JP2988119B2 - Echo suppression device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reverberation suppressor, and more particularly to a reverberation suppressor used in a digital mobile communication system such as a mobile phone or a mobile phone.

[0002]

2. Description of the Related Art As shown in FIG. 4, an echo suppressor used in a conventional analog mobile communication type mobile telephone mobile unit is an amplifier 44 for adjusting the level of a received signal demodulated by a radio unit 43, and the amplifier 44. The analog-to-digital converter (A / D) 45 for converting the output signal of the analog signal from an analog signal to a digital signal (PCM signal) 45 and the two-to-four-wire converter (HYB) 51 for converting the digitally-converted received signal due to mismatch.
A line echo canceller 22 that suppresses the generated reverberation signal (line echo signal), and a digital-to-analog converter (D) that converts the received signal in which the line echo signal is suppressed by the line echo canceller 22 from a digital signal to an analog signal. / A) 15, an amplifier 16 for adjusting the level of the converted analog signal, and a speaker 17 for inputting the signal of the amplifier 16. Further, a microphone 11 for inputting the voice of the near-end speaker, an amplifier 12 for adjusting the level of the transmission signal output from the microphone 11, and an analog for converting the signal of the amplifier 12 from an analog signal to a digital signal (PCM signal) A digital converter 13, an acoustic echo canceller 14 for suppressing an echo signal (acoustic echo signal) generated by acoustic coupling between the microphone 11 and the speaker 17 from the digitally converted transmission signal, and an acoustic echo canceller 14 A digital-to-analog converter 41 for converting a transmission signal in which a system echo signal is suppressed from a digital signal to an analog signal, and an amplifier 42 for adjusting the level of the analog signal digital-to-analog converted by the digital-to-analog converter 41
And a radio unit 43 for inputting the output signal of the amplifier 42. This radio unit 43 is connected to a base station (not shown) by radio, and furthermore, an exchange (not shown),
It is connected to the other party's subscriber 52 via a two-wire / four-wire converter 51.

[0003] The echo cancellers 14 and 22 are adaptive digital filters (Adaptive Digital Filters) that simulate the characteristics of a transmission path (echo path).
r: hereinafter referred to as ADF) A pseudo echo signal obtained by inputting a reference signal to each of 14a and 22a is added to an adder 14b.
22b is a circuit for suppressing the echo signal by adding the echo signal to the echo signal. The echo path in the case of the acoustic echo canceller 14 is transmitted from the reference signal input terminal of the ADF 14a to the speaker 1.
7, a path up to the addition of the pseudo echo signal including the microphone 11 to the echo signal. The echo path of the line-based echo canceller 22 is a path from the reference signal input terminal of the ADF 22a to the addition of the pseudo echo signal to the reverberation signal, including the radio unit 43 and the 2-wire 4-wire converter 51. Each ADF 14 inside the echo cancellers 14 and 22
Reference numerals a and 22a add the pseudo echo signal generated by the ADF to the echo signal, apply feedback based on the output signal (error signal), and identify the echo path. For example, AD
F updates the filter coefficients by an updating algorithm called a learning identification method, and identifies an echo path.

[0004] This ADF is used when the signal input to the echo canceller is only an echo signal (single talk).
It is possible to sequentially modify the tap coefficients of the filter to generate the same characteristics as the echo path. However, when a signal other than the reverberation signal is input (double talk state), the signal other than the reverberation signal is not deleted, and the converged tap coefficients of the ADF are disturbed by the signal. In order to solve this problem, a double talk detector is required to control so as to stop updating the tap coefficient when a signal other than the echo signal is input. For example, a double talk detector receives a reference signal and a reverberation signal and calculates the power of each. For the detection of double talk, the power value of the reference signal is multiplied by a constant, and the threshold value is compared with the power value of the echo signal. If the comparison result indicates that the threshold value is small, a state (double talk) in which a signal of another speaker is input to the echo signal is detected, and updating of the tap coefficient is stopped.

[0005]

In a mobile telephone for analog mobile communication, a carrier is frequency-modulated with an analog signal and transmitted / received. Therefore, when a digital processing reverberation suppression device is used, the reverberation suppression is performed as described above. An analog-to-digital converter and a digital-to-analog converter are required before and after the device. For example, a PCM codec of a μ-law compression processing method is used for the analog-digital converter and the digital-analog converter. This PCM
The delay due to codec encoding and decoding is 300
It is about μsec. When such an echo suppressor is applied to a mobile telephone of a digital mobile communication system that modulates a carrier wave with a digital signal, a mobile phone of a mobile telephone, and the like, the following problem occurs. That is, in the case of a digital mobile communication system mobile telephone, a low bit rate voice codec is used to perform time division multiplexing.
This audio codec is provided between the line-based echo canceller and the radio unit, and its encoding and decoding processing (64k
The digital-to-digital conversion between the b / s PCM encoded speech signal and the 11.2 kb / s highly efficient encoded speech signal (VSELP) has a delay of about 100 msec.
Therefore, the line echo canceller must suppress the echo signal for 100 msec or more.

For example, assuming that 100 msec is required for the encoding / decoding processing in the voice codec and the impulse response of the two-wire / four-wire converter is 40 msec, the line-based echo canceller needs an ADF for identifying an echo path of 140 msec. Become. 140 kHz signal at 8 KHz
Sampled at 1,120 samples. Therefore, the number of taps of the ADF is 1120 taps. 140m
DSP with echo canceller to suppress reverberation signal
(Digital Signal Processo
In the case of realizing r), a command of 4 steps is required for each tap (one step for generating a pseudo echo signal and three steps for updating tap coefficients), so a processing capability of about 36 MIPS is required. Also, to realize ADF,
2K to hold filter tap coefficients and received signal
Word RAM is required. The actual time required for echo control is 40 m for the impulse response of the 2-wire 4-wire converter.
Since the ADF for 320 seconds is sufficient, the ADF for 320 samples has only to be realized.

In addition, when connected to an international line or a satellite line, the impulse response time of the line-based echo path becomes longer, becomes longer than the tap of the ADF, and the line echo signal cannot be suppressed.

An object of the present invention is to provide an AD of an echo canceller.
It is an object of the present invention to provide a sound suppressor capable of reducing the number of taps of F and shortening the time required to simulate an echo path.

[0009]

A reverberation suppression apparatus according to the present invention is a reverberation suppression apparatus for suppressing a line reverberation signal due to a transmission signal output to a transmission side line among reception signals from a reception side line. A signal is converted from a first code format to a second code format to be a transmission signal to the transmission line, and a digital reception signal from the reception line is converted from the second code format to the first code format. Digital-to-digital conversion means for inverting the digital
The transmission signal of the first code format input to the digital conversion means is delayed by a time corresponding to an externally controlled delay amount.
Delay means for varying the amount of delay for branch output , an adaptive digital filter, and an adder. Each of the taps of the adaptive digital filter is referred to by using the output of the delay means as a reference signal.
Generates appropriate updated estimated echo signal tap coefficients flop, and line-based echo canceling means for suppressing the line system echo signal in the received signal of the first code format output from the digital-to-digital converter , Said line eco
-Each tap of adaptive digital filter of cancellation means
Read the value of the tap coefficient, and based on the situation,
Line-based echo path including digital-to-digital conversion means
Of the fixed delay of the echo signal in the
The delay is set so that the delay amount is equal to the delay amount of the delay means.
Controlling the delay means and the delay means and the line
System echo canceling means to suppress the line echo signal.
Control signal indicating whether or not pressure is possible,
Output from the delay control means,
When the control signal indicates whether or not the line echo signal is suppressed.
Is output from the digital-to-digital conversion means.
The received signal of the first code format is
Outputs the signal by bypassing the cancel means, and
Between the input reference signal of the
Based on the level comparison result, it is determined whether or not the
In the middle case, a level loss is given to this diverted received signal.
Means for suppressing the line system echo signal .

[0010]

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

FIG. 1 shows the operation of an echo suppressor to which the present invention is applied .
FIG. 5 is a block diagram for explaining the operation , and portions equivalent to those in FIG. 4 are denoted by the same reference numerals. The radio unit 24 transmits and receives signals to and from a base station (not shown) by radio, and is connected to a subscriber 52 via a switching center (not shown) and a two-wire / four-wire converter 51. Audio codec 23 1 from the radio unit 24
64 kb / s by inputting 1.2 kb / s VSELP signal
The signal is decoded into a PCM audio signal and output. The audio signal output from the audio codec 23 is input to a line echo canceller 22 that suppresses a line echo signal. The output signal of the line echo canceller 22 is input to the acoustic echo canceller 14 and the digital / analog converter 15. The digital / analog converter 15 converts the output signal of the line echo canceller 22 from a digital signal to an analog signal. The signal output from the digital / analog converter 15 has its level adjusted by the amplifier 16 and is output to the speaker 17. The signal from the microphone 11 is input to the amplifier 12, the level of which is adjusted, and input to the analog / digital converter 13. The analog / digital converter 13 converts the analog signal into a digital signal (64 k
b / s PCM audio signal). The signal digitally converted by the analog-to-digital converter 13 is input to an acoustic echo canceller 14, and an acoustic echo signal is suppressed. The output signal of the acoustic echo canceller 14 is input to the audio codec 23 and the delay circuit 21. The line echo canceller 22 includes the acoustic echo canceller 14.
Is input via a delay circuit 21 for realizing a delay corresponding to the encoding and decoding processing time of the audio codec 23. The audio codec 23 is an acoustic echo canceller 1
A 64kb / sPCM signal from the 4 11.2 k b / sV
The signal is encoded into a SELP signal and input to the radio unit 24.

The operation of the acoustic echo canceller 14 will be described. For example, when the acoustic reverberation signal has a reverberation of 40 msec, the ADF 14a needs the FIR filter of 320 taps as described above. The pseudo echo signal generated by the ADF 14a is added to the echo signal by the adder 14b to suppress the echo signal. The tap coefficient of the ADF 14a is
It is updated by an algorithm called a learning identification method of Expression 1.

[0014]

(Equation 1)

Here, H _n is the ADF tap coefficient row vector at time n, X _n is the reference signal row vector at time n, e
_n is a cancellation error signal of the echo signal at time n. α is called a convergence coefficient and is a value in the range of 0 <α <2. T represents the transposition of the vector. The tap coefficients of the ADF 14a are sequentially updated by this learning identification method, and the characteristics of the echo path are identified. The echo path referred to here is a digital / analog converter 15, an amplifier 16, a speaker 17, an acoustic coupling between the speaker 17 and the microphone 11, the microphone 11, the amplifier 12, and the analog / digital converter 13 This is the acoustic signal propagation path of the acoustic system consisting of:

Similarly, the line-based echo canceller 22 identifies a line-based echo path including the voice codec 23, the radio unit 24, a network such as a base station and an exchange, and a two-wire / four-wire switch 51. The delay time of the echo signal due to the encoding and decoding processing time of the audio codec 23 in the echo path is a fixed time of about 100 ms, and the delay circuit 21 delays the output signal of the acoustic echo canceller 14 by the delay time. Line system echo canceller 22
To the reference signal input terminal 22c. This delay circuit 2
1, the ADF 22a of the line-based echo canceller 22 has the impulse response time (impulse response time ( For example, 4
The echo path can be simulated with 320 taps corresponding to 0 ms), and the time required for simulating can be reduced. If no delay circuit is inserted, an ADF with 1120 taps is required. The simulation of the echo path is repeated at a predetermined cycle.

Incidentally, the radio unit 24 and the two-wire / four-wire converter 5
The time at which the reverberation of the line between lines 1 and 2 occurs differs greatly depending on the type of line, such as an international line or a domestic line.
In the configuration shown in (1), the number of ADF taps may need to be increased. In order to cope with such fluctuations in the echo time of the line system, a delay control circuit 27 is newly provided as shown in the block diagram of FIG. ADF
The control processing of the tap position and coefficient of 26a is performed to estimate the propagation delay amount of the line system, and the delay amount of the variable delay circuit 25 is controlled to be equal to the estimated amount. Other circuit configurations are the same as those in FIG.

Next, the flow of the process of estimating the amount of delay by the delay control circuit 27 will be described with reference to the flowchart of FIG. In addition,
In the delay circuit 25, a delay amount corresponding to the signal processing time of the audio codec 23 is set in advance. (1) Set an initial value. That is, the estimated delay amount D =
0, the delay correction amount Dh = 0, and the position Imax = 0 where the absolute value of the tap coefficient is maximum (step 101). (2) After the ADF is updated once, a tap coefficient is fetched (step 102). (However, α1 is designated by the delay control circuit for the convergence coefficient α of the ADF). (3) The above-mentioned (2) until the coefficient of the same tap among the captured tap coefficients is successively detected as the maximum absolute value P times.
The processing of the term is repeated (step 103). (4) The position of the maximum absolute value of the tap coefficient detected P times consecutively is defined as Imax. (5) Let Imax-L be the delay correction amount Dh (step 105). L is a margin for when Imax fluctuates. (6) The delay correction amount Dh is added to the estimated delay amount D (D
= D + Dh). If the value of D when the correction amount Dh is added becomes negative, D and Dh are set to 0 (step 10).
6). (7) After the ADF updates the tap coefficient Q times, the tap coefficient is fetched (step 107). (However, α2 is designated by the delay control circuit for the convergence coefficient α of the ADF). (8) Im the position where the acquired tap coefficient has the maximum absolute value
ax (step 108). (9) The processing of the above items (5) to (8) is repeated until Imax reaches the same position R times (step 109). Here, each variable uses, for example, the following numerical value. α
1 = 1.0, α2 = 0.25, P = 20, L = 5, Q =
10, R = 20 Estimated delay amount D estimated by the above procedure and voice codec 2
The delay circuit 25 is controlled so that the value added to the delay amount generated in the encoding / decoding processing of No. 3 is equal to the delay amount of the delay circuit 25. When changing the delay amount, the tap coefficient and the reference signal of the echo canceller 26 are shifted by the delay correction amount. If the correction amount Dh is negative, the tap coefficient and the reference signal are shifted, and then Dh is cleared to 0 from the beginning. Note that the amount of delay changes when the line switches,
The processing of the above items (5) to (9) is performed periodically.

As described above, the line-based echo canceller 26
When simulating a line system echo path, the fixed delay can be removed and only the impulse response needs to be processed, so that the time required for simulating can be reduced.

The delay amount realized by the delay circuit and the ADF
In the case of an impulse response in an international line or a satellite line that exceeds the impulse response that can be simulated by the above, the line-based echo canceller cannot accurately simulate the echo path and cannot suppress the echo signal. Therefore, an embodiment of the present invention
As shown in FIG. 3, a delay control circuit 31 for determining whether or not the impulse response time of the line echo path exceeds the echo controllable range from the value and the position of the tap coefficient of the line echo canceller 26, and the delay control circuit 31 The output signal of the voice codec 23 is diverted from the line-based echo canceller 26 in accordance with the determination result of
Is higher than the level of the output signal of the audio codec 23 (when transmitting),
(A selector circuit 32, a variable gain amplifier 33, and a gain control circuit 34) for giving a level loss to the output signal of (1) and suppressing an echo signal. Other circuit configurations are the same as those in FIG.

In FIG. 3, the delay control circuit 31 has the same function as the delay control circuit 27 of FIG. 2, and furthermore, the maximum value of the tap coefficient is within 5 taps from the last tap of the ADF 26a of the line echo canceller 26. If detected,
That is, when the range exceeds the range in which the echo can be controlled by the line-system echo canceller 26, the selection control signal SS is set to logic "1".
Otherwise, it is set to logic "0" and output to the selector circuit 32. When the selection control signal SS is logic “0”, the selector circuit 32
When the signal SS is logic "1", the signal SA obtained by adjusting the output signal SR of the audio codec 23 by the variable gain amplifier 33 is selected, and the digital / analog converter 15 and the sound are output. System echo canceller 1
Output to 4. The gain control circuit 34 controls the level of the output signal ST of the delay circuit 25 and the output signal SR of the audio codec 23.
, And when the level of the signal ST is equal to or higher than the level of the signal SR, it is determined that transmission is in progress, and the variable gain amplifier 3
3, the signal SR is attenuated by a predetermined amount, and the signal SR
If the level is high, it is determined that a call is being received and the variable amplifier 3
The gain of No. 3 is set to 0 db, and the signal SR is passed as it is.
With this configuration, when the signal exceeds the range that can be controlled by the line-based echo canceller connected to the international line or the satellite line, the output signal of the variable gain amplifier is selected, and the echo signal is suppressed by the gain control.

[0022]

As described above, according to the present invention, a connection to a line is established.
Transmission signal to the following digital-to-digital conversion means
Delay means for variably delaying the delay, and an output of the delay means
Is used as a reference signal for each tap of the adaptive digital filter.
Digital / digital while updating tap coefficients appropriately
Suppress line-system echo signals in the received signal from the conversion means
Line echo canceling means and line echo cancel
Tap of each tap of adaptive digital filter
Digital-to-digital conversion based on number value situations
Fixed delay of reverberation signal in circuit echo path
Estimate delay amount and control delay means so that delay amount is equal
And echo by the line-based echo canceling means
Delay for outputting control signal indicating whether signal suppression is possible
The control signal from the control means and the delay control means
To indicate whether or not suppression is performed, the received signal is
Output while bypassing the
Level comparison between input reference signal of cancel means and received signal
If it is determined that a call is being made based on the result, this bypass
A method for suppressing the reverberation signal by giving a level loss to the received signal.
And a digital-to-digital converter.
Even if the delay amount of the stage or line is large, the delay amount is delayed.
In the echo cancellation means of the line system
-If you can shorten the time to simulate the pass
The impulse response of the line echo path
Of adaptive digital filter for line echo cancellation
Longer than the number of taps
Even if the signal cannot be suppressed,
The reverberation signal can be suppressed by giving a loss, and the call quality
Can be improved .

[0023]

[0024]

[Brief description of the drawings]

FIG. 1 illustrates the operation of an echo suppressor to which the present invention is applied.
It is a block diagram of the order.

FIG. 2 illustrates the operation of an echo suppressor to which the present invention is applied.
It is a block diagram of the order.

FIG. 3 is a block diagram of an embodiment of the present invention.

FIG. 4 is a block diagram of a conventional echo suppression device.

5 is a flow diagram of the tap position estimation operation of the fixed delay of the time line based echo canceller.

[Explanation of symbols]

 Reference Signs List 11 microphone 12, 16, 42, 44 amplifier 13, 45 analog-digital converter (A / D) 14 acoustic echo canceller 14a adder 14b ADF (adaptive digital filter) 15, 41 digital-analog converter (D / A) 17 speaker 21, 25 delay circuit 22, 26 line-system echo canceller 22 a, 26 a adder 22 b, 26 b ADF (adaptive digital filter) 23 voice codec 24, 43 radio unit 27, 31 delay control circuit 32 selector circuit 33 variable gain amplifier 34 Gain control circuit 51 2-wire 4-wire converter (HYB) 52 Other subscriber

Continuation of the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) H04B 1/76-3/44 H04M 1/60

Claims (1)

(57) [Claims] A reverberation suppression apparatus for suppressing a line-based reverberation signal due to a transmission signal output to a transmission side line in a reception signal from a reception side line, wherein a digital transmission signal is converted from a first code format to a second code. A digital signal for converting a digital received signal from the receiving line from the second code format to the first code format by converting the received signal into a signal to be transmitted to the transmitting line.
Digital conversion means, and an externally controlled delay for transmitting the transmission signal of the first code format input to the digital / digital conversion means.
It has a variable delay of the delay means for delaying the amount worth of time only branch output, the adaptive digital filter and an adder, the adaptive digital filter output of said delay means as a reference signal
Update the tap coefficient of each tap as needed to generate a pseudo echo signal.
Forms, a line-based echo canceling means for suppressing the line system echo signal in the received signal of the first code format output from the digital-to-digital converter, the adaptive digital Fi of the line-based echo canceling means
Read the tap coefficient value of each tap of the filter
Based on the digital-to-digital conversion means
Estimation of fixed delay amount of echo signal in circuit echo path
The estimated delay amount is not equal to the delay amount of the delay means.
Controlling the delay means so that
The line by means of a stage and the line echo canceling means
It is determined whether or not the system echo signal can be suppressed.
And delay control means for outputting a control signal indicating, said control signal said circuit system echo from said delay control means
To indicate whether or not to suppress the signal, use the digital
Received signal of the first code format output from the
Output by bypassing the line-based echo canceling means
And input to the line-based echo canceling means.
Speak based on level comparison result between reference signal and received signal
Judge whether it is in the middle or not.
A level loss is given to the signal to suppress the line-based echo signal.
Means for performing the echo suppression.