JPH01318325A - Echo canceller - Google Patents

Abstract

PURPOSE:To prevent the adaptation of an echo canceller at bidirectional simultaneous communication and to continue the adaptation in the fluctuation of the echo path by providing 1st and 2nd echo cancellers and an echo canceller selections means. CONSTITUTION:An echo canceller selection means 8 compares a transmission output level of the 1st echo canceller 5 with a transmission output level of the 2nd canceller 6 so as to check which of them cancels the echo better. Then the echo canceller 5(6) identifying better the echo path better at all times is selected by the echo canceller selection means 8 and the echo canceller 5(6) to be selected and offering the better echo cancelling quantity gives a transmission output signal and the echo canceller 6(5) with worse echo cancellation identifies the characteristic of the echo path. Thus, when a nearby talker at bidirectional simultaneous communication utters, the echo cancellers 5, 6 sending an output can stop the adaptation and when the echo path is fluctuated, the adaptation is implemented immediately to follow the fluctuation of the echo path.

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, the practical use of teleconference systems, and the spread of hands-free telephones, 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 cancelers that cancel 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). The presence or absence must be detected. Toshiba Minami and Kawasaki have shown a method for detecting two-way simultaneous calls (for example, 1989 Telecommunications Institute National Conference 589).

The conventional echo canceling device as described above will be described below with reference to the drawings.

FIG. 2 shows the configuration of a conventional echo canceling device.

In FIG. 2, 1 is a receiving side input terminal, 2 is a receiving side output terminal, 3 is a transmitting side input terminal, 4 is a transmitting side output terminal, 50
An echo canceller provided between terminals 1 to 4 cancels echoes returning from the receiving output terminal to the transmitting output terminal.

56 is the transmission output signal level Ls of the echo canceller 50
57 is a reception signal level detection means for detecting the reception signal level Lr1n 1. 58 is the transmission output signal level Lsout, the reception signal level Lr1n 1 and the dark value kt used last time are used this time. Dark value updating means for calculating the dark value kt, 59 is the dark value kt
and reception signal level estimating means 60 for estimating the reception signal level from the transmission output signal level, and 60 identifies the echo path of the echo canceller 50 by comparing the estimated reception signal level Lr1n2 with the actual reception side signal level L, rinl. This is a comparison means that controls whether or not the process is performed.

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 50 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.

If there is no near-end speaker signal, the signal Rout enters the transmitting side input terminal 3 as the transmitting power Sin through the echo path.

In the echo canceller 50, the series of receiving human power Rin and the transmitting output Sou are adjusted so that the transmitting output signal Sout becomes zero.
Create an echo path from t to pseudo i and eliminate the echo signal. However, if a near-end speaker exists, the echo path cannot be estimated successfully. Therefore, it is necessary to detect the presence of a near-end speaker signal and stop the adaptation of the echo canceller 50. The near-end speaker detection method will be described below.

In a complete reception state in which only the far-end speaker is speaking, the threshold value updating means 58 gradually increases the threshold value kl so as not to exceed the convergence speed of the echo canceller. The received signal level estimating means 59 sets the erased residual signal level Lsout to the darkness value kt.
Multiplying Lr1n2 calculates the estimated received signal level Lr1n2. The comparison means 60 compares the estimated received signal level Lr1n2 and the received signal level Lr1n1, and determines that Lr1n2≦Lr1nl.
If Lr1n 2 >Lr1n 1, then the adaptation is stopped. Since I am currently listening, Lr1n2≦Lr1n
1, and adaptation continues. Here, in a two-way simultaneous conversation state in which the near-end speaker speaks, the transmission output level Lsout of the echo canceller increases and the estimated received signal level Lr
1n2 becomes larger. Therefore, Lr1n 2 >Lr
1nl, and the adaptation of the echo canceller 50 can be stopped. However, even if the characteristics of the echo path change rapidly, the transmission output signal level Ls of the echo canceller 50
The out force becomes strong and adaptation stops. Therefore, the threshold value updating means 58 gradually lowers the threshold value kt. Therefore, the estimated received signal level Lr1n2 of the received signal level estimation means 59 becomes small, and after a while, Lr1n2≦
1rin 1 and adaptation is restarted. Since the same operation is performed in the two-way simultaneous communication state and when the echo path fluctuates, the speed at which the darkness value kt is increased is adjusted to set the time for restarting adaptation to 3 seconds. This uses the fact that 96% of simultaneous two-way calls are completed within three seconds.

In the transmitting state where only the near-end speaker is speaking, the threshold value updating means 58 holds the threshold value kt.

Problems to be Solved by the Invention However, with the above configuration, if the echo path fluctuates, it may not be possible to cancel the echo for more than 3 seconds. Also,
When holding a conference with a large number of people using a teleconference system, etc., there is a high possibility that simultaneous two-way calls lasting more than 3 seconds will occur, so it is necessary to increase the judgment time, and it is said that it is slow to follow changes in the echo path. It had drawbacks.

In view of the above-mentioned problems, the present invention provides an echo canceling device that can stop adapting the echo canceller to simultaneous two-way communication, but can continue adapting when the echo path changes.

Means for Solving the Problems In order to achieve this object, the echo cancellation device of the present invention includes a first echo canceler and a second echo canceller that cancel echo signals returned to the transmission side input terminal, and a first echo canceller selection means for comparing the transmission output signal level of the second echo canceller with the transmission output signal level of the second echo canceller to determine which echo canceller identifies the characteristics of the echo path better; a transmission output switch that sends the transmission output of the echo canceller selected by the canceller selection means to a transmission side output terminal as a transmission output signal; and reception state detection means that considers a reception state when the transmission input signal level is lower than the reception output signal level;
a first adaptive control that performs control to identify characteristics of the echo path of the first echo canceller when the echo canceller selection means selects the second echo canceller and the reception state detection means determines that the reception state is in the reception state; a second echo canceller for controlling the echo canceller to identify characteristics of the echo path of the second echo canceller when the echo canceller selecting means selects the first echo canceller and the receiving state detecting means determines that the receiving state is in the receiving state; Adaptive control means.

Effect: With this configuration, the echo canceller selection means always selects the echo canceller that identifies the echo path better, so that in the receiving state, the one with the better echo cancellation outputs the transmission output signal, and the one with the worse echo cancellation outputs the transmit output signal, and the one with the worse echo cancellation One echo canceller attempts to identify the characteristics of the echo path. In a complete listening situation where only the far-end speaker is speaking, the first and second echo cancellers alternately adapt to identify the characteristics of the echo path. At this time, since the echo canceller selection means has hysteresis, the selection does not change at very fast intervals, for example, every sample, and the characteristics of the echo path can be stably identified. In the case of simultaneous two-way communication, an echo canceller that performs adaptation cannot identify the echo path well, so the amount of echo cancellation is worse than an echo canceller that does not perform adaptation. Therefore, the adaptation of the echo canceller that actually performs echo cancellation can be stopped. When the echo path changes, the echo canceller that was not selected by the echo canceller selection means is adapted, so the amount of echo cancellation becomes better than the selected echo canceller, and the echo canceller selection means switches the selection and immediately cancels the echo. It is possible to follow changes in the route.

Further, if the near-end speaker is speaking louder than the far-end speaker, or if only the near-end speaker is speaking, this is detected by the receiving state detection means, and the first. Adaptation of the second echo canceller is stopped. As described above, when the near-end speaker speaks during a simultaneous two-way conversation, the echo canceller that outputs the transmission output can stop adapting, and can continue adapting even when the echo path changes. This makes it possible to follow changes in the echo path.

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 Figure 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 the gain of the echo path from the receiving output terminal 2 to the transmitting input terminal 3 is 1. It is as follows. 5 is a first echo canceller, and 6 is a second echo canceller, which cancels the echo returning from the receiving side output terminal 2 to the transmitting side input terminal 3. 20 is a reception output level detection means for detecting the reception output signal level, 21 is a transmission input level detection means for detecting the transmission input signal level, and 22 is a first transmission output for detecting the transmission output signal level of the first echo canceller. level detection means, 23 is a second transmission output level detection means for detecting the transmission output signal level of the second echo canceller, 7 is the transmission output level of the first echo canceller and the second transmission output level;
echo canceller selection means for comparing the transmission output levels of the echo cancellers and selecting which echo canceller cancels the echo better; 8 connects the transmission output of the echo canceller selected by the echo canceller detection means to the transmission output terminal; 9 is a receiving state determining means for comparing the transmitting input level and receiving output level to determine whether or not the receiving state is in the receiving state; 10 is a second echo canceller selecting means when the receiving state determining means determines that the receiving state is the receiving state; a first adaptive control means that performs control to adapt the first echo canceller when the echo canceller is selected;
Reference numeral 11 denotes second adaptive control means for performing control to adapt the second echo canceller when the receiving state determining means determines that the receiving state is the receiving state and the echo canceller selecting means selects the first echo canceller. It is.

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

First, the far-end speaker's voice signal is supplied to the first and second echo cancellers 5.6 via the receiving side input terminal 1 as the receiving side input signal Rin. Further, the reception input signal Rin is sent from the reception side output terminal 2 to the near-end speaker as the reception side output signal Rou.
Sent as t. If there is no near-end talker signal, the signal Ro
A part of the signal u t passes through the echo path and enters the transmitting side input terminal 3 as the transmitting side input signal Sin. Now, the first echo canceller 5 is selected by the echo canceller selection means 7,
It is assumed that the transmission output switch 8 is turned to the a side to connect the transmission output of the first echo canceller 5 to the transmission output terminal 4. At this time, the receiving input signal level Lr1n detected by the receiver's power level detecting means 20 is equal to the transmitter's power level detecting means 2.
Since the transmission input signal level Lsin is higher than the transmitted input signal level Lsin detected in step 1, the receiving state detection means 9 determines that the receiving state is in the receiving state. The second echo canceller 6 is stopped and adapted by the second adaptive control means 11. The transmission output signal level Lsou of the second echo canceller 6 detected by the second transmission output level detection means 23 as the adaptation of the second echo canceller 6 progresses.
t2 becomes smaller, and the first transmission output level detection means 22
When the transmission output signal level Lsout of the first echo canceller 5 detected by Lsout becomes smaller than 1 and Lsout2<k*Lsoutli k=0.5, the echo canceller selection means 7 selects the second echo canceller 6 and transmits. The output switch 8 is tilted to the b side and the second
The transmission output of the echo canceller 6 is connected to the transmission output terminal 4. The first echo canceller 5 starts adaptation, and as the adaptation progresses, the transmission output signal level Lsoutl of the first echo canceller 5 detected by the first transmission output level detection means 22 becomes smaller, and the second It becomes smaller than the transmission output signal level Lsout2 of the second echo canceller 6 detected by the transmission output level detection means 23, and Ls
outl<k*Lsout2; When k=0.5, the echo canceller selection means 7 selects the first echo canceller 5 again, and the transmission output switch 8 also falls to the a side, so that the transmission output of the first echo canceller 5 Connect to the transmission output terminal 4. The above operations are repeated to adapt the first and second echo cancellers.

Next, a two-way simultaneous conversation state in which a near-end speaker interrupts the far-end speaker while the far-end speaker is speaking will be described.

It is now assumed that the first echo canceller 5 is selected by the echo canceller selection means 7 and the transmission output switch 8 is turned to the a side to connect the transmission output of the first echo canceller 5 to the transmission output terminal 4. During simultaneous two-way communication, the receiving input signal level Lr1n detected by the receiver's power level detecting means 2° is higher than the transmitting input signal level Lsin detected by the transmitting human power level detecting means 21, and the receiving state detecting means 9 is determined to be in the listening state, so
The first echo canceller 5 is stopped from adapting by the first adaptive control means 1°, and the second echo canceller 6 is stopped from adapting by the first adaptive control means 1°.
Adaptation is performed by the adaptive control means 11. Since the near-end speaker signal is mixed in the transmitted input signal Sin, the second echo canceller 6 cannot successfully identify the characteristics of the echo path. Therefore, the transmission output signal level Lsout2 of the second echo canceller 6 detected by the second transmission output level detection means 23 does not become smaller, and the first echo detected by the first transmission output level detection means 22 does not decrease. The transmission output signal level Lsout of the canceller 5 does not become smaller than 1, and L
soout2>k*Lsoutl; The relationship of k-0,5 is maintained, and the echo canceller selection means 7 continues to select the first echo canceller 6.

As described above, during simultaneous two-way communication, the echo canceller actually used for canceling the echo is not switched, and the echo can be canceled using the coefficients used before the simultaneous two-way communication starts.

In addition, if the near-end speaker is speaking louder than the far-end speaker, or if only the near-end speaker is speaking, the reception output signal level Lr detected by the reception output level detection means 20
1n is smaller than the transmission input signal level Lsin detected by the sender level detection means, the receiving state detecting means 9 determines that the receiving state is not in the receiving state, and the first. The adaptation of the first and second echo cancellers is stopped by the second adaptive control means.

Next, the operation when the echo path changes will be explained. now,
It is assumed that the first echo canceller 5 is selected by the echo canceller selection means 7, and the transmission output switch 8 is turned to the a side to connect the transmission output of the first echo canceller 5 to the transmission output terminal 4.

At this time, the reception input signal level Lr1n detected by the reception input level detection means 20 is determined by the transmission input level detection means 21.
Since the transmission input signal level Lsin is higher than the detected transmission input signal level Lsin, the receiving state detecting means 9 determines that the receiving state is in the receiving state. Therefore, the first echo canceller 5 stops adaptation by the first adaptive control means 1o. The second echo canceller 6 performs adaptation by the second adaptive control means 11. If the echo path fluctuates dramatically in this state, the first
．． Transmission output signal level Lso of the second echo canceller
Since both ut 1 and Lsout 2 become large, the echo canceller selection means 7 still selects the first echo canceller. However, since the second echo canceller 6 continues to adapt, the second echo canceller 6 detected by the second transmission output level detection means 23
The transmission output signal level Lsout2 of the first echo canceller 5 decreases rapidly, and the transmission output signal level Ls of the first echo canceller 5 is immediately detected by the first transmission output level detection means 22.
out becomes smaller than 1, Lsout2<k*Lsoutl; k=0.5, the echo canceller selection means 7 selects the second echo canceller 6, and the transmission output switch 8 falls to the b side and the second
The transmission output of the echo canceller 6 is connected to the transmission output terminal 4. Then, the first echo canceller 5 starts adaptation, and as the adaptation progresses, the transmission output signal level Lsout 1 of the first echo canceller 5 detected by the first transmission output level detection means 22 becomes smaller. It becomes smaller than the transmission output signal level Lsout2 of the second echo canceller 6 detected by the second transmission output level detection means 23, and Lsout 1 < k * Lsout 2 ;
When k = 0.5, the echo canceller selection means 7 selects the first echo canceller 5 again, and the transmission output switch 8 also falls to side a, connecting the transmission output of the first echo canceller 5 to the transmission output terminal 4. do. The above operations are repeated to adapt the first and second echo cancellers. Even if the echo path changes in this way, one of the echo cancellers is constantly adapting, so it can instantly follow the changes in the echo path.

As described above, according to this embodiment, the echo canceller selection means always selects an echo canceller that can better identify the echo path, and the selected echo canceller with a good echo canceling amount outputs a transmission output signal and echoes the echo. The echo canceller with worse cancellation attempts to identify the characteristics of the echo path, so that in a complete listening situation where only the far-end speaker is speaking, the first and second echo cancellers adapt alternately. We will identify the characteristics of the echo path. At this time, since the echo canceller selection means has hysteresis, the selection does not change at very fast intervals, for example every l samples, and the characteristics of the echo path can be stably identified. In addition, in the case of two-way simultaneous calls, the echo canceller that performs adaptation cannot identify the echo path well, so the amount of echo cancellation is worse than that of the echo canceller that does not perform adaptation, and the echo canceller is actually performing echo cancellation. The adaptation of the echo canceller can be stopped.

When the echo path changes, the echo canceller that was not selected by the echo canceller selection means is adapted, so the echo canceller selection means switches the selection and the two echo cancellers are adapted alternately in the same way as in a complete receiving state. Therefore, it is possible to immediately follow changes in the echo path. Further, if the near-end speaker is speaking louder than the far-end speaker, or if only the near-end speaker is speaking, this is detected by the receiving state detection means, and the first. Adaptation of the second echo canceller is stopped. As described above, when the near-end speaker speaks during a two-way simultaneous conversation, etc., the echo canceller that outputs the transmission output can stop adapting, and when the echo path changes, it can immediately adapt and change the echo path. can follow fluctuations in In addition, since two echo cancellers do not perform adaptation at the same time, the identified echo path only needs to be corrected for one unit, and the adaptive control logic for each echo canceller can be determined by simple level calculations and comparisons. The hardware can be increased with less hardware.

Note that the condition for switching the echo canceller selection of the echo canceller selection means from the second echo canceller to the first echo canceller is Lsoutl<k*Lsout2; k-0,5 first
From the echo canceller to the second echo canceller, Lsout2<k*Lsoutl; is set to = 0.5, but the value of k may be changed to a value of 1 or less depending on the environment in which it is applied.

Effects of the Invention As described above, the present invention compares the transmission output signal levels of two echo cancellers using the echo canceller selection means, always selects the echo canceller that identifies the echo path better, and selects the echo canceller that identifies the echo path better. The echo canceller with good echo cancellation outputs the transmitted output signal, and the echo canceller with poor echo cancellation attempts to identify the characteristics of the echo path. , the first and second echo cancellers alternately adapt to identify characteristics of the echo path. At this time, since the echo canceller selection means has hysteresis, the selection does not change at very fast intervals, for example, every sample, and the characteristics of the echo path can be stably identified. In addition, in the case of two-way simultaneous calls, the echo canceller that performs adaptation cannot identify the echo path well, so the amount of echo cancellation is worse than that of the echo canceller that does not perform adaptation, and the echo canceller is actually performing echo cancellation. The adaptation of the echo canceller can be stopped.

When the echo path changes, the echo canceller that was not selected by the echo canceller selection means is adapted, so the echo canceller selection means switches the selection and the two echo cancellers are adapted alternately in the same way as in a complete receiving state. Therefore, it is possible to immediately follow changes in the echo path. Further, if the near-end speaker is speaking louder than the far-end speaker, but only the near-end speaker is speaking, this is detected by the listening state detection means, and the first. Adaptation of the second echo canceller is stopped. As described above, when the near-end speaker speaks during a two-way simultaneous conversation, etc., the echo canceller that outputs the transmission output can stop adapting, and when the echo path changes, it can immediately adapt and change the echo path. can follow fluctuations in In addition, since two echo cancellers do not perform adaptation at the same time, only one echo path needs to be corrected, and the adaptive control logic for each echo canceller can be determined by simple level calculations and comparisons. It requires less hardware, can follow changes in the echo path at high speed, and is excellent in eliminating echo signals in the acoustic echo path of multi-person teleconference systems, hands-free telephone IM, etc. can be carried out, and its practical effects are great.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an echo canceling device according to an embodiment of the present invention, and FIG. 2 is a block diagram of a conventional echo canceling device. 5...First echo canceller, 6...Old...Second echo canceller, 7...Echo canceller selection means, 8...Transmission output switch, 9...
. . . Receiving state detection means, 10 . . . First adaptive control means, 11 . . . Second adaptive control means, 2
0... Reception output level detection means, 21...
. . . Transmission input level detection means, 22 . . . First transmission output level detection means, 23 . . . Second transmission output level detection means.

Claims (1)

[Claims] Identify the characteristics of the receiving side input terminal, the receiving side output terminal, the transmitting side input terminal, the transmitting side output terminal, and the echo path between the receiving side output terminal and the transmitting side input terminal, and a first echo canceller and a second echo canceller for canceling the echo signal returned to the transmitting side input terminal; a receiving output level detecting means for detecting the signal level of the receiving output signal; and detecting the signal level of the transmitting input signal. a first transmission output level detection means for detecting the signal level of the transmission output signal of the first echo canceller; and a second transmission output level detection means for detecting the signal level of the transmission output signal of the second echo canceller. a transmission output level detection means, a first transmission output signal level detected by the first transmission output level detection means, and a second transmission output signal level detected by the second transmission output level detection means. In comparison, if the first transmission output signal level is smaller than the second transmission output signal level, which is a certain value smaller than 1, it is considered that the first echo canceller identifies the characteristics of the echo path better; If the second transmitting output signal level is smaller than the fixed value of the first transmitting output signal level, it is assumed that the second echo canceller identifies the characteristics of the echo path better; otherwise, the previously selected an echo canceller selection means that considers that the echo canceller that has been selected better identifies the characteristics of the echo path; a transmission output switch that sends the transmission output of the echo canceller selected by the echo canceller selection means to a transmission side output terminal; Listening state detecting means which compares the receiving output signal level detected by the output level detecting means and the transmitting input signal level detected by the transmitting input level detecting means and deeming the case where the transmitting input signal level is smaller than the receiving output signal level as a receiving state. and the echo canceller selecting means selects a second echo canceller, and when the receiving state detecting means determines that the receiving state is in the receiving state, the echo canceller is controlled to identify characteristics of the echo path of the first echo canceller. first adaptive control means and the echo canceller selecting means select the first echo canceller, and when the receiving state detecting means determines that the receiving state is the receiving state, the characteristics of the echo path of the second echo canceller are determined. and second adaptive control means for controlling to perform identification.