JPS6161579B2 - - Google Patents

Description

[Detailed description of the invention]

(Technical Field of the Invention) The present invention relates to an adaptive echo canceller that cancels echo while estimating the transmission characteristics of an echo path using sequentially received signals and echo signals. (Prior art and its problems) Echoes in long-delay telephone lines such as satellite lines are a cause of significant deterioration in call quality. Although echo suppressors currently in use can effectively suppress echoes, they have the disadvantage that they cannot in principle avoid deterioration in speech quality, such as speech disconnection and echo leakage during superimposed speech. For this reason, echo cancellers are attracting attention as a new echo control device. The principle of an echo canceller is to estimate the transmission characteristics of the echo path from the received signal and echo signal, generate a pseudo echo signal based on the results, and cancel the echo by subtracting it from the true echo signal. . The main algorithm for estimating the transmission characteristics of the echo path of conventional echo cancellers is based on the learning identification method. Therefore, the convergence time was longer and the amount of echo cancellation was insufficient compared to when white noise was input. In order to eliminate the above drawbacks, the present inventors used received signals for each predetermined time length to find the optimal autoregressive coefficient in terms of the squared error when the received signal is the output of an autoregressive model. Find the difference signal 〓 _j between the predicted value of the received signal using this autoregressive coefficient and the received signal and the difference signal 〓 _j between the predicted value of the echo signal using the autoregressive coefficient and the echo signal,
Using the difference signal 〓 _j of the received signal and the difference signal 〓 _j between the echo signals, perform learning such that the modified signal 〓 _j+1 = 〓 _j + α 〓 _j 〓 _j ／‖〓 _ｊ ‖ ² Fixed method algorithm (J.
Nagumo and A.Noda: “A learning method”
for system identification”, IEEE Trans., AC
-12, 3, P.282 (June 1976). ), the echo path can be canceled by successively estimating the transmission characteristics of the echo path, creating a pseudo echo signal using the successively estimated transmission characteristics, and subtracting it from the true echo signal. By appropriately selecting the order of the autoregressive model,
We proposed an echo control method that can whiten the difference signal of the received signal, has a short convergence time, and has an extremely large amount of echo cancellation compared to conventional methods (Japanese Patent Application No. 53-57129). (See "Echo Control Method" and Japanese Patent Application No. 165196/1982 "Echo Control Method"). These methods are excellent as echo path identification and echo cancellation methods, but
It is necessary to separately provide a register for storing the received signal and a register for storing the residual signal of the received signal, and to perform controls such as delaying the received signal until the regression coefficient is determined. (Object of the Invention) The present invention has a simple configuration and has characteristics comparable to those of the prior application method by inserting an inverse matrix of the autocorrelation matrix of the received signal into the error matrix of the echo canceller. This provides an echo control method. (Structure and operation of the invention) Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 shows a block diagram of an embodiment of an echo canceller using the present invention. 1 is an echo canceller, 2 is an input terminal on the receiving side, 3 is an output terminal on the transmitting side,
4 is a receiving side output terminal, 5 is a transmitting side input terminal, 6 is a hybrid coil, 7 is a terminal device such as a telephone,
8 is a regression coefficient calculator, 9 is a register, 10 is a convolution operator, 11 is a register, 12 is a corrector,
13 is a matrix operator, and 14 is a subtracter. In order to simplify the explanation, it is assumed that the signals are digitized in the echo canceller 1, and that clock pulses are naturally supplied to each part, although they are omitted in FIG. To explain the operation according to the diagram, the received signal x _j input from the receiving side input terminal 2 is sent to the terminal device 7 through the receiving side output terminal 4 and the hybrid coil 6, but a part of the received signal is It passes through the hybrid coil 6 and enters the transmitter input terminal 5 as an echo y _i . On the other hand, inside the echo canceller 1, the received signal x _j is sent to the regression coefficient calculator 8 and also to the register 9. The regression coefficient calculator 8 calculates the received signal x=x ₁ within a predetermined time,
x ₂ , ..., x _L ), regression coefficient a=a ₁ , a ₂ ,
..., find a _M , and use it to find the inverse matrix. For example, the algorithm to find the autoregressive coefficient is
Durbin's method (Reference, Durbin, J. (1960) The
Fitting of time−series models, Rev.Inst.Stat.
, 28, 233-244). That is, Then, the initial condition E ⁽⁰⁾ = R ₀ (1) a ⁽¹⁾ = R ₁ /E ⁽⁰⁾ (2) E ⁽¹⁾ = E ⁽⁰⁾ (1-[a ₁ ⁽¹⁾ ] ² ) (3) a _j ⁽ⁱ⁾ = a _j ^(i-1) −a _i ⁽ⁱ⁾ a _ij ^(i-1) j=1, 2, ..., i-1 (5) E ⁽ⁱ⁾ = E ^{(i- 1)} (1-[a _i ⁽ⁱ⁾ ] ² ) i=2, 3, ..., M (6) It is found recursively. Note that the regression coefficient a ₁ ,
a ₂ , ..., a _M are a ₁ ^(M) , a ₂ ^(M) ... in equations (4) and (5), respectively.
..., corresponds to a _M ^(M) . Next, find the autoregressive coefficient a ₁ , ..., a _M ,
Find the inverse matrix Q of the autocorrelation matrix. this is however, It is required as. B' is the inverse matrix of B. Since the regression coefficient calculator 8 has a lower processing speed than other circuits, it can be configured mainly using a microprocessor. The inverse matrix Q obtained by the regression coefficient calculator 8 is sent to the matrix calculator 13. The configuration of the matrix calculator 13 is as shown in FIG. 2, for example, and FIG. 2 shows the case where M=1. 11, 12, ..., 44 are multipliers, q ₁₁ for 11, q ₁₂ for 12, etc.
ij contains the contents of q _ij . However, q _ij is the element at the i-th row and j-column of the inverse matrix Q. That is, the signal x _j of register 9 = (x _j-1 , x _j-2 ,
..., x _jN ) is transferred to the matrix operator 13,
Therefore, the correction vector Qx _j is obtained, and the corrector 1
Transferred to 2. On the other hand, the signal x _j in register 9
is transferred to the convolution calculator 10, and the pseudo echo signal 〓 _j = 〓 _{j , 〓 j is} obtained using the signal x _j and the contents 〓 _j of the register 11 . Pseudo echo signal〓 A difference signal e _j between _j and echo y _j is obtained by the subtracter 14, and using the difference signal e _j and the output Qx _j from the matrix operator 13, the corrector 12 calculates the correction vector Qx The following equation holds true when _j is regarded as the difference signal between the received signal and its predicted value of the learning identification method shown in the previous application 〓 _{j +1} ＝〓 _j +αQe _j x _j ／〓x _j 〓 ^2. Modify the value 〓 _j = (〓 ₁ , 〓 ₂ , . . . , 〓 _N ) in the register 11 according to the algorithm in (8). In equation (8), 〓 _j indicates the value in the register 11 before being modified, and 〓 _j+1 indicates the value in the register 11 after being modified. Here, α is an arbitrary value of 0<α<2, but usually α=1. In addition to the learning identification method, it is of course possible to apply a learning identification method such as the steepest descent method. Next, the regression coefficient calculator 8 will be explained in detail. FIG. 3a is a configuration example diagram of the regression coefficient calculator 8.
In FIG. 3a, 201 is a serial-parallel converter, 20
2, 203 is a register, 204 is a flip-flop, 205 is an adder, 206 is a multiplier, 207 is a gate, 208 is an accumulator, 209 is a constant multiplier, 2
10 is a register, 211 is a flip-flop, 2
12, 213, and 214 are counters, 215 is a gate, 216 is a microprocessor, and 217 is a memory. To explain the operation according to the diagram, the received signal x input from the receiving side input terminal 2 in FIG.
_j is transferred to the serial-parallel converter 201, and when L signals are accumulated in the serial-parallel converter 201, they are transferred to the registers 202, 203 and the flip-flop 204. register 2
02, 203 and the flip-flop 204 circulate synchronously, so that the signals of the multiplier 206
x _j x _j is determined and transferred to the accumulator 208. In the accumulator 208, the signal x from the multiplier 206
Cumulative value of _j x _j

Find [formula]. Counter 212 outputs a signal every L clock pulses and sends the contents of accumulator 208 to constant multiplier 209. In the constant multiplier 209, the value from the accumulator 208

[Formula] is multiplied by a constant (1/L), and register 21
Send to 0. The signal from counter 212 is passed through the interrupt line of microprocessor 216 to instruct microprocessor 216 to read the signal in register 210. On the other hand, microprocessor 216 writes the signal in register 210 to a specific address in memory 217 using normal computer operation. On the other hand, the counter 213 receives the clock pulse (L+
1) Flip-flop 2 to output signals individually
11 is for every L clock pulses and between consecutive clock pulses 1, 2, 3... as shown in FIG.
The state "1" is maintained, during which the gate 207 is closed. In Figure 3b, 300 is a clock pulse, 301
is the output of the counter 212, 302 is the output of the counter 21
The output of 3, 303, indicates the state of flip-flop 211. Therefore, the signal in the register 203 is changed to the register 203 every L clock pulses.
The result is as shown in FIG. 3c. 400 in FIG. 3c is the register 202
The signals 401, 402, 403 are each L,
Register 203 after 2L and 3L clock pulses
and the signals in flip-flop 204.
As a result, the register 210 sequentially receives the signals as before.

【formula】

[Formula]... is transferred. When the counter 214 counts M signals from the counter 212, it closes the gate 215, and as a result, the register 21
M signals within 0 are loaded into memory 217 by microprocessor 216 . Microprocessor 216 registers 210
A ₁ ^{( M)} , a ₂ ^(M) ,…
_. ^_ (Effects of the Invention) When the present invention is used, the convergence speed of the echo canceller is faster and the amount of cancellation is larger than in the conventional method. Also, unlike the earlier applications, Japanese Patent Application No. 53-57129 (echo control method) and Japanese Patent Application No. 53-165796 (echo control method), control is easier and the configuration is simpler and can be realized using an integrated circuit. This has the advantage that it is easy to do. As described above, the present invention dramatically improves the performance of the echo canceller, such as the amount of cancellation and convergence speed, with only a slight increase in the amount of calculation and hardware. This also has the effect of making it possible to remove attachments such as the center clipper that had been attached.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a configuration example of an echo canceller according to the present invention, FIG. 2 is a configuration example diagram of a matrix calculator used in the present invention, and FIG. 3a is a detailed configuration example diagram of a regression coefficient calculator used in the present invention. Figure 3b,c is Figure 3a
FIG. 3 is a diagram for explaining the operation of the regression coefficient calculator of FIG.

Claims (1)

[Claims] 1. An adaptive echo canceller that generates a pseudo echo signal while estimating the transmission characteristics of an echo path using successive received signals and echo signals, and cancels the echo signal by subtracting the pseudo echo signal from the echo signal, The received signal for each predetermined time length is regarded as an output signal of an autoregressive model having a certain autoregressive coefficient, and the autocorrelation matrix of the received signal is calculated using the autoregressive coefficient obtained by the received signal within the time length. a regression coefficient calculator that calculates an inverse matrix of , a matrix calculator that creates a correction vector using the received signal and the inverse matrix, a storage device that stores estimated transmission characteristics of the echo path, and a storage device that stores the estimated transmission characteristics of the echo path; a corrector that corrects according to a learning identification method using an echo canceller output, the correction vector, and the contents of the storage device; and a device that creates the pseudo echo signal using the contents of the storage device and the received signal. An echo control method characterized by: