JPS6251836A - Echo erasing device - Google Patents

Abstract

PURPOSE:To discriminate bidirectional simultaneous speaking and the variation of an echo path quickly by providing the first and second adaptive filters which approximate and synthesize an echo signal included in a transmission-side input terminal. CONSTITUTION:When a near-end speaker is absent, the first and second adaptive filters 5 and 7 simulate the echo path and the echo signal is erased by the first and second subtractors 6 and 8. When the near-end speaker or the echo path is varied, estimating of the echo path of the filter 5 is stopped immediately, and the variance of the echo path or bidirectional simultaneous speaking is discriminated by a bidirectional simultaneous speaking detecting part 9. If the variation of the echo path is discriminated, the detecting part 6 switches the first switch 10 to the side of the second subtractor 8 and switches the second switch 11 to the side of the last step of the filter 5 and resets the filter 7 to start updating echo path characteristics of filters 5 and 7. If bidirectional simultaneous speaking is discriminated, estimating of the echo path of the filter 5 is stopped as it is.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to an echo canceling device for controlling echoes in long-distance circuits such as satellite communications, teleconference systems, and the like.

2. Description of the Related Art In recent years, with the rapid increase in long-distance lines, especially satellite communications, and the practical use of teleconference systems, clear and natural voice transmission has been desired. Reverberation signals are a phenomenon in which a signal generated by a speaker returns to the speaker due to reflection on the receiver, but in teleconference systems that include long-distance lines and acoustic systems, long-delayed echoes occur. This causes deterioration in call quality. Recently, adaptive echo canceling devices that eliminate this echo are being put into practical use, but these are remote-operated devices (identifying the echo path based on the far-end speaker's voice).
Therefore, the presence or absence of a near-end speaker must be detected. (For example, a method for detecting two-way simultaneous calls, Minami, Kawasaki, 1989 National Conference of the Telecommunications Division of the Institute of Electronics and Communication Engineers, 589) The conventional echo canceling device as described above will be described below with reference to the drawings.

FIG. 3 shows the configuration of a conventional echo canceling device.

In Fig. 3, 1 is the receiving side input terminal, 2 is the receiving side output terminal, 3 is the transmitting side input terminal, 4 is the transmitting side output terminal, and 5 is the echo canceller provided between each terminal 1 to 4. Eliminates the echo that returns from the side output terminal to the transmitting side output terminal. Reference numeral 16 indicates an erased residual signal level detection unit that detects the level of the erased residual signal which is the transmission output of the echo canceller 6, 17 indicates a receiving side signal level detector, and 18 indicates the erased residual signal level LSout, the received signal level LRin, and the previous signal. 19 is a threshold storage unit that stores the threshold value Xt sent from the threshold value Kt, and 20 estimates the receiving side signal level from the threshold value Kt and the erasure residual signal level. The reception side signal level estimator 21 is a comparison unit that compares the estimated reception side signal level LHin with the actual reception side signal level LR4n and controls whether or not to identify the echo path of the echo canceller.

The operation of the echo canceling device configured as described above will be explained below.

First, the voice signal of the far end speaker is supplied to the echo canceller 6 via the receiving side input terminal 1 as the receiving human power Rin. Further, the signal Rin is sent from the receiving side output terminal 2 to the near-end speaker as a reception signal Rout. When there is no near-end speaker signal, the signal Rout enters the transmitter input terminal 3 as the transmitter input signal Sin through the echo path. In the echo canceller 5, the transmitting side output signal 5out (also called cancellation residual signal)
The series of receiving human power Rin and the transmitting output 5o so that becomes zero
Create a pseudo echo path from the ut. Although the echo signal is canceled in this way, if a near-end speaker is present, the echo path cannot be estimated successfully. Therefore, it is necessary to detect the presence of a near-end speaker signal and stop estimating the echo path.

Therefore, the presence or absence of a near-end speaker is detected from the relationship between the receiving side input signal Rin and the erasure residual signal 5out which is the transmitting side output signal. A receiving side signal level LRin is detected by the receiving side signal level detection section 7. The erased residual signal level detection section 16 detects the erased residual signal level LSout. The threshold updating unit 18 compares the receiving side signal level LJn and the erased residual signal level L.
Sout, the previously used threshold Kt from the threshold storage section 9, and the signal Mdp from the comparison section 21 indicating the stop or continuation of echo path estimation.
In response to this, the threshold value Kt' to be used this time is calculated and sent to the threshold value storage section 9. Next, the receiving side signal level estimator 20 uses a threshold value K.
An estimated value LRin of the receiving side signal level is calculated from t' and the erased residual signal level LS□ut.

The comparator 21 compares the estimated receiving side signal level and the actual receiving side signal level LJn of /ゞ＼, and determines that LRin>LJn
When , it is assumed that a near-end speaker exists and estimation of the echo path is stopped.

The threshold value is updated by increasing the threshold value little by little as the echo path estimation progresses so as not to exceed the convergence speed of the echo canceller. If there are only near-end speakers, the threshold value is not updated. In the case of simultaneous two-way communication or a change in the echo path, the threshold value is decreased by a value proportional to the amount of cancellation attenuation so that the time required to determine that the change in the echo path is caused by the signal level on the transmitting side does not change significantly. The above judgment time is approximately 3 seconds.

Problems to be Solved by the Invention However, with the above configuration, when holding a conference with a large number of people using a teleconference system, there is a high possibility that two-way simultaneous calls lasting more than 3 seconds will occur, so the determination time will be longer. However, it has the disadvantage that it is slow to follow changes in the echo path.

In view of the above-mentioned problems, the present invention provides an echo canceling device that can quickly determine simultaneous two-way communication and fluctuations in the echo path.

Means for Solving the Problems In order to achieve this object, the echo canceling device of the present invention includes first and second adaptive filters that approximate synthesize the echo signals included in the transmitting input terminal, and a receiving input terminal. a first subtracter that subtracts the output signal of the first adaptive filter from the signal; and a second subtractor that subtracts the output signal of the second adaptive filter from the output of the first subtracter; A first switch that selects the output signal of the first subtracter, and otherwise selects the output signal of the second subtractor, and the input signal of the second adaptive filter are connected at °C, during simultaneous two-way communication. a second switch that selects a receiving side signal and otherwise selects a final tap received signal stored in the first adaptive filter; and a two-way simultaneous call detection unit that controls these adaptive filters and the switch. Become. The two-way simultaneous call detection unit includes a reception input signal level detection means, a first echo cancellation effect calculation means for calculating a first echo cancellation effect that is a ratio of the input and output signals of the first subtracter, and a first a second echo cancellation effect calculation means for calculating a second echo cancellation effect that is a ratio of the input signal of the subtracter to the output signal of the second subtracter;
adaptive stopping means for detecting simultaneous two-way communication or a change in the echo path from the echo cancellation effect calculating means and stopping estimation of the echo path of the first adaptive filter; and storing the echo cancellation effect of the first adaptive filter. estimating the echo path characteristics of the first adaptive filter by determining whether it is a two-way simultaneous call or a change in the echo path from the effect storage means and the first echo cancellation effect and the second echo cancellation effect stored in the effect storage unit; , restart or stop of the second adaptive filter, restart/start of the second adaptive filter, and determination means for controlling the first and second switches.

Operation With this configuration, when there is no near-end speaker, the first and second adaptive filters simulate the echo path and the first and second subtractors cancel the echo signal.

When the near-end speaker or the echo path changes, the first echo cancellation effect calculated by the first echo cancellation effect calculation means decreases,
The adaptive stopping means detects this and immediately stops estimating the echo path of the first adaptive filter, and the determining means determines whether it is a change in the echo path or a simultaneous two-way conversation. The determining means compares the second echo-cancelling effect calculated by the second echo-cancelling effect calculation means with the first echo-cancelling effect stored in the storage means, and determines whether the second echo-cancelling effect is equal to the first echo-cancelling effect. If it is larger than the echo cancellation effect, it is determined that there is a variation in the echo path, the first switch is switched from the first subtracter to the second subtracter, and the second switch is switched from the receiving side signal to the final tap of the first adaptive filter. At the same time as switching, the second adaptive filter is reset and updating of the echo path characteristics of the first and second adaptive filters is started. In cases other than the above, it is assumed that there is a two-way simultaneous conversation, and the estimation of the echo path of the first adaptive filter remains stopped. At this time, the second
The adaptive filter is reset and started at regular intervals by the determining means. In this way, fluctuations in the echo path and simultaneous two-way communication can be determined within the convergence time of the second adaptive filter, and furthermore, when there is no near-end speaker, the combined tap length of the first and second adaptive filters can be determined. This makes it possible to eliminate long-term echo signals as an adaptive filter.

EXAMPLE An example of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of an echo canceling device in one embodiment of the present invention.

In FIG. 1, 1 is a receiving side input terminal, 2 is a receiving side output terminal, 3 is a transmitting side input terminal, and 4 is a transmitting side output terminal. The above configuration is the same as the configuration shown in FIG.

6 is a first adaptive filter that approximately synthesizes echoes returning from the receiving output terminal to the transmitting output terminal; 6 is a first subtracter that subtracts the signal Sin approximately synthesized by the first adaptive filter from the transmitting side input signal Sin; 7 is a second adaptive filter that approximately synthesizes the output Din of the first subtracter 6; 8 is a Output Din of the first subtractor 6
A second subtractor 1o subtracts the output Din of the second adaptive filter 7 from the output signal 5out on the transmitting side, and the output signal Sin of the first subtractor 6 or the output signal Dout of the second subtractor 8 on the receiving side. Signal R1n and second subtractor D
A first switch 11 selects whether to output or output the receiving side input signal R as the receiving input of the second adaptive filter 7.
A second switch 9 selects between the input signal R and the signal from the final tap of the first adaptive filter 6;
in, transmitting side output signal Sin, and transmitting side output signal 5out
The presence or absence of a near-end speaker is detected from the output signal Dout of the second subtractor 7 and the first and second adaptive filters 6.7 and the first . This is a two-way simultaneous call detection unit that controls the second Sui Notes 10 and 11.

FIG. 2 is a block diagram showing the detailed configuration of the two-way simultaneous call detection section 9. As shown in FIG. In FIG. 2, 91 is the ratio c of the output signal Din of the first subtractor 6 and the input signal Sin on the transmitting side.
nc1 (hereinafter referred to as the first echo cancellation effect); 92 is the ratio CnC of the output signal DOut of the second subtracter 8 and the transmission side input signal Sin;
2 (hereinafter referred to as the second echo cancellation effect), the second echo cancellation effect calculating means 93 calculates the receiving side input signal Ri.
received signal level detection means for detecting the signal level of n;
4 is a detection means for detecting two-way simultaneous communication from the first echo canceller CNC1 or a change in the echo path; 96 is a first echo when the first adaptive filter is estimating the echo path; Elimination effect Mcnc1 Echo cancellation effect ACnC
96 is an effect storage means 95 for storing
The echo cancellation effect ACnC2 stored in the echo cancellation effect ACnC2 is compared with the second echo cancellation effect ACnC2 to determine two-way simultaneous communication and fluctuations in the echo path, and the echo paths of the first and second adaptive filters are estimated. This is determination means for controlling the second switch. The tap lengths of the first and second adaptive filters are the same.

The operation of the echo canceling device configured as described above will be explained below.

First, the voice signal of the far end speaker is supplied to the first adaptive filter 5 via the receiving side input terminal 1 as the receiving human power Rin. Further, the signal Rin is sent from the receiving side output terminal 2 to the near-end speaker as a reception signal Rout. When there is no near-end speaker signal, the signal Rout enters the transmitter input terminal 3 as the transmitter input signal Sin through the echo path. When there is no near-end talker, the first switch is connected to the second subtractor, and the second switch is connected to the last tap of the first adaptive filter. Therefore, the first and second adaptive filters 5.6 function as one adaptive filter and operate to estimate the echo path from the receiving side output terminal 2 to the transmitting side input terminal 3.

The specific operations of the first and second adaptive filters 6.7 will be explained. The first adaptive filter 5 creates a pseudo echo path from the series of receiving human power Rin and the output signal Din so that the output signal Din of the first subtractor 6 becomes zero. Similarly to the first adaptive filter 5, the second adaptive filter 7 also receives the output signal D of the second subtracter 8.
The first adaptive filter 6 so that out becomes zero.
A pseudo echo path is created from the sequence of the receiving power Rin and the output signal Dout taken from the last signal. In this way, a pseudo echo path with the combined tap length of the first and second adaptive filters 5.7 is created and the echo signal is canceled, but if there is a near-end speaker, the echo path cannot be estimated successfully. I can't do it. Therefore, it is necessary to detect the presence of a near-end speaker signal and stop estimating the echo path. The two-way simultaneous call detection unit 7 detects the presence or absence of a near-end speaker and controls the first and second adaptive filters 5.7 and the first and second switches 6.8 in the following manner. near-end speaker,
Alternatively, when the echo path changes, the first echo cancellation effect ACnO1 calculated by the first echo cancellation effect calculation means 91 decreases, and the adaptive stop means 93 detects this and outputs the stop signal Stp.
is served. Upon receiving the stop signal Stp, the determining means 96 sets the adaptive control signal Mdp to 1 so as to stop estimating the echo path of the first adaptive filter, and sets R/S to '1' to temporarily reset the second adaptive filter 7. ) A reset signal R/S is issued to immediately start adaptation (R/S is set to '0'). Further, the first switch is controlled to select the output of the first subtracter, and the second switch is controlled to select the receiving side input signal Rin. As a result, the first adaptive filter 6 holds the simulated echo path, and the second adaptive filter 7 starts estimating the echo path from the receiving side input terminal 2 to the first subtractor 6. Next, in the determining means 96, the second
The second echo canceling effect kon02 calculated by the echo canceling effect calculating means 92 is compared with the first echo canceling effect Acna stored in the effect storage means 96. When the echo canceling effect kono and AOnO2 are equal, , the adaptive control signal Adp is set to '0' so that it is determined that there is a change in the echo path, and estimation of the echo path of the first adaptive filter 5 is started, and the reset node start signal R/S is set to reset the second adaptive filter 7. 111), immediately start adaptation (set the reset start signal R/S to '○'), and at the same time set the first switch 1o to the output of the second subtracter 8 and the second switch 11 to the first adaptive switch. Controls the selection of the final tap of the filter. This returns to the initial near-end speakerless state. When the second echo cancellation effect kon02 is worse than the first echo cancellation effect ACnO stored in the effect storage means 95, it is assumed that a near-end speaker exists, and the echo path of the first adaptive filter 5 The adaptive control signal Adp is set to '1' to keep the estimation of
Leave it as is. At this time, the second echo canceller 6
is reset and started by a reset and start signal R/S issued from the determination means 96 at regular intervals.

However, when the level LH4n of the receiving side input signal Rin is smaller than a predetermined threshold in the received signal level detecting means 93, it is regarded as noise and the estimation of the echo paths of the τ first and second adaptive filters is stopped. In this way, fluctuations in the echo path and simultaneous bidirectional communication can be determined within the convergence time of the second adaptive filter 7.

As described above, according to this embodiment, when there is no near-end speaker, the first and second adaptive filters simulate the echo path, and the first and second subtracters cancel the echo signal. When the near-end speaker or the echo path changes, the first echo cancellation effect calculated by the first echo cancellation effect calculation means decreases, and the adaptive stop means detects this and immediately stops the first adaptive filter. The estimation of the echo path is stopped, and the determining means determines whether the echo path is changing or whether there is a simultaneous two-way conversation. In the determination means, the second
The second echo-cancelling effect calculated by the echo-cancelling effect calculating means is compared with the first echo-cancelling effect stored in the storage means, and if the second echo-cancelling effect is greater than the first echo-cancelling effect. It is determined that there is a change in the echo path, and the first switch is switched from the first subtractor to the second subtractor, and the second switch is changed.
Simultaneously with switching the Sochi from the receiving signal to the final tap of the first adaptive filter, the second adaptive filter is reset and updating of the echo path characteristics of the first and second adaptive filters is initiated. In cases other than the above, it is assumed that there is a two-way simultaneous conversation, and the estimation of the echo path of the first adaptive filter remains stopped. At this time, the second adaptive filter is reset and started at regular intervals by the determining means. When determining whether it is a near-end talker or a variation in the echo path, and when a near-end talker is present, a first adaptive filter cancels the echo signal. The determination time is short, and when a near-end speaker is present, the echo signal is not audible. In this way, fluctuations in the echo path and simultaneous two-way communication can be determined within the convergence time of the second adaptive filter,
Furthermore, when there is no near-end speaker, the first and second adaptive filters can be combined or used as a long-length adaptive filter to eliminate long-term echo signals.

Note that in this embodiment, the second adaptive filter is
Although the adaptive filter has the same delay amount as the first adaptive filter, the adaptive filter may have a delay amount smaller than the delay amount of the first adaptive filter. In this case, in the determination means, when comparing the first echo cancellation effect and the second echo cancellation effect, the deterioration of the echo cancellation effect corresponding to the smaller delay amount of the second adaptive filter is Just add it to the echo cancellation effect. By reducing the delay amount of the second adaptive filter, the scale of the device can be reduced, and the convergence speed of the second adaptive filter becomes faster, making it easier to distinguish between simultaneous two-way communication and fluctuations in the echo path. Becomes faster.

Although the first echo cancellation effect is defined as the ratio between the output signal of the first subtracter and the input signal on the transmitting side, it may also be the ratio between the output signal on the transmitting side and the input signal on the transmitting side. In this case, in the determination means, when comparing the first echo cancellation effect and the second echo cancellation effect, it is determined that the delay amount of the second adaptive filter is smaller than the delay amount of the first and second adaptive filters. It is sufficient to add the deterioration of the echo canceling effect corresponding to the second echo canceling effect to the second echo canceling effect.

Effects of the Invention As described above, in the present invention, when there is no near-end speaker, the first and second adaptive filters simulate the echo path, and the first and second subtractors cancel the echo signal. When the near-end speaker or the echo path changes, the first echo cancellation effect calculated by the first echo cancellation effect calculation means decreases, and the adaptive stop means detects this and immediately stops the first adaptive filter. The estimation of the echo path is stopped, and the determining means determines whether the echo path is changing or whether there is a simultaneous two-way conversation. The determining means compares the second echo-cancelling effect calculated by the second echo-cancelling effect calculation means with the first echo-cancelling effect stored in the storage means, and compares the second echo-cancelling effect with the first echo-cancelling effect. If it is larger than the echo cancellation effect, it is determined that there is a variation in the echo path, the first switch is switched from the first subtracter to the second subtracter, and the second switch is switched from the receiving side signal to the final tap of the first adaptive filter. At the same time as switching, the second adaptive filter is reset and updating of the echo path characteristics of the first and second adaptive filters is started. In cases other than the above, it is assumed that there is a two-way simultaneous conversation, and the estimation of the echo path of the first adaptive filter remains stopped.

At this time, the second adaptive filter is reset and started at regular intervals by the determining means.

When determining whether it is a near-end talker or a variation in the echo path, and when a near-end talker is present, a first adaptive filter cancels the echo signal. The determination time is short, and when a near-end speaker is present, the echo signal is not audible.

In this way, fluctuations in the echo path and simultaneous two-way communication can be
It can be determined within the convergence time of the adaptive filter, and furthermore, when there is no near-end speaker, the long-term echo signal can be canceled as a long-length adaptive filter that combines the first and second adaptive filters. The adaptive filter always simulates the echo path well and can perform excellent echo cancellation even in a multi-person teleconfluence system, and its practical effects are outstanding.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an echo canceling device in an embodiment of the present invention, FIG. 2 is a block diagram of a two-way simultaneous call detection section of the echo canceling device in an embodiment of the present invention, and FIG. 3 is a block diagram showing the configuration of an echo canceling device in an embodiment of the present invention. FIG. 1 is a block diagram of a conventional echo cancellation device. 6...First adaptive filter, 6...
...First subtractor, 7...Second adaptive filter, 8...Second subtractor, 9...
...Two-way simultaneous call detection unit, 1o...First switch, 11...Second switch, 91...
...First echo cancellation effect calculation means, 92...
Second echo cancellation effect calculation means, 93... Received signal level detection means, 94... Adaptive stop means, 9
5... Effect storage means, 96... Judgment means. Name of agent: Patent attorney Toshio Nakao and 1 other person No. 1
Figure 2 Figure 3

Claims (3)

[Claims] (1) First and second adaptive filters that adaptively simulate the characteristics of the echo path and approximately synthesize the echo signals included in the transmission side input signal, and the output signal of the first adaptive filter from the transmission side input signal. a first subtracter that subtracts a first pseudo-echo signal, which is the output signal of the second adaptive filter; a two-way simultaneous call detection means for detecting a two-way simultaneous call from the output signals of the first and second subtracters and the input/output signals of the transmitting and receiving sides; a first switch that selects the output signal of the first subtractor at a certain time or at the beginning of a variation of the echo path and selects the output signal of the second subtractor at other times; As the received input signal of the filter, the receiving side signal is selected when there is a near-end talker or at the beginning of the fluctuation of the echo path, and at other times, the received signal of the last tap stored in the first adaptive filter is selected. An echo canceling device characterized in that it is equipped with a second switch. (2) A receiving signal level detection means in which the two-way simultaneous call detection unit detects the input signal level on the receiving side, and a first echo cancellation effect that is the ratio of the output signal level of the first subtracter and the input signal level on the transmitting side. a first echo cancellation effect calculation means for calculating the echo cancellation effect, and a first echo cancellation effect calculation means for detecting a change in the near-end speaker or the echo path from the receiving side input signal level and the first echo cancellation effect, and detecting a change in the echo path characteristic of the first adaptive filter. At the same time as stopping the update, controlling a first switch to switch from a second subtracter to the first subtractor, and a second switch to switch from the final tap of the first adaptive filter to the receiving signal. adaptive stopping means; second echo cancellation effect calculation means for calculating a second echo cancellation effect that is a ratio of the output signal level of the second subtractor and the input signal level of the first subtractor; The first when updating the coefficients of the first adaptive filter
effect storage means for storing a first echo cancellation effect stored in the effect storage means when the adaptive stop means detects that there is a change in the near-end speaker or the echo path; It is determined from the second echo canceling effect whether it is a near-end speaker or a change in the echo path, and when the echo path is changing, the first switch is switched to the first switch.
switch from the subtracter of the first adaptive filter to the second subtractor, switch the second switch from the receiving signal to the last tap of the first adaptive filter, and at the same time reset the second adaptive filter, determining means for starting updating of the echo path characteristics of the second adaptive filter, and stopping updating of the echo path characteristics of the first adaptive filter when the user is a near-end speaker; An echo canceling device according to claim 1, characterized in that: (3) After the determining means stops updating the echo path characteristics of the first echo canceller, the second adaptive filter is reset and restarted at regular time intervals. The echo canceling device described in Section 2.