JP2009218849A - Acoustic echo cancel control method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve accuracy in discriminating a reception single-talk state and a double-talk state, when performing a hands-free talk that accompanies acoustic echoes. <P>SOLUTION: A sub echo canceler SEC and a main echo canceler MEC are provided which operate in accordance with an instruction from an adaptive operation control unit. In the sub echo canceler SEC, an echo cancel error signal bs required for deciding double talk is outputted; and since this is not sent to a counter terminal side but is sent to the adaptive operation control unit so as to precisely detect a double-talk state and is used only for adaptively controlling the main echo canceler MEC during a double talk, a speech quality during the double talk in a hands-free speech is improved. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an acoustic echo cancellation control method and apparatus. Specifically, an object of the present invention is to provide a novel control method and apparatus for acoustic echo cancellation in a communication device capable of hands-free communication in which acoustic echo due to acoustic wrapping is likely to occur.

  FIG. 5 is a loop gain diagram for explaining the conventional loop gain. When a hands-free call is made using the communication device shown here, an acoustic echo is generated due to an acoustic wraparound, and in order to prevent howling, the loop gain formed by reverberation or the like is set to 0 dB or less. It is necessary to design to.

  When only the 4-wire system shown in the figure without a 2-wire to 4-wire conversion hybrid circuit is used, the receiver terminal or speaker R of the partner terminal and the loop from the transmitter or microphone T and the speaker S of the terminal itself are used. And the loop gain formed by the wraparound by the acoustic path between the microphone M and the microphone M is set to 0 dB or less. At this time, considering that the partner terminal may also make a hands-free call, the gain from the input terminal TM1 to the output terminal TM2 of the terminal needs to be 0 dB or less. Here, it is assumed that the acoustic wraparound attenuation amount due to the acoustic path between the speaker S and the microphone M is ZdB.

  FIG. 6 is a call route diagram for explaining a call route of a conventional example. The own terminal is used as a terminal of a conference device or a button telephone device capable of hands-free communication. The amplifiers 53 and 57 are inserted so that the speaker can talk even if the speaker is about 0.5 to 1 m away from the speaker S or microphone M of the terminal. Similarly, the amplifiers 54 and 56 are inserted on the partner terminal side. Between the own terminal and the partner terminal, a hybrid circuit that performs two-wire to four-wire conversion (not shown) is interposed in each of the own terminal and the partner terminal.

  Here, assuming that the transmission system gain of the amplifier 53 is YdB, the reception system gain of the amplifier 57 is XdB, and the acoustic wraparound attenuation is ZdB, generally, if the echo canceling device 51 does not exist, the gain of the own terminal X + Y-Z becomes larger than 0 dB. The same applies to the counterpart terminal side. When such a terminal and the partner terminal are connected, the loop gain described with reference to FIG. 5 becomes larger than 0 dB, so that howling occurs and communication is impossible.

  Therefore, in order to enable a hands-free call, it is necessary to mount an echo cancellation device 51 for eliminating acoustic echo. When the echo canceling amount of the echo canceling device 51 is α dB, it is designed to satisfy X + Y−Z−α <0. The acoustic wraparound attenuation amount ZdB of the terminal itself and the maximum echo cancellation amount αdB of the echo cancellation device 51 are obtained by measurement. Ultimately, it is necessary to determine what value each of the transmission system gain YdB and the reception system gain XdB should be.

  FIG. 7 is a circuit configuration diagram of an echo cancellation apparatus 51 for canceling acoustic echoes according to a conventional example. There is an echo canceller EC which is instructed by an adaptive operation instruction signal A from the adaptive operation control unit 10B and performs an adaptive operation. The echo canceller EC has an adaptive filter 11B and an adder 16. The adder 16 adds the pre-echo cancel (EC) transmission signal y and the negative-polarity pseudo echo signal y 'which is the output of the adaptive filter 11B. That is, the adder 16 performs subtraction of y-y 'and operates as subtraction means for outputting the transmission signal b (= y-y') after echo cancellation (EC).

  FIG. 8 is a circuit configuration diagram showing details of the internal configuration of adaptive operation control unit 10B, which is a component of FIG. 7 showing the conventional example. The adaptive operation control unit 10B includes a power calculation unit 21B, an acoustic coupling amount calculation unit 23, a received sound determination unit 24, a transmitted sound determination unit 25, and an echo canceller adaptation permission determination unit 26.

  When the reception signal x, the transmission signal y before echo cancellation (EC), and the transmission signal b after echo cancellation (EC) are applied to the power calculation unit 21B, each power is, for example, 8 kHz sampling Is always calculated at intervals of 125 μs, and the reception signal power Px, the transmission signal power Py before the echo cancellation (EC), and the transmission signal power Pb after the echo cancellation (EC) are obtained.

  To the acoustic coupling amount calculator 23, received signal power Px, transmission signal power Py before echo cancellation (EC), and transmission signal power Pb after echo cancellation (EC) are applied. There, the acoustic coupling amount 36 is obtained by calculating the acoustic coupling amount by the acoustic echo that circulates from the speaker S to the microphone M. That is, the acoustic coupling amount 36 represents how much the acoustic echo from the speaker S based on the received signal power Px is coupled to the voice signal of the sender input to the microphone M.

  The transmitted sound determination unit 25 receives the received signal power Px and the acoustic coupling amount 36, determines the voice signal of the transmitter, and sets the transmitted signal power Pb after echo cancellation (EC) to be transmitted in advance. When it is larger than the transmitted voice presence threshold value Thb (Pb> Thb), it is determined that there is a voice transmission, and a transmission voice determination signal 37 is output.

  Here, the transmitted sound threshold value Thb will be described. The acoustic echo of the reception signal x output from the speaker S wraps around the microphone M, which causes a false determination that the transmission signal is a speech signal of a voice signal by the speaker. Therefore, the transmission / reception ratio (Py / Px) between the transmission signal power Py before echo cancellation (EC) and the reception signal power Px during the reception single talk, and the transmission signal power Pb after echo cancellation (EC). Echo cancellation (EC) Multiplication value ((Py / Px) × (Pb / Py)) with echo cancellation amount (Pb / Py), which is a ratio to transmission signal power Py before transmission, is obtained as acoustic coupling amount 36 Yes.

  The value of the acoustic coupling amount 36 added with a margin is multiplied by the reception signal power Px to determine the transmission sound as the transmission sound threshold Thb. For example, if the echo canceller EC is not converged after the power is turned on or immediately after the start of the call, the transmission voice threshold value Thb is set large, and if it is converged, it is set small. Yes.

  When setting the transmission voice threshold value Thb, if almost no margin is added, it is possible to accurately determine the transmission voice due to the speaker signal, so the state of adaptive stop is detected. Sensitivity to be increased. However, when used in an environment where the wraparound attenuation is smaller than the assumed environment, or when the characteristics of the acoustic echo path change due to environmental changes and the amount of acoustic coupling becomes large, the acoustic echo signal is mistakenly regarded as transmitted speech. The so-called blocking phenomenon occurs in which the adaptive operation of the echo canceller EC stops despite the received single talk requiring the adaptive operation.

  On the other hand, if a large margin is added when setting the value of the transmit voice threshold, the blocking phenomenon due to the received single talk can be suppressed, but the level of the speaker voice is low. Echo canceller EC is used in cases where the level is low or where the level is relatively low, such as consonants, because it is determined that the transmission is silent, and even during double talk, it is misjudged as received single talk. Will be disturbed.

  8 receives the received signal power Px, and determines that the received sound is received when the received signal power Px is larger than a preset received sound threshold value Thx (Px> Thx). Then, an incoming / outgoing voice determination signal 38 is output. Here, the received sound threshold value Thx is set to a substantially fixed level that does not determine that the noise level is sound.

  The echo canceller adaptive permission determination unit 26 that has received the received sound determination signal 38 and the transmitted sound determination signal 37 is an adaptive operation instruction signal that permits an adaptive operation when the received sound is active (transmission silence). A is output to cause the echo canceller EC to execute the adaptive process, otherwise, the adaptive process is stopped and the filter coefficient H (n) of the adaptive filter 11B is held as it is.

  The echo canceller EC of FIG. 7 includes an adaptive filter 11B and an adder 16 that performs subtraction processing. The adaptive filter 11B receives the echo cancel (EC) post-transmission signal b and the reception signal x which are error signals (= y−y ′), and the pseudo echo signal y ′ is generated according to the instruction of the adaptive operation instruction signal A. It is created by a known method. The adder 16 that performs subtraction processing is applied with the echo cancel signal (EC) before the echo cancel (EC) y and the pseudo echo signal y ′ from the microphone M, and the pseudo echo signal y from the echo cancel (EC) signal before the echo cancel (EC). 'Is subtracted, and an echo cancel (EC) post-transmission signal b which is an error signal (= y-y') is output. The transmitted signal y before echo cancellation (EC) includes the received signal x as an acoustic echo component that circulates as sound from the speaker S.

FIG. 9 shows a circuit of a conventional adaptive filter 11B with N taps. There, a received signal x (n) is applied at time n, and a pseudo echo signal y ′ (n) is output. There are N−1 delay units Z ⁻¹ , an adder d that performs a subtraction action, and N tap amplifiers a. In order to perform adaptive operation with high accuracy, it is necessary to increase the value of the number of taps N There is. The value of the tap number N is N = 800 to 1200 in the adaptive filter 11B of FIG.

The filter coefficients represent the characteristic of the adaptive filter H (n) is the tap amplifier a ₀ ~a _N-1 of the amplification degree _{h 0 (n) ~h N-} 1 determined by the (n). If this filter coefficient H (n) is expressed as a determinant, it can be expressed as follows.
H (n) = [h ₀ (n), h ₁ (n), h ₂ (n),..., H _N−2 (n), h _N−1 (n)]

  FIG. 10 shows an adaptive operation of the echo canceller EC according to an instruction of the adaptive operation control unit 10B, which is a component of FIG. When the transmission and reception are silent, the adaptive operation is stopped. In the case of a single talk where the transmission is silent and the reception is only voiced, an adaptive operation is performed. The adaptive operation is stopped at the time of a single talk where the incoming call is silent and the voice is only sent. That is, the problem of malfunction does not occur because the adaptive operation is stopped when both transmission and reception are silent and during transmission single talk.

In Patent Document 1, in a hands-free loudspeaker, in order to accurately detect the double talk state, the accuracy of echo path estimation is deteriorated based on the fact that adaptive control has been performed in the double talk state. This prevents the residual echo (Patent Document 1, paragraphs 0001 and 0009). The first echo cancellation unit 11 in which the filter coefficient for canceling the echo through the direct propagation path is fixedly set according to the direct propagation path, and the output signal of the first echo cancellation unit 11 is used Double talk detector 12 that detects the talk state, and second echo canceler 13 that cancels the echo that has passed through the indirect propagation path from the output signal of the first echo canceler 11, and its filter coefficients are adaptive The coefficient update is stopped while the double talk detection unit 12 detects the double talk state (Patent Document 1, paragraph 0010).
JP 5-48547 A Japanese Patent Application No. 2006-111424

  When setting the value of the threshold value for transmitting voice, if there is almost no margin, it is possible to accurately execute the determination of the transmitting voice based on the speaker signal that the speaker inputs into the microphone. This increases the sensitivity to adaptive stoppage. However, the acoustic echo signal is smaller when the wraparound attenuation is smaller than the assumed environment, that is, when it is used in a large acoustic echo environment, or when the acoustic echo path characteristics change depending on the environment and the acoustic echo becomes large. Is determined as a transmission signal, and sometimes it is erroneously determined as a transmission sound. In such a case, a blocking phenomenon occurs in which the adaptive operation of the echo canceller stops despite the reception single talk with no speaker signal.

  On the other hand, if a large margin is added when setting the value of the transmit voice threshold, the blocking phenomenon due to the received single talk can be suppressed, but the level of the speaker voice is low. If it is small or a part with a relatively low level such as a consonant, it is determined that the transmission is silent, and even during double talk, the adaptive operation is performed while misjudging as received single talk. Adaptive operation will be disturbed.

  Therefore, the problem to be solved by the present invention is that when the terminal having an echo canceller makes a call using a hands-free call device and the other terminal also makes a call using a hands-free call device or a handset device at the time of a call with acoustic echo Even when the usage environment and acoustic echo path characteristics change, double talk is accurately detected, no blocking phenomenon occurs, and adaptive operation with minimal disturbance of the echo canceller is performed.

  In Patent Document 1, the filter coefficient for canceling the echo that has passed through the direct propagation path is set to be fixed according to the direct propagation path, or the first echo cancellation unit that adapts to gradual characteristic fluctuations such as aging of the direct propagation path And a double talk detection unit that detects a double talk state using the output signal of the first echo cancellation unit, and a second unit that cancels the echo that has passed through the indirect propagation path from the output signal of the first echo cancellation unit. In the second echo cancellation unit, the filter coefficient is adaptively updated, and the coefficient update is stopped while the double talk detection unit detects the double talk state.

  In Patent Document 1, the filter coefficient of the first echo canceling unit is basically fixedly set in accordance with the direct propagation path, including the case of adapting to gradual characteristic fluctuations such as aging. In addition, in the double talk detection unit, the double talk determination threshold for the reverberation signal level ratio between the received signal and the direct sound is reduced on the assumption that the direct echo component is eliminated by the first echo cancellation unit. It is set. Hands-free phone keys and touch panel operations allow the arm to block the direct propagation path between the speaker and microphone, or place the phone itself near a wall, or indirect propagation in an environment where the installation on the desk is nearby The influence of the path cannot be ignored, the reverberation signal increases, and the double talk control cannot be performed properly.

In response to the instruction of the sub-adaptive operation instruction signal, a sub-pseudo echo signal is created in order to remove the acoustic echo that wraps around the microphone from the speaker that converts the received signal into speech, and the transmission signal before echo cancellation from the microphone side A sub echo canceller for subtracting the sub pseudo echo signal from the signal to obtain an echo cancel error signal,
When an echo cancellation error signal, a pre-echo cancel transmission signal and a reception signal are received, and the reception signal power is smaller than the predetermined reception voice threshold, the sub-adaptive operation instruction signal is set and is large Next, the sub-adaptive operation instruction signal is cleared, the echo cancellation amount in the sub-echo canceller is calculated from the echo cancellation error signal and the transmission signal before echo cancellation, and the echo cancellation amount is less than the predetermined cancellation amount threshold value. When large, the main adaptive operation instruction signal is set, and when small, the adaptive operation control unit for clearing the main adaptive operation instruction signal;
In response to the instruction of the main adaptive operation instruction signal, a main pseudo echo signal is created from the received signal, the main pseudo echo signal is subtracted from the pre-echo cancel transmission signal from the microphone side, and the post-echo cancel transmission signal A main echo canceller for creating and sending

  The echo cancellation amount of the sub-echo canceller distinguishes between received single talk and double talk, and avoids the situation where blocking does not occur during received single talk and the blocking phenomenon occurs. Detect double talk. Since the adaptive operation control unit reflects this in the adaptive operation instruction of the main echo canceller, even if the adaptive operation of the sub echo canceller is disturbed, as long as the operation does not diverge, the main echo canceller It has become possible to minimize disturbance of the filter coefficient of the main echo canceller without directly affecting the adaptive operation of the canceller.

  Since the adaptive operation of the sub-echo canceller does not directly affect the adaptive operation of the main echo canceller, the adaptive filter coefficient of the sub-echo canceller is increased so that the adaptive operation does not diverge and double -Talk detection sensitivity and operating speed can be increased. Even when the partner terminal makes a call using a hands-free call device or a handset call device, the bidirectionality is improved and the call quality in the hands-free call is improved.

In response to the instruction of the sub-adaptive operation instruction signal, a sub-pseudo echo signal is created in order to remove the acoustic echo that wraps around the microphone from the speaker that converts the received signal into speech, and the transmission signal before echo cancellation from the microphone side A sub echo canceller for subtracting the sub pseudo echo signal from the signal to obtain an echo cancel error signal,
When the echo response error signal, the pre-echo cancel transmission signal and the reception signal are received, and the reception signal power is greater than the predetermined reception voice threshold, the sub-adaptive operation instruction signal is set. Next, the sub-adaptive operation instruction signal is cleared, the echo cancellation amount in the sub-echo canceller is calculated from the echo cancellation error signal and the transmission signal before echo cancellation, and the echo cancellation amount is less than the predetermined cancellation amount threshold value. An adaptive operation control unit for setting the main adaptive operation instruction signal when it is small, and clearing the main adaptive operation instruction signal when it is large;
In response to the instruction of the main adaptive operation instruction signal, a main pseudo echo signal is created from the received signal, the main pseudo echo signal is subtracted from the pre-echo cancel transmission signal from the microphone side, and the post-echo cancel transmission signal And a main echo canceller for creating and transmitting

  FIG. 1 is a circuit configuration diagram of an echo canceling apparatus showing an embodiment of the present invention. Corresponding symbols are attached to those corresponding to the components shown in FIG. There are a sub-echo canceller SEC and a main-echo canceller MEC that are instructed by the sub-adaptive operation instruction signal As and the main adaptive operation instruction signal Am from the adaptive operation control unit 10, respectively, and perform an adaptive operation.

  The main echo canceller MEC includes an adaptive filter 11 and an adder 16. The adaptive filter 11 and the adder 16 are the same as those shown in FIGS. The adder 16 adds the echo cancel (EC) pre-transmission signal y and the main pseudo echo signal y'm having a negative polarity which is the output of the adaptive filter 11. That is, the adder 16 performs subtraction of y-y 'and operates as subtraction means for outputting the transmission signal bm (= y-y'm) after echo cancellation (EC). The adaptive filter 11 is applied with an echo signal after echo cancellation (EC) bm and a reception signal x, which are error signals, and a main pseudo echo signal y′m is generated by a known method in accordance with the instruction of the main adaptive operation instruction signal Am. is doing.

  The sub echo canceller SEC includes an adaptive filter 12 and an adder 17. The adder 17 adds the echo cancellation (EC) pre-transmission signal y and the negative polarity pseudo echo signal y's which is the output of the adaptive filter 12. That is, the adder 17 operates as a subtracting unit that performs subtraction of y-y's and outputs an echo cancel (EC) error signal bs (= y-y's).

  An echo cancel (EC) error signal bs and a received signal x are applied to the adaptive filter 12, and a sub pseudo echo signal y's is generated by a known method in accordance with the instruction of the sub adaptive operation instruction signal As. The echo cancel (EC) pre-transmission signal y includes an acoustic echo component in which the reception signal x turns around as sound from the speaker S. The echo cancellation (EC) error signal bs has a small value during single talk and a large value during double talk if echo cancellation is sufficient.

  The adaptive operation control unit 10 is applied with the received signal x, the pre-echo cancel (EC) transmission signal y, and the echo cancel (EC) error signal bs, and the main adaptive operation instruction signal Am and the sub adaptive operation instruction signal. As is output.

  FIG. 2 is an adaptive operation diagram of the sub echo canceller according to the instruction of the sub adaptive operation instruction signal As from the adaptive operation control unit 10. This corresponds to the conventional example of FIG. When the transmission and reception are silent, the adaptive operation is stopped. In the case of a single talk where the transmission is silent and the reception is only voiced, an adaptive operation is performed. The adaptive operation is stopped at the time of a single talk where the incoming call is silent and the voice is only sent. What is different from the conventional example of FIG. 10 is that both transmission and reception are performed during a double talk indicated by a thick frame with sound. In other words, it is characterized in that an adaptive operation is performed even during double talk.

  FIG. 3 is a circuit configuration diagram showing details of the internal configuration of the adaptive operation control unit 10. As shown in FIG. This includes a power calculation unit 21, an echo cancellation amount calculation unit 27, a sub-echo canceller (EC) received voice presence determination unit 28, and a main echo canceller adaptive permission determination unit 29. The adaptive operation control unit 10 is applied with the received signal x, the pre-echo cancel (EC) transmission signal y, and the echo cancellation error signal bs, and outputs the sub adaptive operation instruction signal As and the main adaptive operation instruction signal Am. The

  A reception signal x, an echo cancellation (EC) pre-transmission signal y, and an echo cancellation error signal bs are applied to the power calculation unit 21, and the reception signal power Px and echo cancellation (EC) pre-transmission are the respective powers. The speech signal power Py and echo cancellation error signal power Pbs are output.

  The echo cancellation amount calculator 27 receives the echo cancellation (EC) pre-transmission signal power Py and the echo cancellation error signal power Pbs to obtain the echo cancellation amount Ers of the sub echo canceller (SEC). . The main echo canceller adaptation permission determination unit 29 receives the sub-adaptive operation instruction signal As indicating the received voice and the echo cancellation amount Ers of the sub-echo canceller (SEC), and receives the sub-echo canceller (SEC). Based on the value of the echo cancellation amount Ers, the main adaptive operation instruction signal Am for discriminating whether the received single talk or the double talk is used is output.

  The sub-echo canceller SEC is not directly aimed at removing acoustic echoes, but outputs an echo cancellation error signal bs used to distinguish between received single talk and double talk. Based on this, the echo cancellation amount calculation unit 27 calculates the echo cancellation amount Ers of the sub echo canceller (SEC).

  The tap number N (see FIG. 9) of the adaptive filter 12 of the sub-echo canceller SEC may be a value that can remove the acoustic echo due to the direct propagation path, so the number of taps of the adaptive filter 11 of the main echo canceller MEC. Since a value considerably smaller than N is sufficient, it can be realized with a small number of elements. The number of taps N of the sub-echo canceller SEC is greater than the number of taps N required for adaptive operation with sufficient accuracy in each of the first and second echo cancellation units of Patent Document 1. The value of is fairly small and may be a number between 10 and 20%.

  The echo cancellation error signal bs, which is the output of the sub echo canceller SEC, is a signal used to calculate the echo cancellation amount Ers of the sub echo canceller (SEC), and is used only in the adaptive operation control unit 10. Therefore, the convergence speed of the filter coefficient H (n) (refer to the description of FIG. 9) representing the characteristics of the adaptive filter 12 is set to a large value and is not used as a transmission signal to the partner terminal. Sensitivity and speed for detecting whether the talk is single talk or double talk can be increased. Therefore, even when the received single talk is in progress, it is possible to detect the transmitted voice, and the main echo canceller MEC can perform the adaptive operation for the double talk.

  FIG. 4 is a flowchart showing an operation flow of the adaptive operation control unit 10. When the operation of the adaptive motion control unit 10 is started, the sub-adaptive motion instruction signal As representing the reception voice is 1 (As = 1), and the echo cancellation error signal bs indicating a large value and the echo when the reception voice is present. Cancel (EC) When the value Pbs / Py (erase amount value) of the transmission signal power Py before transmission is smaller than a preset erase amount threshold value Thy (Pbs / Py ≦ Thy) (S1Y), the main In order to instruct the echo canceller MEC to perform an adaptive operation, the main adaptive operation instruction signal Am is set to 1 (S2, S3).

  When the value of the erasure amount value (Pbs / Py) is not smaller than the preset erasure amount threshold value Thy (S1N), in order to instruct the main echo canceller MEC to stop the adaptation operation, the main adaptation The operation instruction signal Am is cleared to 0 (S4). When the received signal power Px is larger than the preset received voice threshold value Thx (Px> Thx) (S5Y), the received voice is present, and the sub echo canceller SEC is instructed to perform an adaptive operation. The sub-adaptive operation instruction signal As is set to 1 (S6, S7).

When the received signal power Px is not greater than the preset received sound threshold value Thx (S5N), there is no received sound and the sub-adaptive operation instruction signal As is cleared to 0 (S8). The sub echo canceller SEC stops its operation. Thereafter, the above-described operation is repeated.

  A sub-echo canceller SEC and a main echo canceller SEC are provided, and the sub-echo canceller SEC outputs an echo cancellation error signal bs necessary for determination of double talk, which is sent to the partner terminal side. Is used only for adaptive control of the main echo canceller MEC during double talk, so the call quality during double talk in a hands-free call is improved.

  The echo cancellation device of the present invention can be replaced with the echo canceller of Patent Document 2, and if combined with the voice switch shown in Patent Document 2, it is a high-quality hands-free device that is not affected by acoustic echo. It will be clear from the above description that a call can be made.

It is the circuit block diagram which showed the Example of the echo cancellation apparatus of this invention. (Example 1) FIG. 2 is an adaptive operation diagram of a sub-echo canceller according to an instruction of an adaptive operation control unit that is an important component of FIG. FIG. 2 is a circuit configuration diagram showing details of an internal configuration of an adaptive operation control unit that is an important component of FIG. FIG. 2 is a flowchart showing an operation flow of an adaptive operation control unit which is an important component of FIG. 1. FIG. It is a loop gain figure for demonstrating the loop gain of a prior art example. It is a call route figure for demonstrating the call route of a prior art example. It is a circuit block diagram of the echo cancellation apparatus which shows a prior art example. FIG. 8 is a circuit configuration diagram showing details of an internal configuration of an adaptive operation control unit that is a component of FIG. 7 showing a conventional example. FIG. 8 is a circuit diagram of an N-tap adaptive filter that is a component of FIG. 7 showing a conventional example. FIG. 8 is an adaptive operation diagram of an echo canceller according to an instruction of an adaptive operation control unit that is a component of FIG. 7 showing a conventional example.

Explanation of symbols

10, 10B Adaptive motion controller
11, 11B, 12 Adaptive filter
16,17 Adder
21, 21B Power calculation part
23 Acoustic coupling amount calculator
24 Received sound determination unit
25 Transmitted sound determination unit
26 Echo canceller adaptation permission judgment part
27 Echo cancellation amount calculator
28 Sub-echo canceller (EC) receiving voice determination unit
29 Main echo canceller adaptation permission judgment part
36 Acoustic coupling
37 Transmitted sound determination signal
38 Received voice determination signal
51 Echo cancellation device
53, 54, 56, 57 Amplifier A Adaptive operation instruction signal a Tap amplifier Am Main adaptive operation instruction signal As Sub adaptive operation instruction signal b, bm Transmission signal after echo cancellation (EC) bs Echo cancellation error signal d Adder EC Echo canceller Ers Echo cancellation amount of sub echo canceller (SEC) h Amplifier output M Microphone MEC Main echo canceler Pb Echo cancel (EC) post-transmission signal power Pbs Echo cancel error signal power Px Received signal Power Py Echo cancellation (EC) Pre-transmission signal power S Speaker SEC Sub echo canceller Thx Erasing threshold Thy Received sound threshold TM1 Input end TM2 Output end x Received signal y Echo cancel (EC) Previous transmission signal y'm Main pseudo echo signal y's Sub pseudo echo signal

Claims (8)

In response to the instruction of the sub adaptive operation instruction signal (As), the sub pseudo echo signal (y's) is used to remove the acoustic echo that wraps around the microphone (M) from the speaker (S) that converts the received signal (x) into sound. ) And sub-echo signal (y's) is subtracted from the echo cancel (EC) pre-transmission signal (y) from the microphone (M) side to generate an echo cancel (EC) error signal. Sub echo canceller (SEC) processing to obtain (bs = y−y ′s),
Receiving the echo cancellation (EC) error signal (bs), the echo cancellation (EC) pre-transmission signal (y), and the reception signal (x), and receiving the power of the reception signal (x) When the signal power (Px) is larger than the preset received voice threshold (Thx) (Px> Thx), the sub-adaptive operation instruction signal (As) is set (As = 1) and is small ( When Px ≦ Thx), the sub-adaptive operation instruction signal (As) is cleared (As = 0), and the power (Pbs) of the echo cancellation (EC) error signal (bs) and before the echo cancellation (EC) The echo cancellation amount (Ers = Pbs / Py) in the sub echo cancellation (SEC) process is calculated from the power (Py) of the transmission signal (y), and the echo cancellation amount (Ers) is set in advance. The main adaptive operation instruction signal (Am) is set (Am = 1) when it is smaller than the erase amount threshold (Thy) (Ers ≦ Thy), and the main adaptive operation instruction is issued when it is larger (Ers> Thy). Adaptive operation control processing (10) for clearing the signal (Am) (Am = 0),
In response to the instruction of the main adaptive operation instruction signal (Am), main adaptive filter processing (11) is performed from the received signal (x) to generate a main pseudo echo signal (y′m), and the microphone (M) The main pseudo echo signal (y'm) is subtracted from the transmission signal (y) before echo cancellation (EC) from the side, and the transmission signal (bm) after echo cancellation (EC) is generated and transmitted. Main echo canceller (MEC) processing to
Acoustic echo cancellation control method. In the adaptive motion control process (10),
The echo cancellation (EC) error signal (bs), the echo cancellation (EC) pre-transmission signal (y), and the reception signal (x) are received, the respective powers are calculated, and echo cancellation ( EC) error signal power (Pbs), echo cancel (EC) pre-transmission signal power (Py) and power calculation processing (21) to obtain reception signal power (Px),
When the received signal power (Px) is greater than a preset received voice threshold (Thx) (Px> Thx), it is determined that the received voice is present, and the sub-adaptive operation instruction signal (As) is set ( As = 1), if it is small (Px ≦ Thx), it is determined that there is no reception sound and the sub-adaptive operation instruction signal (As) is cleared (As = 0). Processing (28)
Echo cancellation for calculating the echo cancellation amount (Ers = Pbs / Py) from the ratio of the echo cancellation (EC) error signal power (Pbs) and the transmission signal power (Py) before the echo cancellation (EC). Do the amount calculation process (27),
When the sub-adaptive operation instruction signal (As) indicating the received voice and the echo cancellation amount (Ers) are received and the echo cancellation amount (Ers) is smaller than a preset cancellation amount threshold (Thy) (Ers When it is determined that the adaptive operation is permitted to the main echo canceller (MEC) at ≦ Thy, the main adaptive operation instruction signal (Am) is set (Am = 1) and output, and if large (Ers> (Thy), a main echo cancellation adaptation permission determination process (29) for clearing (Am = 0) the main adaptation operation instruction signal (Am).
2. The acoustic echo cancellation control method according to claim 1. In the sub echo cancellation (SEC) process,
Upon receiving the received signal (x) and the echo cancel (EC) error signal (bs), an adaptive filter operation is performed in accordance with an instruction of the sub adaptive operation instruction signal (As), and the sub pseudo echo signal (y do sub-adaptive filtering (12) to get
A subtraction process (17) for subtracting the sub pseudo echo signal (y's) from the transmission signal (y) before the echo cancellation (EC) to obtain an echo cancellation (EC) error signal (bs). 2. The acoustic echo cancellation control method according to claim 1. The acoustic echo cancellation control method according to claim 3, wherein the number of taps used in the sub adaptive filter processing (12) is smaller than the number of taps (N) used in the main adaptive filter processing (11). In response to the instruction of the sub adaptive operation instruction signal (As), the sub pseudo echo signal (y's) is used to remove the acoustic echo that wraps around the microphone (M) from the speaker (S) that converts the received signal (x) into sound. ) And sub-echo signal (y's) is subtracted from the echo cancel (EC) pre-transmission signal (y) from the microphone (M) side to generate an echo cancel (EC) error signal. Sub-echo canceller (SEC) means to obtain (bs = y-y's);
Receiving the echo cancellation (EC) error signal (bs), the echo cancellation (EC) pre-transmission signal (y), and the reception signal (x), and receiving the power of the reception signal (x) When the signal power (Px) is larger than the preset received voice threshold (Thx) (Px> Thx), the sub-adaptive operation instruction signal (As) is set (As = 1) and is small ( Px ≦ Thx), the sub-adaptive operation instruction signal (As) is cleared (As = 0), the echo cancel (EC) error signal (bs) and the echo cancel (EC) pre-transmission signal (y) ) To calculate the echo cancellation amount (Ers = Pbs / Py) in the sub-echo canceller (SEC) means, and the echo cancellation amount (Ers) is calculated based on the preset cancellation amount threshold (Thy). When it is small (Ers ≦ Thy), the main adaptive operation instruction signal (Am) is set (Am = 1), and when it is large (Ers> Thy), the main adaptive operation instruction signal (Am) is cleared (Am = 0). ) Adaptive motion control means (10),
In response to the instruction of the main adaptive operation instruction signal (Am), a main pseudo-echo signal (y′m) is created from the received signal (x) by the main adaptive filter means (11), and from the microphone (M) side. To subtract the main pseudo-echo signal (y'm) from the transmission signal (y) before echo cancellation (EC), and generate and transmit the transmission signal (bm) after echo cancellation (EC). Including the main echo canceller (MEC) means,
Acoustic echo cancellation control device. The adaptive motion control means (10)
The echo cancellation (EC) error signal (bs), the echo cancellation (EC) pre-transmission signal (y), and the reception signal (x) are received, the respective powers are calculated, and echo cancellation ( EC) error signal power (Pbs), power calculation means (21) for obtaining transmission signal power (Py) and reception signal power (Px) before echo cancellation (EC);
When the received signal power (Px) is greater than a preset received voice threshold (Thx) (Px> Thx), it is determined that the received voice is present, and the sub-adaptive operation instruction signal (As) is set ( As = 1), if it is small (Px ≦ Thx), it is determined that there is no reception sound and the sub-adaptive operation instruction signal (As) is cleared (As = 0). Means (28);
Echo cancellation for calculating the echo cancellation amount (Ers = Pbs / Py) from the ratio of the echo cancellation (EC) error signal power (Pbs) and the transmission signal power (Py) before the echo cancellation (EC). Amount calculation means (27);
When the sub-adaptive operation instruction signal (As) indicating the received voice and the echo cancellation amount (Ers) are received and the echo cancellation amount (Ers) is smaller than a preset cancellation amount threshold (Thy) (Ers When it is determined that the adaptive operation is permitted to the main echo canceller (MEC) at ≦ Thy, the main adaptive operation instruction signal (Am) is set (Am = 1) and output, and if large (Ers> (Thy) includes main echo cancellation adaptation permission determination means (29) for clearing (Am = 0) the main adaptation operation instruction signal (Am).
6. The acoustic echo cancellation control device according to claim 5. The sub echo canceller (SEC) means includes:
Upon receiving the received signal (x) and the echo cancel (EC) error signal (bs), an adaptive filter operation is performed in accordance with an instruction of the sub adaptive operation instruction signal (As), and the sub pseudo echo signal (y sub-adaptive filter means (12) to obtain 's);
Subtracting means (17) for subtracting the sub pseudo echo signal (y's) from the transmission signal (y) before the echo cancellation (EC) to obtain an echo cancellation (EC) error signal (bs); 6. The acoustic echo cancellation control device according to claim 5, further comprising: The acoustic echo cancellation control apparatus according to claim 7, wherein the number of taps used in the sub adaptive filter means (12) is smaller than the number of taps (N) used in the main adaptive filter means (11).

Images