JPS5831129B2 - Echo control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an adaptive echo canceller that cancels echo while estimating transmission characteristics of an echo path using sequentially received signals and echo signals.

Echo in long-delay telephone lines such as satellite lines causes a 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 the echo canceller is to estimate the transmission characteristics of the echo path from the received signal and the echo signal, generate a pseudo echo signal based on the results, and cancel the echo by subtracting it from the true echo signal.

The algorithm for estimating the echo path (7) transmission % of conventional echo cancellers is mainly based on the learning identification method. Compared to the case where white noise is input due to correlation, the convergence time is longer and the amount of echo cancellation is insufficient.

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. A difference signal between the predicted value of the received signal using the autoregressive coefficient and the received signal, and a difference signal between the predicted value of the echo signal using the autoregressive coefficient and the echo signal are calculated, and a difference signal between the received signals is calculated. The transmission characteristics of the echo path are estimated sequentially using the difference signal between the By appropriately selecting the order of the autoregressive model, it is possible to whiten the difference signal of the received signal, and the convergence time is short and the amount of echo cancellation is extremely large compared to conventional methods. "Patent Application No. 53-57129 (Echo Control Method)" which proposed an echo control method that has the effect of making dogs bark
reference).

This method maintains real-time processing by inserting a register in the main path.

Although this method is certainly excellent as an echo path identification method, it has the drawback of increasing delay time because a register is inserted into the main path.

Another disadvantage is that when the echo canceller fails, the main path is disconnected, making it impossible to make a call.

The present invention provides an echo control method having characteristics comparable to those of the prior application method without inserting a register in the main path.

Hereinafter, the present invention will be described in detail with reference to the drawings, but for the sake of simplicity, the difference signal between the true signal and the predicted value will be referred to as a residual signal.

FIG. 1 shows a configuration diagram for explaining the principle of an echo canceller according to the present invention.

1 is an echo canceller, 2 is a receiving side input terminal, 3 is a transmitting side output terminal, 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 predictor, 1
1 is a subtracter, 12 is a register, 13 is a convolution operator, 14 is a predictor, 15゜16 is a subtracter, 17 is a corrector,
18 is a register, 19 is an adder, 20 is a subtracter, and 21 is a register.

In order to simplify the explanation, it is assumed that signals are digitized in the echo canceller 1, and that clocks are naturally supplied to each part, although this is omitted in FIG. 1.

To explain the operation according to the diagram, a received signal 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 sent to the hybrid coil. 6 and enters the transmitting side input terminal 5 as an echo.

On the other hand, inside the echo canceller 1, the received signal Xj is sent to the regression coefficient calculator 8, and is also sent to the predictor 10 after being delayed by a register 9 for a certain period of time.

The regression/coefficient calculator 8 calculates regression coefficients a-(al, a2..., Find aM).

The algorithm is, for example, the Durbin method (Reference, Durbin, J. (1960) The Fit
tingof time-series models
Rev, In5t.

5tat, 28, 233-244).

In other words, it is found recursively as .

Since the regression coefficient calculator 8 may have a slower processing speed than other parts, it can be configured mainly using a microprocessor.

In addition, the regression coefficients a1ya2y ”' z aMha are each (
4), (5) Formula (D al (1M).

It corresponds to a(M), . . . , a(M).

Among the regression coefficients calculated by the regression coefficient calculator 8 = (al s
a 2 + ”・s a M) is the predictor 10 and 1
Transferred to 4.

The predictor 10 uses the regression coefficients and the received signal sent through the register 9 to calculate the predicted value Xj of the received signal at time j as XJ2, Σ a 1
Create XJ-1.

l21 The configuration of the predictor 10 is as shown in FIG.
Here, the case where M=5 is shown.

101.102, 103, 104, 105 are delay elements,
111, 112, 113, 114, 115 are multipliers, 1
20 is an adder.

The output Xj of the predictor 10 is transferred to the subtractor 11.

The subtracter 11 outputs the received signal Xj and the predictor 10.
From the input X, a residual signal Xj =Xj-Xj is created and transferred to register 2.

Signal Xj in register 2: (XJ 1.?'-2,”
', X'-N) is a signal 1h2 (hl, h2...
・, hN) and the convolution operation i □h iY J −I are performed in the convolution operator 13, and the operation result ~2, Σ h・X・ ・ is 1=l I J−1 Subtractor 16 will be forwarded to.

On the other hand, the echo signal y・ input from the transmitting side input terminal 5
is sent to register 21 and subtracter 20.

The predictor 14 to which the output of the register 21 is applied receives the regression coefficient wheel transferred from the regression coefficient calculator 8 and the echo signal y.
Using
The subtracter 15 generates the echo signal yj and the predictor
Residual signals yj-y, -yJ are created from the output yj of No. 14,
It is transferred to the subtracter 16.

In the subtracter 16, an error signal ej=yj=y7j is created from the signal yj from the subtracter 15 and the signal .about. from the convolution Ham unit 13, and is transferred to the corrector 17.

The corrector 17 uses the error signal e J L register 2 signal Xj = (Xj-1jXj-2j"'5Xj-N) to
), 1hJ+1=1hj−a eJxj
(7) 11 N・;12 Algorithm of Toiu learning identification method (J, Nagu-m
o and A, Noda: Learning me
thod for system 1 dentifi
cation", IEEE Trans-ns, AC-1
2, 3, P, 282 (June 1976), see)
Accordingly, the value in register 18 is 1h.=(ht.

h2. ......, hN) is corrected.

(In Equation 7, 1h indicates the value in the register 18 before being modified, and Ihj+1 indicates the value in the register 18 after being modified.

Here, α can be any value of O × α × 2, but usually α = 1
It is said that

On the other hand, the output of the predictor 14 transferred to the adder 19
Baka y is added to the output yj of the convolution operator 13, and the result y':=yj+yj is the output of the subtracter 20.
will be forwarded to.

The subtracter 20 subtracts the output of the adder 19 from the echo signal y, and the error e and yj-yj is sent out through the output terminal 3 on the transmitting side.

In this case, if the signals in the register 8 have the same transmission characteristics of the echo paths, the error e becomes 0 and the echo to the speaker disappears.

In FIG. 1, if all the regression coefficients d, which are the outputs of the regression coefficient calculator 8, are forced to O, the outputs of the predictors 10 and 14 will be O, and the outputs of the subtractors 11 and 15 will be the received signal The echo signal becomes y, and has the same configuration as the conventional echo canceller.

Therefore, since the configurations of predictors 10 and 14 have already been described, regression coefficient calculator 8 will now be described in detail.

FIG. 3a is a configuration example diagram of the regression coefficient calculator 8.

In Figure 3, 201 is a serial-parallel converter, 202゜20
3 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, 210 is a register, 211 is a flip-flop, 212゜213.214 is a counter ,
216 is a gate, 217 is a memory, and 218 is a microprocessor.

To explain the operation according to the diagram, the received signal X input from the receiving side input terminal 2 in FIG.
When L signals are accumulated, they are transferred to registers 202 and 203 and flip-flop 204.

Registers 202, 203 and flip-flops 204
By circulating the signals synchronously, X J X J is determined by the multiplier 206 and transferred to the accumulator 208.

The accumulator 208 receives the signals X and XJJ from the multiplier 206. Find the cumulative value of ΣX・X・.

The counter 212''=1 j J outputs a signal every L clocks and sends the contents of the accumulator 208 to the constant multiplier 209.

In the constant multiplier 209, the value XJXJ from the accumulator 208
is multiplied by a constant (1/L) and sent to the register 210.

The signal from counter 212 is passed to microprocessor 218.
The microprocessor 218 is instructed to read the signal in the register 210 through the interrupt line of the microprocessor 218 .

In contrast, microprocessor 218 writes the signal in register 210 to a specific address in memory 217 using normal computer operation.

On the other hand, since the counter 213 outputs a signal every (L-1-1) clocks, the flip-flop 211 outputs a signal every L clocks as shown in FIG. 3b.
The state "1" is maintained for . . . , and the gate 207 is closed during that time.

In FIG. 3B, 300 is a clock, 301 is the output of the counter 212, 302 is the output of the counter 213, and 303 is the state of the flip-flop 211.

Therefore, the signal in the register 203 is shifted one by one with respect to the signal in the register 202 every L clocks, and becomes as shown in FIG.

In FIG. 3c, 400 is the signal in register 202, 401, 4
02 and 403 are the signals in the register 203 and flip-flop 204 after the 2L and 3L clocks, respectively.

As a result, in the same way as before, the register 210 sequentially stores the signals -0ΣXjXj+ls,X:,xJxj+
2. - is transferred as LJ=IJ.

When the counter 214 counts M signals from the counter 212, the gate 216 is closed, and as a result, the M signals in the register 210 are taken into the memory 217 by the microprocessor 218.

The microprocessor 218 uses the data in the memory 211 taken in from the register 210 to formulate equation (i)
, (2) , (3) , (4) , (5) ,
a(M), a(M), . . . according to software created to meet (6).

Determine aM(M) and forward it to predictors 10 and 14.

Next, FIG. 4 shows an operation time chart of the entire system.

In Figure 4, a is the received signal sequence X-b is the processing time required to calculate the regression coefficient wheel, C is the time required to generate the residual component X, d is the pseudo echo residual component sequence, and f is the echo signal train, g
is the predicted signal sequence of the echo signal, h is the residual component error IJ
, i indicates a pseudo echo train.

In correspondence with FIG. 1, a is the input signal to the regression coefficient calculator 8 and the register 9, b is the processing time in the regression coefficient calculator 8, d is the content in the register 12 at each time, and e is the input signal to the regression coefficient calculator 8 and the register 9. The output signal of the convolution operator 13, f is the input signal to the register 21, g is the output signal of the predictor 14, and h is the subtracter 1
6 is the output signal of adder 19, and i is the output signal of adder 19.

In FIG. 4, the signals from Xl to XIO and from X6 to
It is assumed that the autoregressive coefficient of the signal from Xl to X15 is constant, and the autoregressive coefficient of the signal from Xl to X10' is calculated using the signal from xl to The coefficients are obtained using signals from X6 to X10''.

The introduction of registers 9 and 21 enables such real-time processing, thereby preventing signal delays caused by inserting registers into the main path.

Next, an embodiment of the present invention shown in FIG. 5 will be described.

In FIG. 5, the same numbers as in FIG. 1 indicate the same devices.

Therefore, only those with different numbers will be described here: 22 is a register, 23 is a convolution operator,
24 is a subtractor.

In the apparatus shown in FIG. 1, the pseudo echo signal is created using the result of convolution of the prediction error signal and the contents of the register 18, but in the embodiment shown in FIG. The generation of a signal indicating the estimated transmission characteristics of the echo path, which is not necessarily necessary, is realized by a circuit separate from the generation of the pseudo echo signal.

Therefore, the creation of a signal indicating the estimated transmission characteristics of the echo path becomes a complete system in itself, and the interval in which the transmission characteristic is considered constant and the interval in which the autoregression coefficient is calculated can be freely set.

Note that the registers 9 and 21 used for timing alignment in the embodiment of FIG. 1 are included in the predictors 10 and 14, respectively, in the embodiment of FIG.

When the present invention is used, the convergence speed of the echo canceller is faster and the amount of cancellation is also greater than in the conventional method.

Furthermore, unlike the earlier application, Japanese Patent Application No. 53-57129 (echo control method), there is no signal delay caused by inserting a register into the main path.

In addition, signal propagation on the main path is guaranteed even in the event of a failure of the echo canceller.

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, which is different from that used in conventional echo cancellers. It is also possible to remove accessories such as a center clipper.

[Brief explanation of drawings]

FIG. 1 is a block diagram for explaining the principle of the echo canceller according to the present invention, FIG. 2 is a detailed block diagram of the predictor used in the present invention, and FIG. 3a is a detailed block diagram of the regression coefficient calculator used in the present invention. 3b and 3c are diagrams for explaining the operation of the predictor in FIG. 3a, FIG. 4 is a time chart for explaining the operation of the device in FIG. 1, and FIG. 5 is a block diagram of an embodiment of the present invention. be.

Claims (1)

[Claims] a regression coefficient calculator for calculating an autoregressive coefficient when the received signal is regarded as an output signal of an autoregressive model for a received signal for each predetermined time length; and the autoregressive coefficient and the received signal. a first predictor that uses the autoregressive coefficient to create a predicted amount of the received signal; a second predictor that uses the autoregressive coefficient and the echo signal to create a predicted value of the echo signal;
a first subtractor for extracting a first residual signal between the received signal and a predicted value of the received signal; and a second subtracter for creating a second residual signal between the echo signal and the predicted value of the echo signal. a first storage means for storing estimated transmission characteristics of an echo path through which the echo signal passes; and a second storage means for storing a predetermined number of the first residual signals;
a first convolution calculator that calculates an estimated value of the second residual signal by convolving the contents stored in the first storage means and the contents stored in the second storage means; and a third subtracter that obtains the difference between the estimated value and the second residual signal as an estimated error signal of the estimated transmission characteristic, using the error signal and the contents of the second storage means. a corrector that sequentially corrects the estimated transmission characteristics by correcting the first storage content according to an identification method by learning; a third storage means for storing a predetermined number of the received signals; and the third storage. a second convolution calculator for creating a pseudo echo signal by performing a convolution operation between the received signal stored in the means and the contents of the second storage means; and means for canceling the echo signal by pulling the echo signal.