JPH07303062A - Echo canceler and echo path estimating method - Google Patents

Abstract

PURPOSE:To estimate an echo path with high accuracy at high speed at the echo canceler provided at a telephone line network for performing transmission between sounds to be sent through four wire lines and sounds to be sent through two wire lines. CONSTITUTION:An echo path estimating/pseudo echogenerating circuit 3 is composed of an adaptive filter such as an FIR filter and when a near terminal sound signal Sin and a distant terminal sound signal Rout are inputted, a pseudo echo signal Y is outputted to cancel an echo component contained in the former. A delay time measuring circuit 71 measures the delay time of echo based on the signals Sin and Rout and based on this result, the delay time of a delay circuit 75 is set. On the other hand, a spectrum comparator circuit 73 designates any frequency response characteristic in a data base 74 based on signals Rout' and Sin and based on the designated characteristic, the tap coefficient of the FIR filter is set. Thus, the circuit 3 performs high-accurate echo estimation from the beginning.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo canceller suitable for use in a mobile communication network or a long distance telephone line network and an echo path estimation method suitable for use in this echo canceller.

[0002]

2. Description of the Related Art In a long-distance telephone line passing through a submarine cable or a communication satellite, generally, a subscriber line connected to both ends is a two-line type, and an intermediate long-distance transmission part is used for signal amplification. It is a line type. Similarly, mobile phones (or cellular phones).
In the mobile communication network using e)), the subscriber line of the fixed-side analog telephone is a two-line type, and the part from the terminal of the mobile telephone to the exchange or the like is a four-line type. In this case, 2 lines and 4
A hybrid circuit for performing 4-line / 2-line conversion is provided in the connection portion with the line.

This hybrid circuit is designed so as to match the impedance of the two-wire line, but it is difficult to always obtain a good matching, so that the received signal that arrives at the four-wire input side of the hybrid circuit is Leakage to the 4-wire output side causes so-called echo. Such an echo has a lower level than the voice of the talker and reaches the talker with a delay of a certain time, resulting in a call failure. The call disturbance due to such an echo becomes more remarkable as the signal propagation time becomes longer. In particular, in the case of mobile communication using a mobile telephone, various kinds of processing are performed in a wireless communication section up to an exchange or the like, so that a signal delay amount is large and a call failure due to echo is a particular problem. An example of the echo waveform with respect to the impulse response is shown in FIG.

There are echo suppressors and echo cancellers as devices for blocking the above echo. FIG. 2 shows a schematic configuration of an echo canceller used in a mobile communication network.
The echo canceller 1 shown here is provided before the hybrid circuit 2. In this figure, a subscriber of an ordinary analog telephone is called a near-end speaker, and a subscriber of a mobile telephone or the like is called a far-end speaker. Further, the far-end voice signal input to the echo canceller 1 is Rin, the far-end voice signal output from the echo canceller 1 is Rout, and the near-end voice signal input to the echo canceller 1 is Sin, from the echo canceller 1. The output near-end audio signal is indicated by Sout.

The echo canceller 1 shown in FIG. 2 comprises an echo path estimation / pseudo echo generation circuit 3, a control device 4, an adder 5 and a non-linear processing circuit 6. Here, the echo path estimation / pseudo echo generation circuit 3 uses the far-end voice input R
hybrid circuit 2 based on in and near-end voice input Sin
Of the response characteristic of (1), and thus the echo path (that is, the line through which the echo propagates) is estimated. Next, an expected echo (that is, a pseudo echo) from the hybrid circuit 2 is generated by the convolution operation of the estimation result and the far-end voice input Rin. In the adder 5, this pseudo echo is subtracted from the near-end voice input Sin, thereby canceling the echo.

Specifically, the echo path estimation / pseudo echo generation circuit 3 has an FIR filter, and the pseudo echo signal Y (z) output from the FIR filter is obtained by the following equation 1.

[Equation 1]

Where N is the number of taps of the FIR filter and b _i (where i = 0, 1, 2, ... N-1) is the tap coefficient at each tap. When the appropriate values of the number of taps N and the tap coefficient b _i are obtained by the echo path estimation, the pseudo echo signal Y (z) becomes close to an actual echo, and the echo is canceled by the adder 5. As the above-mentioned echo path estimation method, an adaptive filter technology, such as a learning identification method, which has a relatively small amount of calculation among the adaptive algorithms and has good convergence characteristics is used. For details of the learning identification method, see, for example, IEICE Transactions ('77 / 11 Vo
l. J60-A No. 11 “About echo cancellation characteristics of echo canceller using learning identification method”).

The following conditions are required for the above learning to be possible. There is a far-end voice output Rout at a level sufficient for the echo to be returned as the near-end voice input Sin, in other words, the far-end speaker is in a transmitting state. The near-end voice input Sin is composed of only echo (or echo and white noise), in other words, the near-end speaker is not in the transmitting state.

On the other hand, when the far-end talker is in a non-sending state, and when the far-end talker and the near-end talker are simultaneously talking (this state is called double talk), false learning of echo path estimation is performed. Therefore, it is necessary to turn off the learning function.

By the way, a digital signal is transmitted through the transmission line, and D / A conversion (generally, μ is performed between the echo canceller 1 for processing the digital signal and the hybrid circuit 2 for conversion to the analog line. -LAW conversion) is performed. Therefore, a non-linear characteristic relationship is established between the far-end voice output Rout and the near-end voice input Sin, and the echo path estimation / pseudo echo generation circuit 3
Complete echo cancellation cannot be performed only by linear calculation using, for example. Therefore, an echo component that cannot be canceled is generated.

A non-linear processing circuit 6 is provided to eliminate such an echo component (referred to as "residual echo"). The non-linear processing circuit 6 performs a non-linear switching operation. That is, when the near-end voice output Sout is composed of only echoes, that is, when only the far-end speaker is in the transmitting state (this case is referred to as "far-end talker single talk"), the near end. A switching operation is performed to prevent transmission of the audio output Sout, or an operation is performed to replace the near-end audio output Sout with pseudo noise.

The control device 4 controls the echo path estimation / pseudo echo generation circuit 3 and the non-linear processing circuit 6. That is, the far-end person's non-talking state is detected, or double talk is detected, and the learning function for echo path estimation is turned on / off by the double talk signal DT.
The far-end talker's single talk is detected, and the switching operation of the nonlinear processing circuit 6 is controlled.

[0013]

By the way, in the above-mentioned technique, since the adaptive filter technique such as the learning identification method is simply used, if the delay time of the echo to be erased is increased, the number of taps of the adaptive filter is increased. There is a problem that the time increases until the accuracy of estimation becomes sufficient and the calculation amount increases.

Further, in the above-mentioned technique, the echo canceller has no information about the echo path at the beginning of operation. However, the observations made by the present inventors have revealed that the characteristics of the echo path are governed by the characteristics of the hybrid. That is, the echo waveform determined by the hybrid is shifted back and forth on the time axis according to the delay amount in the transmission path,
When the attenuation is performed according to the amount of attenuation in the transmission line, the echo waveform of the transmission line with respect to the impulse input can be obtained with considerable accuracy. The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an echo canceller and an echo path estimation method capable of estimating an echo path at high speed and with high accuracy.

[0015]

In order to solve the above-mentioned problems, the structure according to claim 1 provides a telephone line network for transmitting between voices sent through four lines and voices sent through two lines. In the echo canceller provided,
Storage means for storing frequency response characteristics of a plurality of hybrids, and a transmission signal propagated from the 4 line side to the 2 line side.
Based on the correlation with the echo signals propagated from the line side to the four line side, the echo path is estimated by selecting means for selecting any one of the frequency response characteristics and a predetermined learning algorithm, and the transmission signal is appropriately selected. Pseudo echo generating means for generating a pseudo echo signal by delaying and deforming by a predetermined parameter, and initial parameter setting means for setting the initial value of the parameter based on the frequency response characteristic selected by the selecting means. It is characterized by having.

Further, in the structure according to claim 2, 4
In an echo path estimation method for estimating an echo path for a telephone network that transmits between a voice transmitted through a line and a voice transmitted through a two line, the frequency response characteristics of a plurality of hybrids are stored, One of the frequency response characteristics is selected based on the correlation between the transmission signal propagated to the two-line side and the echo signal propagated from the two-line side to the four-line side, and the echo path is estimated by a predetermined learning algorithm. Characterized in that the transmission signal is appropriately delayed and a pseudo echo signal is generated by deforming the transmission signal with a predetermined parameter, and the initial value of the parameter is set based on the frequency response characteristic selected by the selecting means. There is.

Further, in the structure according to claim 3, in the echo canceller according to claim 1, the transmission signal propagated from the 4 line side to the 2 line side and the transmission signal propagated from the 2 line side to the 4 line side. Delay time measuring means for measuring the delay time of the echo signal with respect to the transmission signal based on the echo signal, and delay for setting the delay time in the pseudo echo generating means to the delay time measured by the delay time measuring means. It is characterized in that a quantity setting means is provided.

[0018]

In the structure according to the first aspect, the storage means stores the frequency response characteristics of a plurality of hybrids, and the selection means stores the transmission signal propagated from the 4 line side to the 2 line side and the 4 line from the 2 line side. One of the frequency response characteristics is selected based on the correlation with the echo signal propagated to the line side. Also,
The pseudo echo generating means estimates the echo path by a predetermined learning algorithm, delays the transmission signal as appropriate, and deforms the transmission signal by a predetermined parameter to generate a pseudo echo signal. Since the initial parameter setting means sets the initial value of the parameter based on the frequency response characteristic selected by the selecting means, the parameter has high accuracy from the beginning.

According to the second aspect of the present invention, the frequency response characteristics of the plurality of hybrids are stored, and the transmission signal propagated from the 4th line side to the 2nd line side and the 4th line from the 2nd line side are stored.
One of the frequency response characteristics is selected based on the correlation with the echo signal propagated to the line side. Further, the echo path is estimated by a predetermined learning algorithm, the transmission signal is appropriately delayed, and a pseudo echo signal is generated by being deformed by a predetermined parameter. Here, since the initial value of the parameter is set based on the selected frequency response characteristic, the parameter has high accuracy from the beginning.

According to the third aspect of the invention, the delay time measuring means measures the delay time of the echo signal with respect to the transmission signal. The delay amount setting means sets the delay time in the pseudo echo generating means to the delay time measured by the delay time measuring means. As a result, the delay time in the pseudo echo generating means becomes more accurate from the beginning.

[0021]

EXAMPLES A. Configuration of Embodiment FIG. 1 is a block diagram of a main part of an echo canceller according to an embodiment of the present invention. In this device, an initial estimation circuit 7 is provided in addition to the echo canceller of FIG. The parts not shown have the same structure as in FIG. In FIG. 1, reference numeral 71 is a delay time measuring circuit, which calculates the delay time T _D (see FIG. 3) based on the near-end audio signal Sin and the far-end audio signal Rout, and outputs this result. As a method of obtaining the delay time T _D , for example, the difference between the time at which the peak of the near-end audio signal Sin occurs and the time at which the peak of the far-end audio signal Rout occurs within a predetermined period may be obtained. It may be obtained by cross-correlation of both signals.

Next, a delay circuit 75 delays the far-end voice signal Rout by the delay time T _D output from the delay time measuring circuit 71. Reference numeral 72 denotes a level ratio measuring circuit, which is a near-end voice signal Sin and a delayed far-end voice signal Ro
The levels of ut are compared, and a signal corresponding to the ratio of the two is supplied to one end of the multiplier 76. The delayed far-end audio signal Rout is supplied to the other end of the multiplier 76. Therefore, the multiplier 76 outputs the far-end audio signal Rout (hereinafter, signal Rout ′) whose delay time and level are normalized.
Is output).

Next, 74 is a frequency response database, which stores frequency response characteristics of various hybrids. Note that there are several tens of hybrids currently used for telephone lines, and it is easy to store the frequency response characteristics of all these hybrids. 7
Reference numeral 3 denotes a spectrum comparison circuit, which selects one of the various frequency response characteristics that most closely approximates the actual characteristics, based on the correlation between the signal Rout ′ and the near-end audio signal Sin. A conversion circuit 77 converts the selected frequency response characteristic into a response characteristic on the time axis. Specifically, the frequency response characteristic is converted into tap coefficients b _i (see Formula 1), and these tap coefficients b _i are set to the initial values of the tap coefficients of the FIR filter of the echo path estimation / pseudo echo generation circuit 3.

The far-end voice signal Rout delayed by the delay circuit 75 is also supplied to the echo path estimation / pseudo echo generation circuit 3. Therefore, in the echo path estimation / pseudo echo generation circuit 3, the near-end audio signal Si
n between the far end voice signal Rout and the delay time T _D (see FIG.
(See) does not exist. Note that the delay time measuring circuit 71, the level ratio measuring circuit 72, and the spectrum comparing circuit 73 stop their operations when double talk is detected in the control device 4 (when the double talk signal DT is output). To do.

B. Operation of Embodiment Next, the operation of this embodiment will be described. First, when the far-end speaker utters some voice, the content of the far-end voice signal Rou
It is supplied to the delay time measuring circuit 71 as t. Eventually,
When the near-end audio signal Sin, which is an echo based on this, is supplied to the delay time measuring circuit 71, the delay time measuring circuit 71
In, the delay time T _D is calculated, and the value is supplied to the delay circuit 75 and the echo path estimation / pseudo echo generation circuit 3.

As a result, the far-end voice signal Rout delayed by the delay time T _D is supplied to the level ratio measuring circuit 72. The level ratio measuring circuit 72 supplies a signal corresponding to the level ratio between the near-end audio signal Sin and the delayed far-end audio signal Rout to one end of the multiplier 76. As a result, the signal Rout ′ whose delay time and level are normalized is supplied to the spectrum comparison circuit 73.

In the spectrum comparison circuit 73, of the frequency response characteristics stored in the frequency response database 74, the one that most closely approximates the actual characteristic is based on the correlation between the signal Rout 'and the near-end voice signal Sin. To be selected. The selected frequency response characteristic is converted by the conversion circuit 77.
After being converted to the tap coefficient b _i via the multiplier 78, the level is corrected via the multiplier 78 and supplied to the echo path estimation / pseudo echo generation circuit 3 to set the initial value of the tap coefficient of the FIR filter. It

As described above, the delay time T _D for the delay circuit 75 is set and the tap coefficient b _i for the FIR filter is set.
The operation of the initial estimation circuit 7 ends when the setting of the initial value of is finished. After that, in the echo path estimation / pseudo-echo generation circuit 3, learning identification is performed based on the near-end audio signal Sin and the far-end audio signal Rout, and the tap coefficient b _i is appropriately changed in order to perform more accurate echo path estimation. To be done. As described above, according to this embodiment, the delay time of the delay circuit 75 can be set by the delay time measuring circuit 71, and the initial value of the tap coefficient b _i can be set by the spectrum comparing circuit 73, the frequency response database 74 and the converting circuit 77. You can set the value.

Of course, the estimation of the echo path is not completed by these initial values, and it is necessary to repeat learning in the echo path estimation / pseudo echo generation circuit 3. By giving it to the generation circuit 3, the time until the learning result converges can be extremely shortened. Further, according to the present embodiment, the far-end voice signal Rout delayed by the delay time T _D is supplied to the echo path estimation / pseudo echo generation circuit 3 via the delay circuit 75. Therefore, the echo path estimation / pseudo echo generation circuit The burden of learning in 3 is reduced, and it is possible to perform echo path estimation with higher speed and higher accuracy.

C. Modifications The present invention is not limited to the embodiments described above, and various modifications can be made, for example, as follows. The echo path estimation / pseudo echo generation circuit 3 is not limited to the one using the learning identification method, and various algorithms such as a Kalman filter may be used.

Echo path estimation / pseudo echo generation circuit 3
May be configured so that the level of the pseudo echo signal Y is initially made smaller than the estimated value, and is returned to the original level with the passage of time. This is because it is difficult to accurately estimate the pseudo echo between the first stages,
If the level of the pseudo echo signal Y is increased from the beginning,
This is because a problem such as noise is generated.

In the above embodiment, the echo path is estimated based on the voice uttered by the far-end speaker, but the echo path may be estimated using another signal. For example,
In the case of near-end calling, the ring back tone is used to estimate the echo path, and in the case of far-end calling, the training signal is sent to the hybrid immediately after the near-end side off-hook, and the echo path is estimated based on these. You may. As a result, when the near-end speaker and the far-end speaker actually start talking, the highly accurate echo path estimation will be completed.

In the above embodiment, the far-end voice signal Rout delayed by the delay circuit 75 is supplied to the echo path estimation / pseudo-echo generation circuit 3, but the non-delayed far-end voice signal Rout is echo path estimated. / It may be supplied to the pseudo echo generation circuit 3. In this case, the far-end voice signal Rout may be delayed inside the echo path estimation / pseudo-echo generation circuit 3 by the delay time T _D measured by the delay time measurement circuit 71 (specifically, in Expression 1. , And the corresponding tap coefficient b _i should be set to “0”). Accordingly, when the delay time T _D is varied also by the echo path estimation / echo replica generation circuit 3 inside the learning, it is possible to follow the delay time T _D after variation pseudo echo path.

[0034]

As described above, according to the echo canceller and the echo path estimation method of the present invention, the delay time when generating the pseudo echo is set in advance, or the initial value of the parameter is set to a highly accurate one. Therefore, it is possible to perform high-speed and highly accurate echo path estimation.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main part of an echo canceller according to an embodiment.

FIG. 2 is a block diagram of a conventional echo canceller.

FIG. 3 is a waveform diagram of an echo.

[Explanation of symbols]

3 Echo path estimation / pseudo echo generation circuit (pseudo echo generation means) 7 Initial estimation circuit (selection means) 71 Delay time measurement circuit (delay time measurement means, delay amount setting means) 74 Frequency response database (storage means) 75 Delay circuit ( Delay amount setting means)

Claims (3)

[Claims] 1. An echo canceller provided in a telephone network that performs transmission between voices transmitted through four lines and voices transmitted through two lines, and storage means for storing frequency response characteristics of a plurality of hybrids. Selecting means for selecting one of the frequency response characteristics based on a correlation between a transmission signal propagating from the line side to the two line side and an echo signal propagating from the two line side to the four line side; The echo path is estimated by a learning algorithm, the transmission signal is appropriately delayed, and a pseudo echo generating unit that deforms by a predetermined parameter to generate a pseudo echo signal, and an initial value of the parameter is selected by the selecting unit. And an initial parameter setting means for setting based on frequency response characteristics. 2. An echo path estimation method for estimating an echo path for a telephone network for transmitting between a voice transmitted through four lines and a voice transmitted through two lines, wherein frequency response characteristics of a plurality of hybrids are stored. Then, one of the frequency response characteristics is selected based on the correlation between the transmission signal propagated from the 4th line side to the 2nd line side and the echo signal propagated from the 2nd line side to the 4th line side, and predetermined learning is performed. An echo path is estimated by an algorithm, the transmission signal is appropriately delayed, and a pseudo echo signal is generated by modifying the transmission signal with a predetermined parameter, and the initial value of the parameter is based on the frequency response characteristic selected by the selecting means. An echo path estimation method characterized by setting. 3. The echo canceller according to claim 1, wherein the transmission signal is transmitted based on the transmission signal propagated from the 4 line side to the 2 line side and the echo signal propagated from the 2 line side to the 4 line side. Delay time measuring means for measuring the delay time of the echo signal, and delay amount setting means for setting the delay time in the pseudo echo generating means to the delay time measured by the delay time measuring means. Echo canceller.