non-voiced determination unit 3 Encoder 6 that inputs the compressed signal and performs compression
And.

In the present invention, the echo canceller unit 4
The output signal of the second sound / silence determination unit 5 is input to the encoder 6 regardless of whether the second sound / non-sound determination unit 5 receives the signal subjected to the echo removal processing and determines whether or not there is sound. In the encoder 6, the signal determined to be silent by the first voiced / silent determining unit 3 or the second voiced / silent determining unit 5 performs a silent compression, and after the echo removal, the second voiced / silent determining unit 5 determines whether or not the signal is present. The signal determined to be sound (therefore, also determined to be voiced by the first voiced / silent determination unit 3) is subjected to voiced compression processing. Since the signals determined to be sound by the first sound / silence determination unit 3 include only echo signals and include signals that are actually soundless, the second sound / silence determination is performed. The determination of the presence or absence of sound is performed again by the determination unit 5 so that the accuracy of the determination of presence or absence of sound is increased. Even if the first presence / absence determination unit 3 determines that there is sound, the second determination is performed. When the sound / silence determination unit 5 determines that there is no sound, the encoder 6 performs the silence compression to increase the compression efficiency and the transmission efficiency.

According to another embodiment of the present invention, there is provided a first sound / silence determining section 3 for judging whether or not an input voice signal is sound, and a decoder 7 for expanding a signal input from a line. An echo canceller unit 4 for performing echo removal processing by referring to an output signal of the decoder from an audio signal determined to be sound by the first voiced / silence determination unit; And an encoder 6 that receives and compresses the audio signal determined to be silent by the first voiced / silent determining unit 3.

As described above, in one embodiment of the present invention, the sound / signal of the signal before input to the echo canceller 4 is output.
A first sound / non-speech determination unit 3 for performing silence judgment is provided, and a signal determined as silence by the sound / non-speech judgment unit is configured to make a path in which no echo cancellation processing is performed. The amount of calculation can be reduced without performing the cancellation process.

In one embodiment of the present invention, the sound / non-speech judging section (VAD5 in FIG. 5) of the encoder 6 is used as the first sound / non-speech judging section 3 for detecting the non-speech / speech of the echo canceller. You may.

[0022]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention; FIG. 1 is a diagram showing the configuration of one embodiment of the present invention. Referring to FIG. 1, one embodiment of the present invention includes a microphone 2 for performing audio input, an analog-to-digital (AD) converter 8 for converting an input (analog input signal) from the microphone 2 into a digital signal, First voiced / silent determination unit (VAD) for determining voiced / silent output of converter 8
3, an echo canceller 4 for removing echoes from audio data determined to be sound by the first VAD 3 with reference to a line signal output to the speaker 1, and a first VAD
A second voiced / silent determination unit (VAD) 5 for performing voiced / silent determination of a signal after echo removal for data including only an echo among voice data determined to be voiced in 3; An encoder 6 for compressing the output of the second VAD 5 or audio data determined to be silent by the first VAD 3, a decoder 7 for expanding a signal input from a line, and converting a digital output of the decoder 7 into an analog signal. Digital
It comprises an analog (DA) converter 9 and a speaker 1 for outputting the output of the DA converter 9 as audio.

Each of the above-described components operates roughly as follows.

The speaker 1 outputs the output of the DA converter 8 as audio. The microphone 2 inputs an audio signal such as a human voice.

The first VAD 3 and the second VAD 5 use a known structure provided in a sound compression process having a silent compression function, which determines whether an input signal is sound or no sound. Note that the VAD3 and VAD5 can be executed without inputting a program capacity by using the same function when inputting a digital audio signal and determining the presence or absence of sound.

The echo canceller 4 is for cutting off the sound coming from the speaker 1 or the like into the sound input to the microphone 2, and is composed of an adaptive digital filter or the like.

The encoder 6 compresses and encodes audio, and reduces the load on the line by compressing the audio signal. In this embodiment, an encoder with a silence compression function is used.

The decoder 7 expands the audio compressed by the same compression method as the encoder 6, and expands the audio compressed by the silent compression function in accordance with the silent compression.

The DA converter 8 converts the audio data (digital signal) expanded by the decoder 7 into an analog signal.
The signal is converted so that it can be output from the speaker 1 as audio.

The AD converter 9 converts an audio signal (analog signal) input to the microphone 2 into a digital signal.

Next, the operation of the embodiment of the present invention will be described in detail with reference to the flowcharts of FIGS.

First, a signal such as voice (step S1) input from the microphone 1 in FIG. 1 is converted from an analog signal to a digital signal by the AD converter 9 so that signal processing such as compression can be performed. The digital signal output from the AD converter 9 is input to the first VAD 3, and a sound / non-sound determination is performed (step S2).

In the first VAD 3, a signal determined to be silent is directly input to the encoder 6 and subjected to a silent compression process (step S3).

On the other hand, the signal determined to be sound by the first VAD 3 is input to the echo canceller 4 and is subjected to echo removal processing (step S3).

In one embodiment of the present invention, the signal from which the echo has been removed by the echo canceller 4 is supplied to the second VAD 5.
(Step S5). That is, the first VA
The signals determined to be sound in D3 only include echo signals, and include signals that are actually silent. Therefore, in the second VAD5, the sound / silence is again determined. Make a decision.

The output signal of the second VAD 5 is input to the encoder 6 irrespective of the sound / non-speech determination result in the second VAD 5, and the signal determined to be silent is subjected to silence compression by the encoder 6. (Step S6) If the sound is determined to be sound, the encoder 6 performs compression processing for sound (step S7).

The decoder 7 expands a compressed signal received from a line or the like, and outputs it to the DA converter 8. The decoder 7 is a decompressor for silent compression.
Automatic extension of voice / silence is performed according to the type of the received signal.

The DA converter 8 converts the received digital signal into an analog signal so that it can be output as a sound, and passes it to the speaker 1.

In one embodiment of the present invention, before performing the echo canceling process, a sound / non-speech determination is performed to detect a silent signal having no echo or voice, and the silent signal is echoed. Since the removal processing is not required, the amount of calculation of the system can be reduced by skipping (passing) the echo removal processing.

In the embodiment of the present invention, the processing of the first VAD 3 is increased as compared with the conventional configuration.
The voice / non-speech VAD is determined by a DSP (Digital Signal Processor) with 2 to 3 MIPS (megainstructions / sec: 1 MIPS executes millions of instructions per second).
On the other hand, the echo canceller process requires a calculation amount (calculation performance) of about 20 MIPS, although it depends on the echo removal time. The effect of reducing the amount of computation by being able to pass the echo cancellation process is more significant,
Processing performance has been improved.

Next, another embodiment of the present invention will be described. FIG. 3 is a diagram showing the configuration of the second embodiment of the present invention. Referring to FIG. 3, a second embodiment of the present invention is shown in FIG.
VAD5 is deleted from the embodiment described with reference to FIG. That is, the microphone 2 for performing voice input,
AD converter 8 that performs AD conversion on the input from microphone 2
And the first V that determines sound / non-sound of the output of the AD converter 8
AD3, an echo canceller 4 that performs echo removal by referring to a signal of a line output to the speaker 1 from audio data determined to be sound by the first VAD3, and an output of the echo canceller 4 or the first VAD3. An encoder 6 for compressing audio data determined to be silent, a decoder 7 for expanding a signal input from the line, and a decoder 7
A DA converter 9 for performing DA conversion of the output of
And a speaker 1 for outputting the output as a sound.

Each of the above elements operates as follows.

The speaker 1 outputs the output of the DA converter 8 as sound. The microphone 2 can input a voice such as a human voice. The VAD 3 determines whether the input signal is sound or silence. Echo canceller 4
Is to cut off the sound coming from a speaker or the like to the sound input to the microphone 2, and is used in an audio communication system such as a telephone.

The encoder 6 compresses audio, and can reduce the load on the line by the compression function. In this embodiment, an encoder without a silence compression function can be used.

The decoder 7 is a means for expanding audio compressed in the same manner as the encoder 5 (the combination of the encoder and the decoder is referred to as a codec), and can be used without a silent compression function. However, even if the encoder has no silence compression function and the decoder has silence compression function,
Since the same decompression process can be performed only without using the decompression function corresponding to the silence compression, the decoder side may have the silence compression function or may not have the silence compression function.

The DA converter 8 performs digital-to-analog conversion of the audio data expanded by the decoder 7, and converts the signal so that the audio data can be output from the speaker 1. The AD converter 9 performs analog-to-digital conversion of the audio signal input to the microphone 2 and converts the signal so that digital processing such as compression / decompression can be performed.

Next, the overall operation of the second embodiment of the present invention will be described in detail with reference to the flowcharts of FIGS.

First, a signal such as audio input from the microphone 1 in FIG. 1 is converted from an analog signal to a digital signal by the AD converter 9 so that signal processing such as compression can be performed. The signal converted into a digital signal by the AD converter 9 is
In the VAD3, a sound / no-sound determination is performed (step S12).

The signal determined to be silent in the VAD 3 is input to the encoder 6 as it is, and compression processing is performed (step S14).

The echo canceller 4 performs an echo removal process on the signal determined to be sound by the VAD 3 (step S13).

The signal after the echo removal processing in the echo canceller 4 is input to the encoder 6, where the signal is subjected to compression processing (step S14) and output to the line (step S15).

The decoder 7 expands a compressed signal received from a line or the like, and outputs it to the DA converter 8. The DA converter 8 converts the received digital signal into an analog signal so that it can be output as a sound, and passes the signal to the speaker 1.

In the embodiment shown in FIG. 1, the encoder 6 and the decoder 7 are configured to have a silence compression function. On the other hand, in the second embodiment of the present invention, even when the silence compression function is provided, It can be realized without the silence compression function.

[0054]

As described above, according to the present invention,
When the input is in a silent state, an effect is obtained that the amount of processing calculations can be reduced.

The reason is that, in the present invention, a sound / non-speech determination is performed before the input signal is input to the echo canceller unit, and when it is determined that there is no sound, the echo cancellation processing is passed, thereby removing the echo. This is because the amount of calculation for processing can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a process according to an embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of another embodiment of the present invention.

FIG. 4 is a flowchart illustrating a process according to another embodiment of the present invention.

FIG. 5 is a diagram showing a configuration of a conventional audio compression / decompression system with an echo canceller.

[Explanation of symbols]

 Reference Signs List 1 speaker 2 microphone 3 VAD (voice existence / non-speech determination unit) 4 key canceller processing unit 5 VAD 6 encoder 7 decoder 8 DA converter 9 AD converter

Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat II (reference) H04M 1/00 G10L 3/00 B H04R 3/02 9/00 F

Claims (6)

[Claims] 1. A first method for determining the presence or absence of sound in an input audio signal.
A voice / silence determining unit, a decoder for expanding a signal input from a line, and a voice signal determined as voiced by the first voice / silence determining unit with reference to an output signal of the decoder. An echo canceller unit that performs an echo removal process; a second voiced / silent determination unit that determines the presence / absence of a signal subjected to the echo removal process in the echo cancellation unit; and an output of the second voiced / silent determination unit And an encoder for inputting and compressing a signal determined to be silent by the first voiced / silent determining unit, and a voice compression / expansion system with an echo canceller. 2. A method according to claim 2, wherein said second voiced / silent determination unit receives the signal after said echo canceling processing by said echo canceller unit and determines whether the second voiced / non-voiced signal is present. An output signal is input to the encoder. In the encoder, a signal determined to be silent by the first or second voiced / silent determination unit performs silent compression, and the first and second voiced / silent determinations are performed. 2. A signal which is determined to be sound by the unit is subjected to a sound compression process.
A voice compression / decompression system with an echo canceller according to the above. 3. A first method for determining the presence or absence of sound in an input audio signal.
A voice / silence determining unit, a decoder for expanding a signal input from a line, and a voice signal determined as voiced by the first voice / silence determining unit with reference to an output signal of the decoder. An echo canceller unit that performs an echo cancellation process, and a signal after the echo cancellation process in the echo canceller unit,
An encoder for compressing an audio signal determined to be silent by the first sound / silence determining unit; and an audio compression / expansion system with an echo canceller. 4. A signal judged as silence by the first sound / silence judging section is inputted to the encoder as it is, and compression processing is performed. The determined signal is input to the encoder after performing an echo removal process in the echo canceller unit, the encoder is an encoder with a silence compression function,
An audio compression / decompression system with an echo canceller, which can be configured by an encoder without a silent compression function. 5. The voice with an echo canceller according to claim 1, wherein a voiced / silent determination unit of said encoder is used as said first voiced / silent determination unit. Compression and expansion system. 6. A signal obtained by converting an analog audio signal input from an audio input means into a digital signal by an analog-to-digital converter is input to the first sound / non-sound determining unit, and an output signal of the decoder is converted to a digital-analog signal. The audio compression / expansion system with an echo canceller according to any one of claims 1 to 4, wherein the signal converted into an analog signal by the converter is supplied to audio output means.