JP4705554B2 - Echo canceling apparatus, method thereof, program thereof, and recording medium thereof - Google Patents

Description

  The present invention relates to an echo canceling device such as an echo canceling device for hands-free communication such as a TV conference or an audio conference, a line echo canceling device, a method thereof, a program thereof, and a recording medium thereof.

  There is one that uses an adaptive filter as a basic technique for modeling an unknown system to which an echo is transmitted using a digital filter and estimating an echo transmission system using a received signal and a collected sound signal. Noise included in the collected sound signal is a factor that greatly degrades the estimation accuracy of the echo transmission system when the echo transmission system is estimated by the adaptive filter. In particular, the mixing of the transmitted voice from the speaker into the echo called double talk or simultaneous speech in the acoustic and line echo canceling device greatly reduces the estimation accuracy of the digital filter that had been included until the mixing, and the echo canceling device Therefore, some countermeasure is required. Hereinafter, a conventional echo canceling apparatus using an adaptive filter and a conventional countermeasure against double talk will be described.

FIG. 1 shows a functional configuration example of a conventional echo canceller. In FIG. 1, the received signal is x (k), the echo estimation signal is y (k), the collected sound signal is d (k), the error signal is e (k), and the vector composed of the coefficients of the adaptive filter is h (k). And k represents a discrete time.
A voice signal from a far-end speaker (not shown) is input to the reproducing means 4 through the echo canceller 10 as a received signal x (k). The echo canceller 10 includes an estimated signal generation unit 11, a filter coefficient update unit 12, a subtraction unit 13, and an adaptive filter coefficient register 14, and the received signal x (k) is estimated with the filter coefficient update unit 12. Input to the signal generator 11.

The estimated signal generator 11 generates an echo estimated signal y (k) by filtering the received signal x (k) with the adaptive filter coefficient h (k) from the adaptive filter coefficient register 14. There are various means for generating the echo estimation signal y (k), but when generating the echo estimation signal y (k) for one time, a vector composed of the received signal x (k) and a vector of adaptive filter coefficients It is generated by the inner product of h (k), that is, y (k) can be expressed by the following equation (1).
y (k) = [x (k−L + 1), x (k−L + 2),. . . , X (k)] h (k)
Formula (1)

However, L represents the number (the number of taps) of the coefficient of the adaptive filter which comprises the estimated signal production | generation part 11. FIG. In a general adaptive filter, when an estimation signal for a plurality of times is generated, it is often performed in the frequency domain in order to reduce the amount of calculation, but details thereof are described in Non-Patent Document 1. The subtractor 13 subtracts the echo estimation signal y (k) from the collected sound signal d (k) collected by the sound collecting means 6 to obtain an error signal e (k). That is, the subtraction unit 13 calculates the following expression (2).
e (k) = d (k) −y (k) Equation (2)

The filter coefficient updating unit 12 obtains an adaptive filter coefficient correction amount Δh from the error signal e (k) and the received signal x (k) by an adaptive algorithm. Then, the adaptive filter coefficient correction amount Δh is multiplied by the step size μ for adjusting the correction amount, and added to the adaptive filter coefficient h (k−1) one time before, so that the adaptive filter in the adaptive filter coefficient register 14 is obtained. Update the coefficient. That is, the update of the filter coefficient can be expressed by the following equation (3).
h (k) = h (k−1) + μΔh (k) Equation (3)

  The adaptive filter coefficient correction amount Δh is obtained by an adaptive algorithm so that the power of the error signal e (k) is reduced. In addition, as a typical adaptive algorithm, the least square (LMS: Least-Mean-Squares) method processed in a time domain, the learning identification (NLMS: Normalized LMS) method, and a recursive least square (RLS) are used. The law is well known. The echo estimation signal y (k) is removed from the collected sound signal d (k) by the subtracting unit 13 and transmitted to the transmission side of the reception signal x (k) as a transmission signal.

<Measures against double talk>
As an example of conventional double-talk countermeasures, means using a duplex filter will be described. This is also called Foreground / Background (hereinafter referred to as FG / BG).
FIG. 2 shows a functional configuration example of the FG / BG echo canceller 20 configured using the conventional FG / BG method. The same functional components as those in FIG. 1 are denoted by the same reference numerals, and redundant description is omitted. The same applies to the following description.
In the FG / BG echo canceller 20, first, the BG filter coefficient update unit 22 estimates the impulse response of the echo path (echo path from the reproduction means 4 to the sound collection means 6) 24, and the estimated value hb (k) is obtained. The data is transferred to the adaptive filter coefficient register 27 as an adaptive filter coefficient hb (k).

The BG estimation signal generation unit 26 performs the same filtering process on the received signal x (k) with the adaptive filter coefficient hb (k) to generate the BG estimation signal yb (k). The adaptive filter coefficient hb (k) is stored in the adaptive filter coefficient register 27.
In the subtracting unit 28, the BG estimation signal yb (k) is subtracted from the collected sound signal d (k) collected from the sound collecting means 6 to generate a BG error signal eb (k). If the impulse response of the reverberation path 24 is estimated satisfactorily, the reverberation signal collected by the sound collection means 6 from the reproduction means 4 and the estimated signal yb (k) are substantially equal.

On the other hand, the FG estimation signal generation unit 30 filters the received signal x (k) with the FG filter coefficient hf (k) to generate the FG estimation signal yf (k). The FG filter coefficient is stored in the FG filter coefficient register 29. Then, the subtraction unit 31 subtracts the FG estimation signal yf (k) from the collected sound signal d (k) to generate an FG error signal ef (k).
If the characteristics of the BG estimation signal generator 26 are close to those of the true echo path 24 (details will be described below), the adaptive filter coefficient hb (k) in the adaptive filter coefficient register 27 is transferred to the FG filter coefficient register 29. , The FG filter coefficient hf (k) is updated to the adaptive filter coefficient hb (k).

Further, the BG error signal eb (k) and the FG error signal ef (k) are input to the error comparison unit 38. In a state where the power of the received signal x (k) is equal to or higher than a predetermined value and is not double talk, the power Peb (k) of the BG error signal eb (k) is equal to the FG error signal ef (k) in the error comparison unit 38 If it is determined that the power Pef (k) is smaller, the adaptive filter coefficient hb (k) better simulates the impulse response h (k) of the actual echo path 24 than the FG filter coefficient hf (k). it seems to do. In this case, the adaptive filter coefficient hb (k) is transferred to the FG estimation signal generation unit 37 to update the FG filter coefficient hf (k).
Here, the power is a time integration value of the signal. When a discretized signal is handled, for example, Px (k) = Σx ² (ki) (Σ is i = 0 to n−1). Is calculated as follows. Here, n represents integration time.

Then, in a state where the voice of the speaker is picked up by the sound pickup means 6, the FG error signal ef (k) is output to the transmission side of the received signal x (k). Details of the FG / BG echo canceller 20 are described in Patent Document 1.
Thus, by adopting the FG / BG configuration, the adaptive filter coefficient hb (k) is disturbed at the time of double talk, and duplication to the FG filter coefficient hf (k) does not occur, and the FG filter coefficient hf (k) is before double talk. Therefore, the influence of double talk does not appear in the FG error signal ef (k).

<Differential FG / BG configuration>
Further, as a technique for reducing the influence of the disturbance of the adaptive filter coefficient hb (k) due to simultaneous calls on the FG filter coefficient hf (k) and maintaining good simultaneous call performance, the FG error signal ef (k) is used as the BG unit. There is a differential FG / BG echo canceller that estimates the amount of change in the FG filter coefficient hf (k) as the adaptive filter coefficient hb (k).
A functional configuration example of the differential FG / BG echo canceller 45 is shown in FIG. The BG unit 40 includes a BG filter coefficient update unit 22, a BG estimation signal generation unit 26, a BG subtraction unit 28, and an adaptive filter coefficient register 27. The FG unit 50 includes an FG filter coefficient register 29, an FG estimation signal generation unit 30, and an FG subtraction unit 31.

  In the BG subtraction unit 28, the FG error signal ef (k) from the FG subtraction unit 31 is input to the error comparison unit 38 and the BG subtraction unit 28. The BG subtraction unit 28 subtracts the FG error signal ef (k) from the BG estimation signal yb (k) from the BG estimation signal generation unit 26, thereby obtaining the BG error signal eb (k). eb (k) is input to the BG filter coefficient update unit 22. When the error comparing unit 38 determines that the power Peb (k) of the BG error signal eb (k) is smaller than the power Pef (k) of the FG error signal ef (k), the FG filter coefficient updating unit 52 The adaptive filter coefficient hb (k) in the adaptive filter coefficient register 27 is added or weighted to the hf (k) in the FG filter coefficient register 29, and the resulting filter coefficient is set as a new FG filter coefficient hf ( Update as k). Details of the differential FG / BG echo canceller are described in Patent Document 2.

As described above, this differential FG / BG echo canceller configuration has a small echo cancellation amount before the adaptive filter coefficient hb (k) is reflected in the FG filter coefficient hf (k) as the adaptive algorithm of the BG unit. There is a feature that it only needs to be obtained.
Japanese Patent No. 3248551, FIG. No. 2836277 “Multidelay block frequency domain adaptive filter,” Jia Sien Soo and Khee K. Pang, IEEE Transaction on Acoustics, Speech, and Signal Processing, vol. 38, No. 2, pp. 373-376 (1990.2) "Frequency-domain and mulirate adaptive filtering," JohnSynk, IEEE SPmagazin, Jabuary 1992 “Connecting Partitioned Frequency-Domain Filters in Parallel or in Cascade, section I, II, III,” M. Joho and GSMoschytz, vol 47.pp. 685-698 (2000.8)

  The conventional differential FG / BG echo canceller described above has a problem that the amount of calculation is large when the number of adaptive filter coefficients is large. Therefore, it was considered that the above calculation is performed not in the time domain but in the frequency domain. In the case of the FG / BG type echo canceller operating in the frequency domain, the constraint condition with the adaptive filter coefficient hb (k) is added when the adaptive filter coefficient hb (k) of the BG unit 40 is added to the FG filter coefficient hf (k). It is necessary to perform addition after conversion to satisfy. The constraint condition is that the result of the filter process performed in the frequency domain in the FG unit 50 is made to coincide with the result when the filter process is performed in the time domain.

  As a specific processing procedure for satisfying this constraint condition, the adaptive filter coefficient h in the frequency domain is converted to the time domain, and more than half of the result is replaced with 0, and then converted to the frequency domain again. It is. This is generally considered to be similar to the case of updating the adaptive filter coefficient when processing in the frequency domain in the adaptive filter, and which element of the adaptive filter coefficient converted into the time domain is replaced with 0. The detailed description of these is described in Non-Patent Document 1 or Non-Patent Document 2. When the FG filter coefficient is a value in the time domain, the conversion to the frequency domain is unnecessary and a coefficient that is not replaced with 0 may be used.

  However, as described above, in the case of the differential FG / BG configuration in which the BG estimation signal generation unit 26 of the BG unit 40 is operated in the frequency domain, the adaptive filter coefficient hb (k) due to double talk or noise superimposed on the echo. As a result, the situation of greatly deviating from the value satisfying the constraint condition may continue. In this situation, the individual elements of the adaptive filter coefficients do not satisfy the constraint, but are updated so that the temporally continuous adaptive filter coefficients cancel out portions where they do not satisfy each other, thus the BG error signal. The level of eb (k) decreases. Therefore, an operation of adding the adaptive filter coefficient hb (k) to the FG filter coefficient hf (k) is performed. At this time, the update of the filter coefficient hf (k) is performed so that the FG error signal ef (k) decreases. However, it is considered that there is a possibility that the level of the FG error signal ef (k) increases with time. In order to solve this problem, in the BG filter coefficient updating unit 22 that processes in the frequency domain, the adaptive filter coefficient hb (k) is set in the time domain in order for the adaptive filter coefficient hb (k) to satisfy the constraint condition. A problem arises in that the amount of computation increases because it must be converted.

In the echo canceling apparatus according to the present invention, the FG estimation signal generation unit filters the received signal with the FG filter coefficient to obtain the FG estimation signal, and the FG subtraction unit subtracts the FG estimation signal from the collected sound signal to generate an FG error signal. FG part for
A BG estimation signal generation unit filters the received signal with an adaptive filter coefficient to generate a BG estimation signal, and an error calculation unit subtracts the BG estimation signal from the FG error signal to obtain a BG error signal, and a BG filter A BG unit that updates the adaptive filter coefficient so that a coefficient updating unit reduces the power of the BG error signal using the BG error signal and the received signal;
An echo cancellation apparatus comprising: an FG filter coefficient updating unit that adds the adaptive filter coefficient to the FG filter coefficient and updates the FG filter coefficient;
The BG section performs processing in the frequency domain,
When the adaptive filter coefficient in the frequency domain is converted into the time domain, the result of cyclic convolution with the filter coefficient of the band rejection filter is such that the first coefficient or the coefficient one half behind is 0, and the others remain unchanged A partial constraint processing unit for storing in the adaptive filter coefficient register as an updated adaptive filter coefficient by performing frequency domain processing on the data in at least one of the processes of updating the adaptive filter coefficient To do.

  According to this configuration, the problem that the level of the FG error signal increases with the update of the FG adaptive filter coefficient is solved by simple processing by the partial constraint processing unit. In addition, since the processing for satisfying the constraint condition is not performed in the adaptive filter update processing in the BG unit, the problem that the amount of calculation increases is solved.

  The best mode for carrying out the present invention will be described below. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated.

FIG. 4 shows a functional configuration example of the echo canceller 120 of the first embodiment in which the present invention is applied to the UMDF method to be described later, and FIG. 5 shows the processing flow. In this invention, at least the BG unit performs processing in the frequency domain, and the FG unit can perform processing in the frequency domain or the time domain. In general, processing performed in the time domain has a small amount of delay but a large amount of calculation. Therefore, the processing of the FG unit may be performed depending on whether it is performed in the frequency domain or in the time domain.
In the first embodiment, the BG filter coefficient updating unit 90 executes the above-described LMS method or NLMS (Normalized LMS) method in the frequency domain, and the MDF method (MultiDelay block Frequency domain adaptive filter) described in Non-Patent Document 1. Alternatively, the UMDF method (Unconstrained MultiDelay block Frequency domain adaptive filter) is targeted.

Here, the time domain adaptive filter coefficient hb used for the frequency domain adaptive filter processing in the BG unit 60 is divided into J for each continuous N / 2 element (N will be described later), and the divided adaptive filter is divided into the frequency domain. This will be described based on the UMDF method, which is an adaptive algorithm with a small amount of calculation to be updated.
Even when the received signal x (k) has a plurality of channels, the following matters can be applied only by increasing the elements of the received signal x (k) and the adaptive filter coefficient h (k) by the number of channels. To do. In the first embodiment, a case where the FG unit 70 performs processing in the frequency domain will be described.

Processing of FG Unit 70 Hereinafter, processing of each unit will be described with reference to FIG. In the FG unit 70, when a time domain received signal x (k) and a sound pickup signal d (k) are newly obtained (FIG. 5, step S2), the time domain received signal x (k) is , And input to the frequency domain converter 62. In the frequency domain transforming unit 62, for example, N discrete frequency domain signals corresponding to N frequencies from ω1 to ωN, x (ω1),. x (ωn),..., x (ωN). However, the received signal x (k) is a digital signal which is sampled at a constant period and each sample is converted into a digital value. This N is a positive integer and is the same as N / 2 of N / 2 when the adaptive filter coefficient hb in the time domain is divided.

The frequency domain transforming unit 62 divides the time domain adaptive filter coefficient hb into N / 2 elements for each of N / 2 elements, and the frequency domain reception corresponding to the jth division h _j (j = 0, 1,..., J−1). The signal _Xj is generated by equation (4) (step S4). Henceforth, the code | symbol of the signal of a frequency domain is represented by capital letters.
X _j = F [x (k−j × (N / 2) −N + 1), x (k−j × (N / 2) −N + 2),..., X (k−j × (N / 2))] ^T formula (4)
Here, F is a Fourier transform, N is the number of samples in one frame (frame shift is N / 2), and ^T is a transposed matrix.

The frequency domain received signal X _j is input to the FG unit 70. As illustrated in FIG. 4, the FG unit 70 includes an FG filter coefficient register 74, an FG estimation signal generation unit 76, an FG time domain conversion unit 78, and an FG subtraction unit 31. The received signal X _j is input to the FG estimation signal generator 76.
The FG estimation signal generation unit 76 performs an operation corresponding to the filtering process using the received signal x (k) and the adaptive filter coefficient h (k) in the time domain in the frequency domain. That is, the frequency domain estimation signal Yf is obtained by taking the inner product of the received signal X _j and the frequency domain adaptive filter coefficient Hf _j recorded in the FG filter coefficient register 74 for each frequency. The frequency domain estimation signal Yf is calculated by Expression (5).
Yf = Σdiag (X _j ) (H _j ) Equation (5)
Here, diag (A) represents the diagonal matrix of A.

The frequency domain estimation signal Yf is input to the time domain conversion unit 78 and converted into a time domain estimation signal yf (k) by a short-time discrete Fourier transform, which is a known technique (step 6).
yf (k) = [y (k−N / 2 + 1), y (k−N / 2 + 2),..., y (k)] ^T = WF ⁻¹ Yf
Formula (6)
Here, W represents a matrix of N / 2 rows and N columns that extracts the lower half of the multiplied vector when the vector is multiplied from the right of W, and F ^-1 represents the inverse Fourier transform. That is, the FG time domain conversion unit 78 converts yf into a time domain signal, and then extracts the lower half elements from the converted time domain signal by N / 2 rows and N columns (step S6).

Each element yf (ki), (i = 0, 1,..., N / 2) of the time domain estimation signal yf (k) is input to the FG subtraction unit 31. The FG subtractor 31 subtracts the estimated signal y (ki) from the collected sound signal d (ki) to obtain an error signal ef (ki) (step S8).
ef (ki) = d (ki) -yf (ki), i = 0, 1,..., N / 2-1
Formula (7)
This FG error signal ef (ki) becomes the output of the echo canceling device. This time domain FG error signal ef (ki) is input to the far-end speaker and the BG unit 60 through a communication network (not shown).

Processing of the BG unit 60 Next, the operation of the BG unit 60 will be described. As shown in FIG. 4, the BG unit 60 includes a BG filter coefficient update unit 90, a partial constraint processing unit 67, an adaptive filter coefficient register 68, a BG estimation signal generation unit 64, a frequency domain error calculation unit 61, and a coefficient update determination unit 96. It consists of and.
On the other hand, in the BG unit 60, the frequency domain estimation signal is obtained by taking the inner product of the adaptive filter coefficient HB _j (hereinafter referred to as adaptive filter coefficient Hb _j ) recorded in the adaptive filter coefficient register 68 and the received signal X _j for each frequency. Yb (hereinafter referred to as BG estimation signal Yb) is obtained (step S10). In other words, the BG estimation signal generation unit 64 calculates the BG estimation signal Yb according to the equation (8).
Yb = Σj (diag (X _i ) Hb _j ) Equation (8)

The FG error signal ef (k) is input to the BG frequency domain converter 61a of the frequency domain error calculator 61 and converted into a frequency domain FG error signal Ef _j (hereinafter referred to as FG error signal Ef _j ). That is, N / 2 error signals ef (ki) from the FG unit 70, i = 0, 1,..., N / 2-1 are preceded by N / 2 zeros to obtain the equation (9). Calculated by
Ef _j = F [0, 0,..., 0, ef (k−N / 2 + 1), ef (k−N / 2 + 2),..., Ef (k)] ^T equation (9)

The BG estimation signal Yb is input to the BG subtraction unit 61 b of the frequency domain error calculation unit 61. The subtraction unit 61b, obtains the BG error signal Eb _j by subtracting the corresponding elements of the BG estimated signal Yb from each element of the FG error signal Ef _j (step S12). That is, formula (10) is calculated.
Eb = Ef−GYb Formula (10)
However, G is extracted elements of the lower half of the vectors multiplied when multiplied by a vector W _N from right W _N, FW _N F ^-1 to or itself upper half as a matrix of N rows and N columns to 0 Is a matrix that approximates

The BG error signal Eb is input to the BG filter coefficient update unit 90. The BG filter coefficient update unit 90 obtains ΔHb _j that is the update amount of the applied filter coefficient in the frequency domain so that the power of the BG error signal Eb decreases using the BG error signal Eb and the received signal X _j (step S14). A specific method for obtaining ΔHb _j will be described later.
Here, the reason why the FG estimation error increases when the portions of the adaptive filter coefficients that are temporally continuous, which are the above-described problems, cancel each other out of the constraint conditions will be described.

First, the convolution in the frequency domain according to equations (8) and (6) will be described. What extracted only the thing regarding j on the right side to Formula (8) is represented by Formula (11),
Yb _j = diag (X _j ) Hb _j expression (11)
This converted into the time domain as shown in equation (6) is expressed as equation (12).
yb _j = WF ⁻¹ Yb _j expression (12)
yb _j is N received signals x (k−j × (N / 2) −N + 1), x (k−j × (N / 2) −N + 2),..., x (k−j × (N / 2)) and F ⁻¹ Hb _{j obtained} by converting the j-th divided Hb _j of the adaptive filter in the frequency domain into the time domain is cyclically convolved, and, for example, N / 2 pieces in the upper half are extracted. When the result yb _j of this cyclic convolution is summed with respect to the division number j, an estimated signal yb (k) is obtained.

If the lower half of F ⁻¹ Hb _j is 0 like the filter coefficient on the FG side, yb _j is N−1 received signals x (k−j × (N / 2) −N + 2). ), X (k−j × (N / 2) −N + 3),..., X (k−j × (N / 2)) and j-th division Hb _j of the frequency domain adaptive filter are converted into the time domain. This coincides with the result of linear convolution in the range where hb _j (= F ⁻¹ Hb _j ) and x overlap.

It should be noted in the convolution in the frequency domain according to the above formulas (8) and (12) that if the lower half of F ⁻¹ Hb _j is 0, yb _j has x (k−j × (N / 2) -N + 1) but does not affect, unless the lower half of ^F -1 Hb _j is 0 so the adaptive filter coefficients is the case when using UMDF method BG side, the yb _j x (k −j × (N / 2) −N + 1) has an effect.

On the other hand, in the data x (k−j × (N / 2) −N + 1), k−j × (N / 2) −N + 1 is k− (j + 1) × (N / 2) − (N / 2). ) As can be seen from the fact that it can be transformed to +1, it exists in the N / 2th received signal sequence obtained by converting X _{j + 1} into the time domain. Therefore, in the MDF method, the division of the adaptive filter coefficient corresponding to X _{j + 1} is only Hb _{j + 1,} but in the UMDF method, the influence of X _{j + 1} appears on the adjacent division Hb _j . Then, in the update of the adaptive filter, Hb _j and Hb _{j + 1} are expressed as diag (X _{j + 1} ) Hb _{j + 1} is the influence of X _{j + 1} that appears in the division Hb _j in Yb obtained by summing up the division numbers as in Expression (8). Updated to negate.

In the time domain, the influence of X _{j + 1 on} the divided Hb _j appears at the beginning of hb _j (that is, the position corresponding to the newest signal) h _j0, and the coefficient in the time domain of H _{j + 1} that cancels this is N of h _{j + 1} / Appears at h _{j + 1 N / 2} in the second position.
By the way, when the adaptive filter coefficient Hb _j on the BG side is transferred to the filter Hf _j on the FG side, constraint condition processing is performed in which Hb _j is converted to the time domain, and the first half is converted to 0 again. During this constraint processing, the head of hb _{j that} has an effect on the divided Hb _j of X _{j + 1} is removed, but h _{j + 1 N / 2} that has canceled it is transferred to Hf _j as it is. In this case, the values that were canceled when the sum of Expression (8) was calculated on the BG side appear as errors on the FG side and do not cancel each other, and the estimation error increases.

To prevent an increase in the estimated error to this h _j0 the influence of X _{j + 1} appear cause is found to the h _j0 may be set to 0.
Depending on how the data is arranged, it is necessary to set the N / 2th coefficient of h _j to 0. The first coefficient of h _j 0, or an N / 2-th coefficient of h _j is called a handle portion constraint processing to zero. In the first embodiment, the partial constraint process is performed on the adaptive filter coefficient HB _j updated by the BG filter coefficient update unit 90 (step S16).

The coefficients of the band rejection filter B that realizes the partial constraint processing unit 67 that _{sets the} 0th coefficient h _j0 of the partial constraint processing 0 and keeps other coefficients as follows are B = [1-1 / N, −1 / N, ..., -1 / N]. Here, the coefficient of -1 / N is (N-1). The operation for cyclically convolving the band elimination filter B and the adaptive filter coefficient Hb _j of the BG unit 60 may be performed as follows, for example. As shown in FIG. 6a, the average value of each element of the adaptive filter coefficient Hb _j of the BG unit is obtained by the H _j average value calculating part 671a, and the average value AH _j is obtained by each element Hb of the adaptive filter coefficient Hb _j of the BG part 60. subtract from _ji . That is, the calculation shown in Formula (13) is performed by the coefficient calculation unit 672a.
H _{j, i} = H _{j, i} −Σi (H _{j, i} ) / N, i = 0, 1, 2,..., N−1
Formula (13)
This partial constraint processing does not require such a large amount of calculation.

Further, the filter coefficients of the band rejection filter B that realizes the partial constraint processing unit 67 for setting the N / 2th coefficient of h _j to 0 are B = [1-1 / N, 1 / N, −1 / N, 1 / N, -1 / N, ..., 1 / N]. The number of coefficients is N. Calculation of circular convolution of the adaptive filter coefficients H _j of the band rejection filter B and BG unit may be performed, for example, as follows. Determined as shown in equation (14) the average value of the difference between the even-numbered and odd-numbered coefficients of the adaptive filter coefficients H _j of the BG portion at H _j average value calculating section 671b as shown in Figure 6b, the average value Htmp Is subtracted from or added to the adaptive filter coefficient Hb _j of the BG part. That is, equation (15) or equation (16) may be calculated by the coefficient calculation unit 672b.
Htmp = Σi ((H _{j, 2 × i} ) − (H _{j, 2 × i + 1} )) / N, i = 0, 1, 2,..., N / 2-1
Formula (14)
H _{j, i} = H _{j, i} −Htmp, i = 0, 2, 4,..., N-2 equation (15)
H _{j, i} = H _{j, i} −Htmp, i = 1, 3, 5,..., N−1 Expression (16)

Whether the expression (13), the expression (15), or the expression (16) is used as the partial constraint processing depends on whether the BG frequency domain conversion unit 61a of the frequency domain error calculation unit 61 uses ef (k) as a frequency domain signal. Depending on the processing method when converting to. When the old data side is converted to the frequency domain with 0, the coefficient one backward from the half of the adaptive filter coefficient h _j converted to the time domain is set to 0.
On the other hand, when the BG frequency domain transform unit 61a sets the new data side to 0, the first coefficient obtained by converting the frequency domain adaptive filter coefficient to the time domain is set to 0. The adaptive filter coefficient Hb _j ′ subjected to the partial constraint processing is overwritten in the adaptive filter coefficient register 68.

FG Filter Coefficient Update Processing Next, a specific configuration example of the FG filter coefficient update unit 72 will be described with reference to FIG. The FG filter coefficient updating unit 72 adds the adaptive filter coefficient Hb _j to the FG filter coefficient Hf _j , and updates the FG filter coefficient Hf _j (step S18). In addition, when the adaptive filter coefficient Hb _j is added to the FG filter coefficient Hf _j in the frequency domain, a constraint condition described later must be satisfied. As shown in FIG. 7, in this embodiment, the FG filter coefficient updating unit 72 includes a time domain conversion unit 802, a half-zeroing unit 804, a frequency domain conversion unit 806, and an addition unit 808. Note that the FG filter coefficient update process of the FG filter coefficient update unit 72 is activated by filling the elements of the frequency ωn that have not been determined to update the FG filter coefficient by the coefficient update determination unit 96 with zeros.

The adaptive filter coefficient Hb _j ′ recorded in the adaptive filter coefficient register 68 is converted into an adaptive filter coefficient hb (k) in the time domain by the time domain conversion means 802.
The adaptive filter coefficient hb (k) is input to the half-zeroing means 804, and the half-zeroing means 804 sets zero or more of the elements of the adaptive filter coefficient hb (k) to zero. Which element of the adaptive filter coefficient converted to the time domain is replaced with 0 is described in Non-Patent Document 1 or Reference. The adaptive filter coefficient hb (k) ′ in which more than half of the elements of the adaptive filter coefficient hb (k) are set to 0 is input to the frequency domain transforming unit 806 and converted again into the frequency domain, and the adaptive filter coefficient Hb _j ″ The frequency domain adaptive filter coefficient Hb _j ″ and the FG filter coefficient Hf _j stored in the FG filter coefficient register 701 are input to the adding means 808.

In addition means 808, are added to the corresponding elements of each element and Hb _j "of Hf _j, FG filter coefficient Hf _j are determined. Newly determined FG filter coefficient Hf _j are stored in the FG filter coefficient register 74 The FG filter coefficients are updated in this manner, and the processes of the FG unit 70 and the BG unit 60 are repeated.
Next, a description will be given based on the PFDLMS method (Partitioned Frequency Domain Least Mean Square) that constrains all the divided Js over a longer time than the adaptive filter coefficient hb.

FIG. 8 shows an example in which the PFDLMS method is applied to the first embodiment. In FIG. 8, the partial constraint processing unit 67, which is a main part of the present invention, is omitted, and the part is shown in FIG.
In the PFDLMS method, for the purpose of reducing the amount of calculation, a constraint condition processing command that imposes constraint condition processing only on the adaptive filter coefficient HB _j corresponding to the division j where learning of the adaptive filter coefficient HB _j has advanced is output. The selection is performed by the selective constraint condition application unit 66 to which the FG error signal Ef _j and the BG error signal Eb _j are input. The selection is determined when the value obtained by subtracting the power PEf _ωn of the frequency ωn of the BG error signal Eb _j from the power PEf _ωn of the frequency ωn of the FG error signal Ef _j is larger than a certain threshold value α, that is, Expression (17) Depending on whether or not

PEf _ωn− PEb _ωn ≧ α Formula (17)
If the power difference is large to some extent, the selective constraint condition application unit 66 determines that learning of the adaptive filter coefficient HB _j has progressed, and the constraint condition processing is performed so as to impose a constraint on the adaptive filter coefficient HB _j corresponding to the division j. Outputs a command.
The selection unit 80 to which the constraint condition processing command is input outputs the adaptive filter coefficient HB _j corresponding to the selected division _j to the constraint condition application unit 69.
The selection judgment, the power PYB _.omega.n frequency .omega.n of BG estimated signal Yb than BG estimation signal generation unit 64, and the power PEb _.omega.n frequency .omega.n of BG error signal Eb _j, it may be performed using. In this case, the value obtained by subtracting the PEb _.omega.n from PYB _.omega.n is, it can be determined that the progress in learning greater than the threshold value beta. That is, if Expression (18) is satisfied, a constraint condition processing command is output so as to impose a constraint on the adaptive filter coefficient HB _j corresponding to the division j, as described above.

This is because if the learning of the applied filter coefficient Hb _j sufficiently proceeds, the BG estimation signal Yb becomes a value close to the FG error signal Ef _j .
PYb _ωn− PEb _ωn ≧ β Formula (18)
The constraint condition is to match the result of the filter process performed in the frequency domain with the result of the filter process performed in the time domain. As a specific processing procedure for satisfying this constraint condition, as shown in FIG. 9, the adaptive filter coefficient Hb _j in the frequency domain is converted into the time domain by the time domain conversion means 691, and half of the result is reduced to half 0. After the value is replaced with 0 by the converting means 692, it is converted again to the frequency domain by the frequency domain converting means 693, which is calculated by the equation (19).
Hb _j ′ = FW′F ⁻¹ Hb _j formula (19)
Here, W ′ represents an N-row N-column matrix that extracts the upper half of the multiplied vector when the vector is multiplied from the right of W ′.
The applied filter coefficient Hb _j subjected to the constraint condition processing, that is, the converged applied filter coefficient Hb _j is input to the applied filter coefficient register 68, and the applied filter coefficient is updated.

In the PFDLMS method, the initial value of the non-selected frequency domain adaptive filter coefficient H ^to b _j or the adaptive filter coefficient Hb _j that is not selected by the selection unit 80 is set to 0, and the adaptive filter coefficient Hb _j is added or overwritten. Partial constraint processing may be imposed on either or both of the update amounts ΔHb _j of the filter coefficient Hb _j .
Here, how to obtain ΔHb _j which is the update amount of the applied filter coefficient in the frequency domain will be described. The calculation method of ΔHb _{j is} an example in which the NLMS method is performed in the frequency domain. The received signal X _j is input to the multiplication unit 904 and the power calculation unit 902 in the BG filter coefficient update unit 90. The power calculator 902 calculates the power of the received signal _Xj , that is, the following equation (20).
Σdiag (X _j ) diag (conj (X _j )) Equation (20)
However, conj (A) indicates that a complex conjugate is taken for each element of a scalar, vector, or matrix in A.
Further, the BG error signal Eb _j in the frequency domain obtained by the BG subtraction unit 94 is input to the multiplication unit 904.

Multiplier 904 multiplies received signal X _j and error signal Eb _j , that is, the following equation (21) is calculated.
diag (conj (X _j )) Eb _j expression (21)
Results and _{diag (conj (X j))} Eb j is a multiplier 904, a Σdiag the result of the power calculation unit _{902 (X j) diag (conj} (X j)) , but is input to the division unit 906, The following equation (22) is calculated, and ΔHb _j which is the update amount of the adaptive filter coefficient in the frequency domain is obtained.
ΔHb _j = (Σ (diag (X _j ) diag (conj (X _j ))) ^{− 1} diag (conj (X _j )) Eb _j equation (22)

A partial constraint processing unit 672 that imposes a partial constraint process on the update amount ΔHb _j may be provided between the division unit 906 and the multiplication unit 908. In that case, in the calculations of the above-described equations (13) to (16), equation (13) corresponds to equation (23) and equation (26) below in order.
ΔHb _{j, i} = ΔHb _{j, i} −Σi (ΔHb _{j, i} ) / N Equation (23)
ΔHbtmp = Σi (ΔHb _{j, 2 × i} −ΔHb _{j, 2 × i + 1} ) / N, i
= 0, 1, 2,..., N / 2-1 Equation (24)
ΔHb _{j, i} = ΔHb _{j, i} −ΔHbtmp, i = 0, 2, 4,..., N−2
Formula (25)
ΔHb _{j, i} = ΔHb _{j, i} −ΔHbtmp, i = 1, 3, 5,..., N−1
Formula (26)
ΔHb _j is input to the multiplier 908.

The step size selection unit 909 selects the step size S of the frequency band manually or by a control unit (not shown). Here, S means a matrix having the step size of each frequency band as a diagonal element. The selected step size S is input to the multiplication unit 908. Multiplier 908 multiplies step size S by ΔHb _j , that is, the following equation (27) is calculated.
SΔHb _j equation (27)
This result is input to the adding unit 910. Adder 910 adds Hb _j−1 stored in adaptive filter coefficient register 912 to SΔHb _j . That is, the following equation (28) is calculated, and the previous adaptive filter coefficient Hb _j-1 is updated based on the result.
Hb _j = Hb _j−1 + SΔHb _j Expression (28)
That is, the adaptive filter coefficients in the frequency domain are updated by the equations (20) to (22), the equations (24), (25), (26), and the following equation (29) (step S6).
Hb _j = Hb _j−1 + S (Σ (diag (X _j )) diag (conj (X _j )))) ^{− 1} diag (conj (X _j )) Eb _j equation (29)

The adaptive filter coefficient Hb _j updated by the adding unit 910 is subjected to constraint condition processing only for the division j determined by the selective constraint condition application unit 66 that learning has progressed as described above.
Then, the non-selected frequency domain adaptive filter coefficients H ^to b _j that have not received the constraint condition processing command are supplied to the partial constraint processing unit 671 by the selection unit 80, and the partial constraint process is performed. The processing for the non-selected frequency domain adaptive filter coefficients H ^to b _{j in} this case is shown in equations (30) to (33).
^{_{H ~ b j, i = H}} ~ b j, i -Σi (H ~ b j, i) / N formula (30)
^{H ~ btmp = Σi (H ~} b j, 2 × i -H ~ b j, 2 × i + 1) / N, i
= 0, 1, 2, ..., N / 2-1 Formula (31)
H ^to bj _{, i} = H ^to bj _{, i−} H ^to btmp, i = 0, 2, 4,..., N−2
Formula (32)
H ^to bj _{, i} = H ^to bj _{, i−} H ^to btmp, i = 1, 3, 5,..., N−1
Formula (33)

The frequency domain adaptive filter Hb _j , the update amount ΔHb _{j of the} frequency domain adaptive filter coefficient H, and the non-selected frequency domain adaptive filters H ^to b _j described above are obtained in the process of updating the adaptive filter coefficient. Partial constraints may be imposed. It may be imposed on a plurality.
As shown in FIG. 8, when a partial constraint processing unit is provided for each filter coefficient and no partial constraint is imposed, the filter coefficients of the band rejection filters constituting each partial constraint processing unit are set to B = [1, 0, 0, ..., 0].

As a modification of the first embodiment, a method of obtaining a difference between signals in the frequency domain by the FG subtraction unit 31 of the FG unit is also conceivable. As shown by a broken line in FIG. 4, an FG frequency domain conversion unit 705 is provided in the subsequent stage of the FG time domain conversion unit 78 in the FG unit 70 so as to satisfy the constraint condition, and an FG estimation signal Yf as an output thereof is provided. Is input to the FG subtraction unit 31. An FG frequency domain conversion unit 708 (shown by a broken line in FIG. 4) is provided for converting the time domain sound pickup signal d (k) into the frequency domain and inputting it to the FG subtraction unit 31. The FG subtraction unit 31 subtracts the corresponding element of the FG estimation signal Yf from each element of the frequency domain sound pickup signal D _j to obtain the frequency domain FG error signal Ef _j . The FG error signal Ef _j is input to the BG subtraction unit 61b and the selective constraint condition application unit 66 in the BG unit 60, as indicated by a broken line in FIG. In this case, the BG frequency domain converter 61a is omitted. Since the FG error signal Ef _j is a frequency domain signal, a time domain conversion unit 710 (indicated by a broken line in FIG. 4) for converting the FG error signal Ef _j to the time domain is provided to provide an FG error signal in the time domain. ef (k) is output to the communication network.

In the second embodiment, the FG estimation signal generation unit 76 in the FG unit 70 performs processing in the time domain. Further, in FIG. 4, performing time domain processing is indicated by parentheses with a dash “′” added to the same number. The FG time domain conversion unit 78 in the FG unit 70 is not provided.
In this case, the received signal x (k) is not input to the frequency domain conversion unit 62 but is directly input to the FG estimation signal generation unit 76 ′ as indicated by a broken line in FIG. The FG filter coefficient register 74 ′ stores the FG filter coefficient hf (k) in the time domain. In the FG estimation signal generation unit 76 ′, the FG estimation signal yf (k) is obtained from the left side of the equation in which h (k) is replaced with hf (k) in the above equation (1).

In the second embodiment, as shown in FIG. 7, the FG filter coefficient updating unit 72 includes a time domain conversion unit 802 and an addition unit 808 ′.
Hb _j ′ stored in the frequency domain in the adaptive filter coefficient register 68 in the BG unit 60 is input to the time domain conversion means 802, converted into time domain adaptive filter coefficient hb (k) ′, and replaced with 0. The data is input to the adding means 808 ′ without being processed. In the adding means 808 ′, hf (k) and hb (k) are added. The added value hf (k) is stored in the FG filter coefficient register 74 ′, and the FG filter coefficient is updated.

Simulation Results FIG. 11 shows a result of comparison between the differential FG / BG echo canceling apparatus that has performed the partial constraint processing of the present invention and the conventional method. The horizontal axis in FIG. 11 represents the time (seconds) after the start of the operation, and the vertical axis represents the coefficient error in dB.
Definition The factor error is an amount representing the difference between the transmission system and the BG portion filter coefficients h, factor error C when the transmission system is expressed as h ₀ by C = ‖h-h ₀ || ^{2 / ‖h} ₀ || ² Is done.

Simulation condition “Number of FFT points of FG unit 70: 256, number of FFT points of BG unit 60: 512, filter length per frequency band converted from time domain adaptive filter coefficient of FG unit 70 to frequency domain: 12, received signal : Male voice, sound collection signal: Received signal + female voice + background noise, change in transmission system: Simulate the system change by reversing the positive and negative of the transmission system approximately 3 seconds after the operation starts.
For example, after changing the transmission system such as shifting the position of the microphone (sound collecting means 6) about 3 seconds after the operation starts, the coefficient error starts to converge in about 7 seconds. It can be seen that the coefficient error of 20 seconds after the start of operation is about 2-3 dB better with the present invention than with the conventional method.

In addition to the above embodiments, the echo canceling apparatus according to the present invention is not limited to the above-described embodiments, and can be appropriately changed without departing from the spirit of the present invention. Further, the processes described in the echo canceling apparatus are not only executed in time series in the order described, but may be executed in parallel or individually as required by the processing capability of the apparatus that executes the processes or as necessary. .
Further, when the processing in the echo cancellation apparatus of the present invention is realized by a computer, the processing contents of the functions that the echo cancellation apparatus should have are described by a program. Then, by executing this program on the computer, the processing function in the echo canceling apparatus is realized on the computer.

The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape, or the like, and an optical disk such as a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only Memory), a CD-R (Recordable). ) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto-Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable Read Only Memory), etc. can be used.
The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its own storage device. When executing the process, the computer reads a program stored in its own recording medium and executes a process according to the read program. As another execution form of the program, the computer may directly read the program from a portable recording medium and execute processing according to the program, and the program is transferred from the server computer to the computer. Each time, the processing according to the received program may be executed sequentially. Further, the above-described processing may be executed by a so-called ASP (Application Service Provider) type service that realizes a processing function only by an execution instruction and result acquisition without transferring a program from the server computer to the computer. Good. Note that the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

  In this embodiment, the echo canceling apparatus is configured by executing a predetermined program on a computer. However, at least a part of these processing contents may be realized by hardware.

The block diagram which shows the function structural example of the echo canceller 10 of a prior art. The block diagram which shows the function structural example of the echo canceller 20 of the FG / BG structure of a prior art. The block diagram which shows the function structural example of the echo canceller 45 of a difference type FG / BG structure of a prior art. 1 is a block diagram showing a functional configuration example of an echo canceller 120 according to Embodiment 1 of the present invention. FIG. 3 is a diagram illustrating a flow of processing according to the first embodiment. 6A is a block diagram illustrating an example of a functional configuration of a partial constraint processing unit that sets the 0th coefficient h _j to 0, and FIG. 6B illustrates an example of a functional configuration of a partial constraint processing unit that sets the N / 2th coefficient h _j to 0. FIG. The block diagram which shows the function structural example of the FG filter coefficient update part 72. FIG. 1 is a block diagram illustrating a functional configuration example in which a PFDLMS method is applied to Embodiment 1. FIG. FIG. 3 is a block diagram illustrating a functional configuration example of a constraint condition application unit 69 according to the first embodiment. FIG. 9 is a block diagram illustrating a functional configuration example of the BG filter coefficient update unit 90 and the partial constraint processing unit 67 in FIG. 8. The figure which shows the simulation result about a coefficient error.

Claims (10)

An FG estimation signal generation unit that obtains an FG estimation signal by filtering the received signal with an FG filter coefficient;
An FG subtracting unit that subtracts the FG estimation signal from the collected sound signal to obtain an FG error signal;
A BG estimation signal generator for filtering the received signal with an adaptive filter coefficient to generate a BG estimation signal;
An error calculator that subtracts the BG estimation signal from the FG error signal to obtain a BG error signal;
A BG unit comprising: a BG filter coefficient updating unit that updates the adaptive filter coefficient so that the power of the BG error signal decreases using the BG error signal and the received signal;
An echo cancellation apparatus comprising: an FG filter coefficient updating unit that adds the adaptive filter coefficient to the FG filter coefficient and updates the FG filter coefficient;
The BG section performs processing in the frequency domain,
When the adaptive filter coefficient in the frequency domain is converted into the time domain, the result of cyclic convolution with the filter coefficient of the band rejection filter is such that the first coefficient or the coefficient one half behind is 0, and the others remain unchanged A partial constraint processing unit for storing in the adaptive filter coefficient register as an updated adaptive filter coefficient by performing frequency domain processing on the data in at least one of the processes of updating the adaptive filter coefficient An echo canceling device characterized by: The echo cancellation apparatus according to claim 1,
The BG unit includes the FG error signal,
A selective constraint application unit that selects a frequency domain adaptive filter coefficient converged from the BG error signal or the BG estimation signal and outputs a constraint command;
A constraint condition application unit that performs constraint condition processing on the selected frequency domain adaptive filter coefficient that has received the constraint condition processing command;
With
The echo cancellation apparatus according to claim 1, wherein the data in at least one of the partial constraint processing units is a non-selected frequency domain adaptive filter coefficient that has not received the constraint condition processing command. The echo cancellation apparatus according to claim 1,
The BG filter coefficient update unit includes an update amount calculation unit that obtains an update amount using the BG error signal and the received signal,
An addition unit that updates the adaptive filter coefficient by adding the update amount and the previous adaptive filter coefficient,
The echo cancellation apparatus according to claim 1, wherein the data in the at least one of the partial constraint processing units is the update amount. In the echo cancellation apparatus in any one of Claims 1 thru | or 3,
The partial constraint processing unit includes an average value calculation unit for obtaining an average value of adaptive filter coefficients in the frequency domain,
A coefficient calculation unit for subtracting the average value from each of the adaptive filter coefficients to obtain the updated adaptive filter coefficient,
Alternatively, an echo canceling unit comprising: a coefficient calculation unit that subtracts or adds each of the even-numbered and odd-numbered adaptive filter coefficients to the average value of the errors to obtain the updated adaptive filter coefficient apparatus. In the echo cancellation apparatus in any one of Claims 1 thru | or 4,
The error calculation unit of the BG unit includes an FG error frequency domain conversion unit that converts the time domain FG error signal into a frequency domain signal by adding 0 which is half the number of data.
If the FG error frequency domain conversion unit performs the conversion process by adding 0 to the old data side,
When the frequency domain adaptive filter coefficient is converted to the time domain, frequency domain processing is performed in which the coefficient one backward from the half becomes 0,
An echo canceling apparatus, wherein the first coefficient obtained by converting the frequency domain adaptive filter coefficient to the time domain is set to 0 when the FG error frequency domain error calculation unit sets the new data side to 0. A process in which an FG estimation signal generation unit filters an incoming signal with an FG filter coefficient to obtain an FG estimation signal;
An FG unit subtracting the FG estimation signal from the collected sound signal to obtain an FG error signal;
A process in which a BG estimation signal generation unit filters the received signal with an adaptive filter coefficient to generate a BG estimation signal;
An error calculating unit subtracting the BG estimation signal from the FG error signal to obtain a BG error signal;
A process in which a BG filter coefficient update unit uses the BG error signal and the received signal to update the adaptive filter coefficient so that the power of the BG error signal decreases;
An echo canceling method comprising: a step of adding an adaptive filter coefficient to the FG filter coefficient and updating the FG filter coefficient;
The BG section performs processing in the frequency domain. When the adaptive filter coefficient in the frequency domain is converted into the time domain, the first coefficient or the coefficient one half behind is zero, and the others remain as they are. The adaptive filter coefficient updated by frequency domain processing on the data in at least one of the processes of updating the adaptive filter coefficient so as to be the same as the result of cyclic convolution with the filter coefficient of a certain band rejection filter An echo canceling method comprising the step of storing in an adaptive filter coefficient register as: The echo cancellation method according to claim 6,
Selecting a frequency domain adaptive filter coefficient converged from the FG error signal and the BG error signal or the BG estimation signal and outputting a constraint condition command;
Echo cancellation characterized in that the data in at least one of the steps of updating the frequency domain adaptive filter coefficient is a frequency domain adaptive filter coefficient that is a non-selected frequency domain adaptive filter coefficient that has not received the constraint processing command Method. The echo cancellation method according to claim 6,
The step of obtaining the BG error signal comprises a step of converting the time domain FG error signal into a frequency domain signal by adding 0 of half the number of data.
If the process of converting the FG error signal to a signal in the frequency domain is performed by adding 0 to the old data side,
When the frequency domain adaptive filter coefficient is converted to the time domain, a frequency domain processing process is performed in which the coefficient one backward from half is 0,
When the process of converting the FG error signal into a frequency domain signal includes a new data side of 0, the process includes a frequency domain processing process of setting the first coefficient obtained by converting the frequency domain adaptive filter coefficient to the time domain as 0. Echo canceling method characterized.   6. An apparatus program for causing a computer to function as each apparatus according to claim 1.   A computer-readable recording medium on which any one of the programs according to claim 9 is recorded.

Images