JPH1013310A - Echo canceller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To control an execution/stop of an adaptive processing of an adaptive filter with high accuracy by allowing an adaptive processing control means to make decision through the use of an estimate echo level and a residual signal level. SOLUTION: An adaptive processing control means is provided on the echo canceller and inputs a reception signal Rin(k), a transmission signal Tin(k), and a residual signal Res(k) being digital signals and outputs an adaptive processing control flag FLGa. Further, level calculation means 5, 16, and 6 respectively calculate a reception signal level Lrin(k), a transmission signal level Ltin(k) and a residual signal level Lres(k) and outputs them. An echo gain estimate circuit 17 calculates an estimate echo gain EG from the reception signal level Lrin(k) and the transmission signal level Ltin(k) when the reception signal level Lrin(k) clearly exceeds a harmful threshold level. Thus, a mean value of level differences of the reception signal is used when the reception signal is clearly harmful.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo canceller for canceling an echo signal, which is a cause of deterioration of communication quality in satellite communications, loudspeakers and the like.

[0002]

2. Description of the Related Art As an adaptation control means in a conventional echo canceller, for example, (1) "Echo canceller technology" (Showa 61-12-20), Japan Industrial Technology Center
129-p130, as a filter coefficient replacement method of a conventional echo canceller,
For example, there is one described in (2) JP-A-62-24726. Hereinafter, the configuration will be described with reference to the drawings. FIG. 8 is a block diagram of a conventional echo canceller described in the above reference (1). k is the time of the digital signal, and Rin (k) represents the received signal at time k, that is, the voice of the far-end speaker. Echo path 1
In the case of an echo canceller for a loudspeaker, it corresponds to a path in which the voice of the far-end speaker reproduced from the speaker goes around from the speaker to the microphone via the acoustic space. In the case of, due to the impedance mismatch of the two-wire / four-wire conversion hybrid transformer, this corresponds to a path in which the voice of the far-end speaker leaks to the far-end side. The received signal Rin (k) passes through the echo path 1 to become an echo signal and is superimposed on the near-end voice Nin (k). Accordingly, the transmission signal Tin (k) is a signal on which echoes of not only the near-end voice but also the far-end voice are superimposed.
If the signal is transmitted to the far end as it is, the communication quality deteriorates. In the echo canceller, a pseudo echo signal Tin '(k) is generated by the adaptive filter 3 and the transmission signal T is generated by the echo subtractor 2.
By subtracting the pseudo echo signal Tin ′ (k) from in (k), the residual signal, that is, the residual signal Re after echo cancellation is obtained.
Generate s (k) and transmit to the far end. The adaptation control means 4 controls the adaptation of the adaptive filter 3, and stops the adaptation of the adaptive filter when double talk is detected, that is, when the far-end speaker and the near-end speaker are simultaneously speaking. Let
When only the far-end speaker is speaking, an adaptation control flag FLG is output to the adaptive filter in order to execute the adaptation.

[0003] Next, referring to FIG.
Will be described in more detail. FIG. 9 is a block diagram of the conventional adaptation control means 4 of the echo canceller described in the above reference (1). The level calculating means 5 receives the received signal Ri.
The logarithmically converted power of n (k), that is, the received signal level Lrin (k) is calculated according to equation (1) and output,
The level calculating means 6 calculates the logarithmically converted power of the residual signal Res (k), that is, the residual signal level Lres (k) according to equation (2) and outputs the result. Then, the level difference calculating means 7 calculates a level difference DL (k) between the received signal level Lrin (k) and the residual signal level Lres (k) according to the equation (3).

[0004]

(Equation 1)

[0005] The level difference DL (k) means the sum of the amount of echo loss by the echo path 1 and the amount of echo cancellation by the echo canceller. The level correcting means 8 calculates the estimated residual signal level Lres' (k) according to the equation (4). Lres ′ (k) = Lrin (k) −DL (k) (4) Here, the estimated residual signal level Lres ′ (k) is obtained when only the echo signal of the reception signal exists in the transmission signal Tin (k). That is, it corresponds to the estimated value of the residual signal level Lres (k) when only the far end speaker is speaking. Comparison means 9
Compares the residual signal level Lres (k) with the estimated residual signal level Lres ′ (k), and when the following conditional expression (5) is satisfied, both the far-end speaker and the near-end speaker are in the utterance state, that is, ,
It is determined that the state is the double talk state, and the adaptation control flag FLG is set to 0 in order to stop the adaptation of the adaptive filter. Lres (k)> Lres ′ (k) (5) Here, FLG = 1 indicates an instruction to execute adaptation for the adaptive filter, and FLG = 0 indicates an instruction to stop adaptation.

The adaptation control determines whether or not the near-end speech is at a level that interferes with the adaptation processing, and executes the adaptation.
This is a process for controlling the stop. Since the near-end voice interferes with the adaptation processing when its level exceeds the level of the echo signal due to the far-end signal, the execution and stop of the adaptation are performed by the echo in the transmission signal Tin (k). It should be controlled by the magnitude relationship between the signal level and the level of the near-end voice. That is, as shown in FIG. 10, when the echo signal level exceeds the level of the near-end audio, or when the level exceeds the value obtained by adding a margin to the level of the near-end audio, the adaptation should be performed. However, in the conventional technique, the residual signal level Lr
The double talk state is detected from es (k) and the estimated residual signal level Lres ′ (k). As shown in FIG. 11 or FIG. 12, the residual signal level Lres (k) and the estimated residual signal level Since Lres ′ (k) takes substantially the same value, even in a section where only the echo signal is included in the transmission signal Tin (k), as in the section a, the double-talk state is erroneously detected and adaptively performed. Will be stopped. At the time of double talk, the residual signal level Lre
Since s (k) exceeds the estimated residual signal level Lres ′ (k), the echo signal level exceeds the level of the near-end voice and adaptation is possible, as shown in section b in FIG. 11 or FIG. The adaptation stops even in the section.

Next, the configuration of a conventional filter canceling method for an echo canceller will be described with reference to the drawings. FIG.
It is a block diagram of the conventional echo canceller described in the above-mentioned literature (2). The conventional example of FIG. 13 differs from the conventional example of FIG. 8 in configuration in that it does not include the adaptation control unit 4 but includes a switching unit 12, a save memory 13, and a coefficient replacement control unit 15. The adaptation memory 10, the product-sum operation unit 11 and the filter coefficient update unit 14 of FIG.
Are divided for each function, and these are not the differences in configuration between the conventional example of FIG. 13 and the conventional example of FIG. In the following description, a description of the same parts as those in FIG. 8 will be omitted. Adaptation memory 1
0 stores a filter coefficient for adaptation updated using an algorithm from the transmission signal and the residual signal after echo cancellation. The save memory 13 stores the filter coefficients of the adaptation memory 10 in parallel for a predetermined interval while the adaptation is determined to be valid. The coefficient replacement control unit 15 controls the update of the filter coefficient for adapting the adaptive filter based on the difference between the transmission signal level and the residual signal level. First, the filter coefficient updating unit 14 converts, for example, the transmission signal Tin (k) and the residual signal Res (k) into, for example,
The first filter coefficient H (j) of the adaptive memory 10 is updated according to an adaptive filter algorithm such as an N-LMS method (learning identification method). The product-sum operation unit 11 generates a pseudo echo signal by performing a convolution operation of the received signal Rin (k) and the first filter coefficient H (j) of the adaptation memory 10. On the other hand, the coefficient replacement control means 15 transmits the transmission signal level Ltin (k) and the residual signal level Lres (k).
Are calculated in accordance with the equations (6) and (2), respectively, and the difference Ds (k) between Ltin (k) and Lres (k) is calculated.
Is calculated according to equation (7).

[0008]

(Equation 2)

Further, the coefficient replacement control means 15 calculates
Observing the time series of (k), when Ds (k) becomes large, it is determined that the adaptation of the adaptive filter is proceeding correctly, and the first filter coefficient of the adaptive memory 10 is stored in the save memory 13. A command to be transferred is given to the switching unit 12, and the switching unit 12 performs the transfer according to the command. As a result, the same value as the first filter coefficient of the adaptation memory 10 becomes
It is stored as the second filter coefficient in the save memory 13. When Ds (k) becomes small, the coefficient replacement control unit 15 determines that the adaptation of the adaptive filter is erroneously progressing, and stores the first filter coefficient of the adaptive memory 10 in the save memory 13. An instruction to replace with the stored value of the second filter coefficient is given to the switching unit 12, and the switching unit 12 performs transfer according to the instruction. FIG.
3 shows the time change of Ds (k) and the timing of the transfer process and the replacement process by the switching unit 12. H in the figure
(J) is a first filter coefficient of the adaptation memory 10, and Hm (j) is a second filter coefficient of the save memory 13. In the figure, Ds (k) increases and the first filter coefficient H (j) of the adaptation memory 10 is transferred to the second filter coefficient Hm (j) of the save memory 13 for storage. Is indicated by a pulse of “save timing from H (j) to Hm (j)”, and Ds (k) decreases, and the second filter coefficient H
m (j) is the first filter coefficient H of the adaptation memory 10
The timing at which data is transferred to (j) is represented by “Hm of H (j)”.
(J) Replacement timing ".

[0010]

Since the conventional echo canceller is configured as described above, the detection of the double talk state is not appropriate, and the adaptation is stopped even in a section where adaptation is possible. However, there is a problem that echoes remain without being canceled because the adaptation progresses slowly due to frequent stoppages. Also, when the amount of echo cancellation by the adaptive filter is small, there is also a problem that the echo remains and it is difficult to hear. Further, in the conventional configuration in which the filter coefficient of the adaptation memory is replaced with the saved value, if the value to be replaced is greatly different, the filter coefficient of the adaptation memory instantaneously changes greatly at the moment of replacement. At that moment, the output of the product-sum operation unit becomes discontinuous, and as a result, there is a problem that abnormal noise is generated in the residual signal.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and has as its object to obtain an echo canceller that accurately determines a double talk section and performs adaptation. It is another object of the present invention to provide an echo canceller with excellent sound quality in which the values are not discontinuous and no abnormal noise occurs at the moment of replacement even when the filter coefficients of the adaptation memory are replaced with the saved values.

[0012]

An echo canceller according to the present invention comprises: means for estimating an echo gain from the level of a received signal and the level of a transmitted signal; and an estimated echo obtained by adding the estimated echo gain to the received signal. Comparing means for comparing the level with the level of the residual signal after echo cancellation is provided, and when the estimated echo level is larger than a value obtained by adding a margin to the residual signal, the echo of the received signal included in the transmission signal is eliminated. And an adaptive control means for performing adaptive control.

[0013] Alternatively, means for estimating an echo gain from the level of the received signal and the level of the transmitted signal, and comparing means for comparing an estimated echo level obtained by adding the estimated echo gain to the received signal with the level of the transmitted signal. And when the estimated echo level is greater than a value obtained by subtracting a margin from the transmission signal, an adaptation control means for performing an adaptation control for canceling an echo of the reception signal included in the transmission signal is provided.

[0014] Alternatively, means for estimating an echo gain from the level of the received signal and the level of the transmitted signal, an estimated echo level obtained by adding the estimated echo gain to the received signal, and a level of the residual signal after echo cancellation. The first to compare
A comparison means, a second comparison means for comparing an estimated echo level obtained by adding an estimated echo gain to the reception signal and a transmission signal level, and echo cancellation from the transmission signal level and the residual signal level An echo canceling amount estimating means for estimating the amount is provided, and when the echo canceling amount is larger than the threshold, the echo of the received signal included in the transmission signal is canceled if the estimated echo level is larger than a value obtained by adding a margin to the residual signal. If the amount of echo cancellation does not exceed the threshold, if the estimated echo level is larger than the value obtained by subtracting the margin from the transmission signal, the adaptive control for canceling the echo of the reception signal included in the transmission signal is performed. The adaptive control means is provided.

Alternatively, an adaptation memory for storing a filter coefficient for adaptation updated by using an algorithm from a transmission signal and a residual signal after echo cancellation, and an adaptation memory while adaptation is determined to be valid. A save memory for storing filter coefficients of a memory in parallel for a predetermined interval, and a coefficient for controlling the suitability of updating the filter coefficient for adapting the adaptive filter based on the difference between the transmission signal level and the residual signal level A replacement control means, and an interpolation means for comparing and interpolating the value of the first filter coefficient in the adaptation memory with the value of the second filter coefficient in the save memory. When the value of the first filter coefficient is replaced with the value of the second filter coefficient, and the value of the first filter coefficient is replaced with the value of the second filter coefficient, Communicating And so as to sequentially updated by interpolation at a predetermined timing number so as to be a value.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1 FIG. An embodiment in which the adaptive control of the adaptive filter is improved will be described. FIG. 1 is a block diagram of the adaptation control means according to Embodiment 1 of the present invention. In this figure, the same parts as those in FIG. 9 are denoted by the same reference numerals, and description thereof is omitted. The adaptation control means is a means provided in the echo canceller shown in FIG. 8, and receives the received signal Rin (k), the transmission signal Tin (k) and the residual signal Res (k) which are digital signals, and performs the adaptation control. Flag FL
Ga is output. Level calculation means 5, 16, 6
Are the reception signal level Lrin (k) and the transmission signal level L
tin (k) and the residual signal level Lres (k) are calculated and output. Here, the level indicates, for example, a logarithmic conversion value of power. When the received signal level Lrin (k) exceeds a threshold value for clearly producing a sound, the echo gain estimating means 17 calculates an estimated echo gain EG from the received signal level Lrin (k) and the transmitted signal level Ltin (k) according to the following equation. Calculate according to (8). That is, the equivalent gain at which the received signal appears at the transmitting end on the echo path is calculated as an average value of several values in the previous section. This is usually a system-determined value, but it is sometimes calculated. Since the adaptation control means uses the average value of the level difference between the reception signal and the transmission signal when the reception signal is clearly a sound, the echo gain can be accurately estimated.

[0017]

(Equation 3)

The echo level estimating means 18 calculates the estimated echo level Lech (k) according to the equation (9).
That is, when there is no near-end signal, the amount by which the received signal appears as an echo in the transmitted signal is estimated. Lech (k) = Lrin (k) + EG (9) The comparing means 19 compares the value obtained by adding the margin α to the residual signal level Lres (k) with the estimated echo level Lech (k), and the following conditional expression ( 10) When the condition is satisfied, the adaptive control flag FLGa is set to 1; Lech (k) ≧ Lres (k) + α (10)

The operation of the present invention will be described with reference to FIG. 2 showing a case where the amount of echo cancellation is large, that is, a case where the adaptation proceeds and the echo is sufficiently canceled. The setting of the adaptation control flag FLGa in the present invention is based on the conditional expression (1).
As shown in (0), the residual signal level Lres (k) is compared with the estimated echo level Lech (k). FIG. 2 shows the residual signal level Lres (k) and the estimated echo level Lech.
The time change of (k) is shown. Residual signal level Lre
When the amount of echo cancellation is large, s (k) is almost equal to the near-end voice level, and the estimated echo level Lech
(K) is substantially equal to the echo signal level. As described above, the ideal parameters used for the adaptation control are the near-end voice level and the echo signal level. Therefore, in the present invention, the residual signal level Lres (k) and the estimated residual A more desirable parameter is used than the signal level Lres' (k). For this reason, adaptation can be performed in many sections as compared with FIG. 11 illustrating the conventional example. That is, as shown in section a of FIG.
In the section where only the echo signal is included in (k), the estimated echo level Lech (k) always exceeds the residual signal level which is an approximation of the near-end voice level, so that the adaptation control flag FLGa is continuously set to 1. This is the operation to be set. Also,
Even in a section where the echo signal level in the double talk section exceeds the near-end voice level as in the section b in FIG. 2, the estimated echo level Lech (k) exceeds the residual signal level as in the section a. Adaptation control flag FLGa is set to 1
Operation.

The margin α in the conditional expression (10) for comparison is obtained by changing the near-end voice level to the residual signal level Lres.
The error approximated by (k) and the estimated echo level Lech
Since (k) is an estimated value, it is for absorbing an error with respect to a true echo level. As described above, according to the present invention, when the amount of echo cancellation is large, good adaptation control can be performed by comparing the signals of the above configuration.

Embodiment 2 FIG. Another embodiment in which the adaptive control of the adaptive filter is improved will be described. FIG. 3 is a block diagram of the adaptation control means according to Embodiment 2 of the present invention. In this figure, the same parts as those in FIGS. 1 and 9 are denoted by the same reference numerals, and description thereof will be omitted. This adaptation control means is a means provided in the echo canceller shown in FIG. 8 and receives a received signal Rin (k) and a transmission signal Tin (k) which are digital signals as inputs, and applies an adaptation control flag FL.
Gb is output. The level calculating means 5 and 16 calculate and output the reception signal level Lrin (k) and the transmission signal level Ltin (k), as in the first embodiment. Echo gain estimating means 17 and echo level estimating means 18
Calculates the estimated echo gain EG and the estimated echo level Lech (k), as in the first embodiment. Comparison means 2
0 is the following conditional expression (1) based on the transmission signal level Ltin (k), the margin β, and the estimated echo level Lech (k).
When 1) is satisfied, the adaptation control flag FLGb is set to 1; Lech (k) ≧ Ltin (k) −β (11)

The operation of the present invention will be described with reference to FIG. 4 showing a case where the echo cancellation amount is small, that is, a case where the adaptation has not yet proceeded and the echo cancellation is not sufficient. In setting the adaptation control flag FLGb in the present invention, as shown in the conditional expression (11), the transmission signal level Ltin (k)
Is compared with the estimated echo level Lech (k). FIG.
Indicates a temporal change of the transmission signal level Ltin (k) and the estimated echo level Lech (k). The transmission signal level Ltin (k) represents an echo signal level when a near-end voice is not included in the transmission signal, and is therefore substantially equal to the estimated echo level Lech (k). At the time of double talk where the end voices overlap, the level of the signal in which the near-end voice and the echo signal are superimposed indicates the level of the signal, and therefore exceeds the estimated echo level Lech (k). Therefore, the transmission signal level Ltin (k) and the estimated echo level Lech
By comparing (k), a section in which only the echo signal is included in the transmission signal Tin (k) is estimated, and the adaptation control flag FLGb is set to 1. FIG. 12 illustrating a conventional example
As can be seen from comparison with FIG. 4, the operation of executing the adaptation in more sections is performed in FIG.

The margin β in the conditional expression (11) for comparison is to absorb an error with respect to a true echo level because the estimated echo level Lech (k) is an estimated value. As described above, according to the present invention, when the amount of echo cancellation is small, good adaptation control can be performed by comparing the signals with the above configuration.

Embodiment 3 FIG. Further, another form of adaptive control of the adaptive filter will be described. FIG.
FIG. 13 is a block diagram of an adaptation control unit according to Embodiment 3 of the present invention. In this figure, the same parts as those in FIGS. 9 and 1 are denoted by the same reference numerals, and description thereof is omitted. This adaptation control means is also means provided in the echo canceller of FIG. 8, and the input and output are the same as those of the configurations of FIGS. The level calculation means 5, 16, and 6 calculate and output the received signal level Lrin (k), the transmitted signal level Ltin (k), and the residual signal level Lres (k), as in the first embodiment. Here, the level indicates, for example, a logarithmic conversion value of power. The echo gain estimating means 17 and the echo level estimating means 18 calculate the estimated echo gain EG and the estimated echo level Lech (k) as in the above embodiment. Since the adaptation control means uses the average value of the level difference between the received signal and the residual signal when the received signal is clearly a sound, the echo cancellation amount can be accurately estimated. The comparing means 19 and 20 set the adaptation control flags FLGa and FLGb and output them to the flag selecting means 22 as in the first and second embodiments, respectively. When the received signal level Lrin (k) exceeds a threshold value for clearly producing a sound, the echo canceling amount estimating means 21 outputs the transmission signal level Lti.
The estimated echo cancellation amount ERLE is calculated from n (k) and the residual signal level Lres (k) according to equation (12).

[0025]

(Equation 4)

When the estimated echo canceling amount ERLE exceeds the threshold, that is, when the estimated echo canceling amount ERLE is large, the flag selecting unit 22 sets the adaptation control flag FLGa output from the comparing unit 19 and the estimated echo canceling amount ERLE exceeds the threshold. When there is no such signal, that is, when it is small, the adaptive control flag FLGb output from the comparing means 20 is output as the final adaptive control flag FLG. Here, FLG = 1 indicates an instruction to execute adaptation to the adaptive filter, and FLG = 0 indicates an instruction to stop adaptation. Here, the reason why the adaptive control flags FLGa and FLGb are selectively used in accordance with the amount of echo cancellation by the adaptive filter is to perform good adaptation control regardless of the magnitude of the amount of echo cancellation. As described in the first and second embodiments, when the amount of echo cancellation is large for good adaptation control, the adaptation control flag FL
When Ga is effective and the amount of echo cancellation is small, the adaptation control flag FLGb is effective.

As described above, according to the present invention, the adaptation control flags FLGa and FLG are set in accordance with the amount of echo cancellation.
Since b is used properly, regardless of the amount of echo cancellation,
Good adaptation control can be performed. The margins α and β have an effect of absorbing an error caused by approximating the near-end voice level with the residual signal level Lres (k) and an error caused by the estimated echo level being an estimated value.

Embodiment 4 A configuration and operation for eliminating generation of abnormal noise when the filter coefficient of the product-sum operation unit is switched by the adaptation control will be described. FIG. 6 is a block diagram of an echo canceller including a filter coefficient replacement unit according to Embodiment 4 of the present invention. In this figure, the description of the parts performing the same operations as in FIG. 13 is omitted. In the drawing, the switching means 12 performs the transfer according to the command when the command for transferring the value of the first filter coefficient of the adaptation memory 10 to the save memory 13 is received from the coefficient replacement control means 15. Is different from the conventional example in that when a command for replacing the value of the first filter coefficient of the adaptation memory 10 with the second filter coefficient value of the save memory 13 is received, nothing is performed. The interpolation means 23, which is a new element, receives a command from the coefficient replacement control means 15 to replace the value of the first filter coefficient of the adaptation memory 10 with the value of the save memory 13, that is, the current adaptation is stopped. When the filter coefficient is replaced improperly, the first filter coefficient H (j) of the adaptive memory 10 is compared with the second filter coefficient Hm (j) of the save memory 13, and the difference between the values is calculated. , M, and one change width DL (j), j =
1, 2,. . . , J according to equation (13), and the change width DL (j), j = 1,
2,. . . , J is saved. Here, J is the number of taps of the adaptive filter.

FIG. 7 shows the time change of Ds (k), the timing of the transfer processing by the switching means 12, and the timing of the interpolation processing by the interpolation means 23. That is, D
s (k) increases, and the timing at which the first filter coefficient H (j) of the adaptation memory 10 is transferred to the second filter coefficient Hm (j) of the save memory 13 for storage is set to “H (k). j) from Hm (j), and Ds (k) becomes smaller, and the first of the adaptation memories 10 is stored in the save memory 1 of the filter coefficient H (j).
The timing of M times of interpolation for changing to the second filter coefficient Hm (j) stored in No. 3 is indicated by a pulse of “H (j) interpolation timing”. In FIG. 14 illustrating the conventional example, the first filter coefficient H (j) of the adaptive memory 10 is replaced with the second filter coefficient Hm of the save memory 13.
While the process of changing to (j) was a single replacement,
In FIG. 8 for explaining the present invention, since the process of changing is M times of interpolation, even if the first filter coefficient and the second filter coefficient before interpolation in the present invention are significantly different,
The first filter coefficient of the adaptation memory 10 can be gradually changed. DL (j) = {Hm (j) -H (j)} / M j = 1, 2,. . . , J (13) Next, at the time of the first interpolation, the interpolation means 23 changes the first filter coefficient H (j) of the adaptation memory 10 to the change width DL.
Processing for adding the value of (j) is performed by j = 1, 2,. . . , J
Do about. Thereafter, the change width DL (j), j = j is added to the first filter coefficient of the adaptive memory 10 until the M-th interpolation.
1, 2,. . . , J are repeated. As a result, when the M-th interpolation is completed, the value of the first filter coefficient of the adaptive memory 10 becomes the second filter coefficient Hm (j), j = 1,
2,. . . , J.

Embodiment 5 Received signal level Lrin (k) and transmission signal level Ltin calculated by level calculation means 5, 16, and 6 in the first, second, and third embodiments.
(K) The residual signal level Lres (k) is a logarithmic converted value of power, but this may be substituted by power. In the present embodiment, the adaptive control means according to the first, second,
In order to perform the same operation as that of the equation (3), the calculation of the equation (8) is performed by the equation (1).
In equation (4), equation (9) is calculated in equation (15), and in conditional equation (1).
The calculation of (0) is changed to Expression (16), the calculation of Conditional Expression (11) is changed to Expression (18), and the calculation of Expression (12) is changed to Expression (20). Here, α ′ is represented by the following equation (17). Here, β ′ is represented by the following equation (19).

[0031]

(Equation 5)

Further, the equation (14) for calculating the estimated echo gain EG and the equation (2) for calculating the estimated echo cancellation amount ERLE
0) may be substituted by equations (21) and (22), respectively.

[0033]

(Equation 6)

[0034]

As described above, according to the present invention, the adaptation control means makes a decision using the estimated echo level and the residual signal level. Can accurately estimate the near-end signal level and the echo signal level, and as a result, there is an effect that the execution and stop of the adaptation of the adaptive filter can be accurately controlled.

Further, since the adaptation control means makes a determination using the estimated echo level and the transmission signal level, it is possible to determine whether or not the transmission signal is only an echo signal even if the amount of echo cancellation is small. There is an effect that execution and stop of adaptation of the adaptive filter can be controlled with high accuracy.

Further, the adaptation control means estimates the amount of echo cancellation from the level of the transmission signal and the level of the residual signal, and the adaptation control means uses different flags depending on the amount of echo cancellation. In addition, there is an effect that the execution and stop of the adaptation of the adaptive filter can be accurately controlled.

Also, at the time of filter coefficient replacement, the filter coefficients of the adaptation memory before replacement and the filter coefficients of the save memory are replaced while being interpolated. Even if the values of the filter coefficients of the save memory are significantly different, the output signal of the product-sum calculator does not become discontinuous, and as a result, there is an effect that abnormal noise does not occur in the residual signal.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an adaptation control unit according to Embodiment 1 of the present invention.

FIG. 2 is a diagram illustrating an operation of an adaptation control unit according to the first embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration of an adaptation control unit according to Embodiment 2 of the present invention.

FIG. 4 is a diagram illustrating an operation of an adaptation control unit according to the second embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of an adaptation control unit according to Embodiment 3 of the present invention.

FIG. 6 is a block diagram showing an echo canceller including a filter coefficient replacement unit according to Embodiment 4 of the present invention.

FIG. 7 is a diagram illustrating an operation of a coefficient replacement unit according to Embodiment 4 of the present invention.

FIG. 8 is a block diagram illustrating an overall configuration of an echo canceller.

FIG. 9 is a block diagram showing a configuration of a conventional adaptation control means.

FIG. 10 is a diagram illustrating an example of a section in which adaptation of an adaptive filter is to be performed.

FIG. 11 is a diagram showing an example of an execution section of adaptation of a conventional adaptive filter.

FIG. 12 is a diagram illustrating an example of an execution section of adaptation of a conventional adaptive filter.

FIG. 13 is a block diagram showing an echo canceller including a conventional filter coefficient replacement method.

FIG. 14 is a diagram illustrating the operation of a conventional coefficient replacement method.

[Explanation of symbols]

Reference Signs List 1 echo path, 2 echo subtractor, 3 adaptive filter, 4 adaptation control means, 5, 6, 16 level calculation means, 7 level difference calculation means, 8 level correction means, 9
Comparison means, 10 adaptation memory, 11 product-sum operation means,
12 switching means, 13 save memory, 14 filter coefficient updating means, 15 coefficient replacement control means, 17 echo gain estimating means, 18 echo level estimating means, 1
9, 20 comparison means, 21 echo cancellation amount estimation means, 22
Flag selection means, 23 interpolation means.

Claims (4)

[Claims] A means for estimating an echo gain from a level of a received signal and a level of a transmitted signal; and an estimated echo level obtained by adding the estimated echo gain to the received signal, and a residual signal after echo cancellation. A comparing means for comparing the residual echo signal with a margin and a margin when the estimated echo level is larger than a value obtained by adding a margin to the residual signal. Echo canceller with adaptive control means. 2. A means for estimating an echo gain from a level of a received signal and a level of a transmitted signal, and comparing an estimated echo level obtained by adding the estimated echo gain to the received signal and a level of the transmitted signal. An adaptive control means for performing an adaptive control for canceling an echo of a received signal included in a transmission signal when the estimated echo level is larger than a value obtained by subtracting a margin from the transmission signal. Equipped echo canceller. 3. A means for estimating an echo gain from a level of a received signal and a level of a transmitted signal, and an estimated echo level obtained by adding the estimated echo gain to the received signal and a residual signal after echo cancellation. First comparing means for comparing the level of the transmission signal with the level of the transmission signal; first comparison means for comparing the level of the transmission signal with an estimated echo level obtained by adding the estimated echo gain to the reception signal; And an echo canceling amount estimating means for estimating an estimated echo canceling amount from the level of the residual signal and a level of the residual signal. If the echo canceling amount is larger than a threshold, the estimated echo level adds a margin to the residual signal. If the echo cancellation amount does not exceed the threshold, the adaptive control for canceling the echo of the reception signal included in the transmission signal is performed. Echo canceller bell with adaptation control means to perform the adaptation control to cancel the echo of the reception signal included in the transmission signal greater than the value obtained by subtracting a margin from the transmission signal. 4. An adaptation memory for storing a filter coefficient for adaptation updated using an algorithm from a transmission signal and a residual signal after echo cancellation, and the adaptation memory while adaptation is determined to be valid. An evacuation memory for storing filter coefficients of a memory for a predetermined interval in parallel; and determining whether the update of the filter coefficients for adapting the adaptive filter is appropriate based on the difference between the transmission signal level and the residual signal level. Coefficient replacement control means for controlling; and interpolation means for comparing and interpolating the value of the first filter coefficient in the adaptation memory with the value of the second filter coefficient in the save memory. When the means controls the replacement of the value of the first filter coefficient with the value of the second filter coefficient when the adaptation is not appropriate, the interpolation means controls the value of the first filter coefficient and the value of the second filter coefficient. Echo canceller to be sequentially updated by interpolation at a predetermined timing number so that the number of values becomes a continuous value.