JP2002290286A - Echo suppressor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an echo suppressor that can obtain a desired echo suppressing amount without causing overflow in an arithmetic operation of an adaptive filter even when the gain of a feedback path is 1 or over. SOLUTION: The echo suppressor EC is provided with an adaptive filter 1 and a subtractor 2, and an amplifier 3 is provided to a pre-stage of the adaptive filter 1. A sum of a gain G0 [dB] of a feedback path HAB measured in advance in a state that the front of a microphone M and a speaker S is set to non reflection and a marginal gain MG [dB] taking into account a reflection condition in an installed environment of an estimated loudspeaker device is selected to be the gain G of the amplifier 3. Thus, even when the adaptive filter 1 comprises a fixed decimal point arithmetic operation processor and the gain of the feedback path HAB reaches 1 or over, no overflow takes place in the arithmetic operation of the adaptive filter 1 and a desired (e.g. the same degree when the gain is less than 1) echo suppression amount can be obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an echo suppressor used for a loudspeaker apparatus (telephone, intercom, PHS, etc.) having a loudspeaker (hands-free) function.

[0002]

2. Description of the Related Art In a loudspeaker apparatus such as a telephone, an intercom, and a PHS, which realizes a loudspeaker call (hands-free call) using a microphone and a speaker, a part of a received voice transmitted from the speaker wraps around the microphone. Since it is transmitted as a transmission voice (transmission signal) to the call terminal of the other party, the voice of the other party can be heard as an echo by the other party, and if the level is high, It will cause discomfort. Therefore, a technique for suppressing the above-described echo has been conventionally proposed or provided, and as one of the means, there is an echo suppressor (a so-called echo canceller).

FIG. 7 is a partially omitted block diagram of a conventional loudspeaker using an echo suppressor EC '. The loudspeaker apparatus includes a microphone M, a speaker S, a microphone amplifier G1 for amplifying a voice signal (sending signal) output from the microphone M, and an A / A for converting the sending signal amplified by the microphone amplifier G1 from analog to digital.
D converter AD, which converts a digital reception signal output from the echo suppressor EC 'into an analog reception signal
A / A converter DA and a speaker amplifier G2 that amplifies the analog reception signal converted by the D / A converter DA and outputs the amplified signal to the speaker S.

The echo suppressor EC 'adaptively identifies an impulse response of a feedback path (echo path) H _AB formed by acoustic coupling between the speaker S and the microphone M, and inputs an input signal (a reception signal) to the feedback path. ) To return path H _AB
And a subtractor 2 for subtracting the echo component estimated by the adaptive filter 1 from an output signal (sending signal) from a feedback path. The adaptive filter 1 is a conventional well-known filter comprising a processor having a variable coefficient and an algorithm for determining the coefficient as needed. The adaptive filter 1 minimizes the root-mean-square value of the output signal of the subtracter 2 (signal at point C in FIG. 7). Algorithm, for example,
Estimating an echo component of the feedback path H _AB (wraparound component of the received signal via a feedback path H _AB) by a variable coefficient adaptive updating the LMS (the Least-Mean-Square ) algorithm. Then, by subtracting the echo component estimated by the adaptive filter 1 from the transmission signal in the subtracter 2, only the echo component included in the transmission signal is canceled, and components other than the echo collected by the microphone M ( It does not give any loss to the microphone M (voice or ambient noise emitted from a caller).

[0005]

The adaptive filter
Eco that data 1 consists of fixed-point arithmetic processor
-In the suppression device EC ', the return path H_ABDepending on the characteristics of
Therefore, there is a characteristic that the obtained echo suppression amounts are different.
Normally, each coefficient of the adaptive filter 1 falls within the range of -1 to 1.
In this case, the feedback path H_ABof
When the gain is more than 1 time, coefficient overflow occurs.
I will. In the operation of the adaptive filter 1, the coefficient
When the bar flow occurs, the echo suppression amount is reduced.
Therefore, adaptive filters can be used with fixed-point arithmetic processors.
In the case where the filter 1 is configured, the return path H _ABGain of 1 or less
The gain of the amplifier in each part of the communication system is set so that
Need to be measured. For example, in the loudspeaker apparatus shown in FIG.
Here, the gain of the speaker amplifier G2 is the D / A converter DA
And the electroacoustic conversion characteristics of the speaker S,
The maximum sound pressure to be transmitted from the speaker S as a communication device
Is inevitably determined. Also, from the speaker S to the microphone
The gain of the sound propagation system to M depends on the structure and installation
This value is determined from the environment (ambient reflection conditions).

Therefore, in order to make the gain of the feedback path H _AB equal to or less than one time, the gain of the microphone amplifier G1 is designed to be equal to or less than a predetermined threshold (the reciprocal of the product of the gain of the speaker amplifier G2 and the gain of the sound propagation system). There is a need to. In other words, it is necessary to design the gain of the microphone amplifier G1 to a smaller value as the maximum sound pressure level to be transmitted from the communication device is higher and the distance between the speaker S and the microphone M is shorter and the gain of the sound propagation system is larger. is there. As a result, the A / D converter A
S / N ratio of the input signal to D (the signal /
(Noise ratio) is reduced, and there is a possibility that the sound quality and the suppression characteristics of the transmission signal deteriorate.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and an object of the present invention is to achieve a desired echo without causing overflow in the operation of an adaptive filter even if the gain of a feedback path is 1 or more. It is an object of the present invention to provide an echo suppression device capable of obtaining a suppression amount.

[0008]

According to a first aspect of the present invention, there is provided a voice communication system for performing a voice communication using a microphone and a loudspeaker. In an echo suppression device for suppressing echo, an adaptive filter that adaptively identifies an impulse response of a feedback path formed by acoustic coupling between a speaker and a microphone and estimates an echo component of the feedback path from an input signal to the feedback path. A subtractor for subtracting the echo component estimated by the adaptive filter from the output signal from the feedback path, and a subtractor provided before or after the adaptive filter so that the reflection between the echo component and the microphone and the speaker is in a non-reflection state. Has a gain obtained by adding a predetermined margin gain to the feedback gain of the feedback path when the predetermined minimum value is obtained. Characterized by including an amplifier,
The input to the adaptive filter or the output from the adaptive filter is
The echo component is amplified by a gain obtained by adding a predetermined margin gain to the feedback gain of the feedback path when the reflection between the microphone and the speaker becomes a non-reflection state and reaches a predetermined minimum value. Even if the gain of the feedback path is 1 or more, it is possible to obtain a desired echo suppression amount (for example, about the same as when the gain is less than 1) without causing overflow in the operation of the adaptive filter.

A second aspect of the present invention is characterized in that, in the first aspect of the present invention, the amplifier comprises a variable amplifier having a variable gain, and the gain is set externally before the start of the voice communication. It is possible to set an appropriate gain in the amplifier according to the communication system with the side communication terminal, and a desired amount of echo suppression can be obtained even when any communication system is formed.

According to a third aspect of the present invention, in the first aspect of the present invention, the amplifier comprises a variable amplifier having a variable gain, and an input signal to the subtractor is compared with an output signal from the subtractor to determine an echo component. An echo suppression amount estimating unit for estimating the amount of suppression, and an optimum gain extracting unit for setting an amplifier by extracting an optimum gain in which the amount of suppression during a call estimated by the echo suppression amount estimating unit is maximum in a finite gain section. The present invention is characterized in that the amplifier gain can be set to an optimum value during a call.

[0011] The invention of claim 4 is the invention of claim 1 or 2 or 3.
The invention according to the invention is characterized in that it comprises an out-of-band removal filter for removing a frequency component other than the frequency band of the voice by the loudspeaker from the input signal to the subtractor, because there are many frequency components other than the voice frequency band. Even when the adaptive filter alone cannot provide a sufficient echo suppression amount, the out-of-band removal filter can remove frequency components other than the voice frequency band, thereby improving howling resistance.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment 1) FIG. 1 is a block diagram, partly omitted, of a loudspeaker apparatus using an echo suppressor EC of this embodiment. Here, of the configuration of the echo suppressor EC of the present embodiment, the conventional echo suppressor EC ′
The same components as those described above are denoted by the same reference numerals, and description thereof is omitted. Further, since the configuration of the loudspeaker apparatus is the same as that of the conventional example, the same components are denoted by the same reference numerals and description thereof will be omitted.

The present embodiment is characterized in that the amplifier 3 is provided in the preceding stage of the adaptive filter 1 and the gain G of the amplifier 3 is set to the following value. That is, the gain G0 [dB] of the feedback path H _AB measured in advance in a state where the front of the microphone M and the speaker S is non-reflective (a state where no object that reflects a sound wave is present ahead).
And the margin gain MG [dB] in consideration of the reflection condition in the assumed environment of the loudspeaker communication device is defined as the gain G of the amplifier 3 (G = G0 + MG). However, when the gain G thus calculated is smaller than 0 [dB], G = 0 [dB].

Thus, the number of input signals to the adaptive filter 1 increases.
It is multiplied by G by the width unit 3, and the gain G of the amplifier 3 is
Return path H in actual use_ABAdaptive gain.
Filter 1 is composed of a fixed-point arithmetic processor.
Return path H _ABGain is more than 1 times
However, overflow occurs in the operation of the adaptive filter 1.
Desired, for example, if the gain is less than 1x
Echo suppression amount).

In this embodiment, the amplifier 3 is provided before the adaptive filter 1. However, it is needless to say that the same effect can be obtained even if the amplifier 3 is provided after the adaptive filter 1. Further, if a variable amplifier having a variable gain is used for the amplifier 3, the gain G
Can be changed, and versatility can be improved.

(Embodiment 2) This embodiment is characterized in that the amplifier 3 is a variable amplifier and the gain G of the amplifier 3 is set from the outside before the start of communication. In addition, the loudspeaker apparatus according to the present embodiment is configured as a master unit of a so-called hands-free intercom capable of realizing a two-way simultaneous call with a door phone slave unit (not shown) as a call terminal of the other party. . As shown in FIG. 2, the master unit receives a call start signal or the like which is generated by pressing a call button (not shown) and serves as a trigger signal for starting a call with the intercom slave unit. A call control unit 10 comprising a CPU mounted with a call path control module 11 for controlling connection and opening of switches and relays, and a gain setting module 12 for setting a gain G for the amplifier 3 of the echo suppressor EC.
And a nonvolatile memory 13 for storing a candidate value of the gain G.

Here, the master unit in the present embodiment is capable of talking with a plurality of intercom units and other communication terminals such as other master units via a wired or wireless transmission system. In this way, when there are a plurality of communication terminals on the other side viewed from the master unit and these types are different, the optimum gain G of the amplifier 3 depends on which of the communication terminals forms a communication path. The values will be different.

Therefore, in the present embodiment, a plurality of candidate values of the gain G of the amplifier 3 corresponding to the various situations described above are stored in the nonvolatile memory 13 in advance, and are stored in the communication path control module 11 of the communication control unit 10. After the communication path between the master unit and the other party's communication terminal is determined, the gain setting module 12 selects a candidate value of the gain G suitable for the determined communication path, reads it out from the nonvolatile memory 13, and starts the communication. The candidate value of the gain G read to the amplifier 3 previously is transferred and set. Thus, the gain G thus optimized for each communication system
Is stored in the nonvolatile memory 13, it is possible to easily cope with various communication systems without changing the circuit configuration.

Although a plurality of candidate values of the gain G are stored in the nonvolatile memory 13 in the present embodiment, they may be stored in a mask ROM provided in a CPU constituting the communication control unit 10.

(Embodiment 3) In this embodiment, the amplifier 3 is a variable amplifier, and the input signal to the subtractor 2 (the transmission signal before echo component suppression) and the output signal from the subtractor 2 (after the echo component suppression) The echo suppression amount estimating unit 4 for estimating the amount of suppression of the echo component by comparing the transmission signals of the echo signals with each other, and the optimum amount of suppression during a call estimated by the echo suppression amount estimating unit 4 becomes maximum in the finite gain section. It is characterized in that it has an optimum gain extraction unit 5 that extracts the gain G and sets it to the amplifier 3. The echo suppression amount estimating unit 4 and the optimum gain extracting unit 5 can be configured by a DSP or a CPU.

Next, the operation of the echo suppression amount estimating unit 4 and the optimum gain extracting unit 5 will be described with reference to the flowchart of FIG. In the optimal gain extracting unit 5, N optimal gain candidate values G
_{0 to} G _N-1 are stored in advance, and when the process of extracting the optimum gain is started during a call, the first candidate value G ₀ is firstly stored in the amplifier 3.
Is set to the gain G. After the elapse of a predetermined wait time, the echo suppression amount estimating unit 4 compares the input / output signals of the subtracter 2 to obtain an echo suppression amount estimation value ERLE ₀ for the candidate value G ₀ . Subsequently, the optimum gain extracting unit 5 sets the second candidate value G ₁ to the gain G of the amplifier 3, and after a predetermined wait time has elapsed, the echo suppression amount estimating unit 4 sets the candidate value G _1.
Of the echo suppression amount ERLE ₁ with respect to. Hereinafter, up to the estimated value ERLE _N-1 of the echo suppression amount for the _N- th candidate value G _N-1 are obtained, and these N estimated values E
A maximum value ERLE _k of RLE _{0 to} ERLE _N−1 is selected, and a candidate value G _{k at} which the maximum value ERLE _k is obtained is set in the amplifier 3 as an optimum value of the gain G.

Since the present embodiment is configured as described above, the gain G of the amplifier 3 can be set to an optimum value during a call. Therefore, if a sufficient number of candidate values G ₀ ,... Corresponding to various assumed communication systems are prepared, it is not necessary to measure in advance for each communication system to obtain an optimum gain, and it is easy to use. There is an advantage that the improvement can be achieved.

(Embodiment 4) By the way, the frequency characteristic (amplitude characteristic) of the feedback path H _AB in the master unit described so far.
Since the amplitude characteristics of the microphone M and the microphone amplifier G1 can be considered to be substantially flat, it is determined by the frequency characteristics of the speaker S and the structure of the housing of the parent device (including the way of mounting the microphone M and the speaker S). Then, depending on the structure of the housing, a sharp peak P may occur in the amplitude characteristic as shown in FIG. However, the input signal (received signal) to the adaptive filter 1 is a received voice transmitted from the other party's call terminal, and most of the energy is in a frequency band corresponding to a human voice (this is called a "call band"). Concentrated in T. Therefore, adaptive filter 1
In this case, the transfer characteristics of the feedback path H _{AB in} the communication band T can be approximated with high accuracy, but the transfer characteristics outside the communication band T have lower approximation accuracy. Therefore, a desired suppression effect cannot be obtained for the peak P of the echo component included in the transmission signal outside the communication band T. For example, as shown in FIG. When a closed loop is formed in the circuit, the loop gain of the closed loop becomes large near the frequency of the peak P outside the communication band T, and howling tends to occur.

Therefore, in this embodiment, as shown in FIG. 5, an out-of-band filter 6 for removing frequency components other than the bandwidth T from the input signal to the subtractor 2 is provided.
The out-of-band filter 6 comprises a band-pass filter (band-pass filter) in which the band T is substantially matched to the passband frequency range, and is provided on the input side of the subtractor 2.

Since the present embodiment is constructed as described above, even if the adaptive filter 1 alone cannot provide a sufficient amount of echo suppression due to the large number of frequency components other than the speech band T, The frequency components other than the communication band T can be removed by the communication band out-of-band removal filter 6, and the resistance to howling can be improved.

[0026]

According to a first aspect of the present invention, there is provided an echo suppressor for use in a loud-speaking communication system for performing a loud-speaking call using a microphone and a loudspeaker, wherein the loudspeaker and the microphone are suppressed. An adaptive filter that adaptively identifies the impulse response of the feedback path formed by acoustic coupling of the filter, and estimates the echo component of the feedback path from the input signal to the feedback path, and returns the echo component estimated by the adaptive filter to the feedback path. A subtractor for subtracting from the output signal of the adaptive filter, and a feedback provided when the echo component has a predetermined minimum value at which the reflection between the microphone and the speaker becomes a non-reflection state, provided before or after the adaptive filter. And an amplifier having a gain obtained by adding a predetermined margin gain to the feedback gain of the path.
The input to the adaptive filter or the output from the adaptive filter is
Since the echo component is amplified by a gain obtained by adding a predetermined margin gain to the feedback gain of the feedback path when the reflection between the microphone and the speaker becomes a non-reflection state and reaches a predetermined minimum value. Even if the gain is 1 or more, the overflow does not occur in the operation of the adaptive filter, and it is desired (for example, about the same as when the gain is less than 1).
This has the effect that the amount of echo suppression can be obtained.

According to a second aspect of the present invention, in the first aspect of the present invention, the amplifier comprises a variable amplifier having a variable gain, and the gain is externally set before the start of the loudspeaking call. Therefore, it is possible to set an appropriate gain according to the communication system of the amplifier in the amplifier, and it is possible to obtain a desired amount of echo suppression even when any communication system is formed.

According to a third aspect of the present invention, in the first aspect, the amplifier is a variable amplifier having a variable gain, and an input signal to the subtractor and an output signal from the subtractor are compared to determine the echo component. An echo suppression amount estimating unit for estimating the amount of suppression, and an optimum gain extracting unit for setting an amplifier by extracting an optimum gain in which the amount of suppression during a call estimated by the echo suppression amount estimating unit is maximum in a finite gain section. With this arrangement, the gain of the amplifier can be set to an optimum value during a call.

The invention of claim 4 is the invention of claim 1 or 2 or 3
In the invention according to the invention, since an out-of-speech band removal filter for removing a frequency component other than the frequency band of the voice by the loudspeaker from the input signal to the subtractor is provided, the adaptive filter Even when only a sufficient amount of echo suppression cannot be obtained with only the above, there is an effect that frequency components other than the frequency band of the voice can be removed by the out-of-band removal filter, and the resistance to howling can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment.

FIG. 2 is a block diagram showing a second embodiment.

FIG. 3 is a block diagram showing a third embodiment.

FIG. 4 is a flowchart for explaining the operation of the above.

FIG. 5 is a block diagram showing a fourth embodiment.

FIG. 6 is an operation explanatory view of the above.

FIG. 7 is a block diagram showing a conventional example.

[Explanation of symbols]

 EC echo suppressor 1 Adaptive filter 2 Subtractor 3 Amplifier

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Akira Terasawa, Inventor 1048, Kazuma Kadoma, Kadoma, Osaka Prefecture F-term in Matsushita Electric Works, Ltd.

Claims (4)

[Claims] 1. An echo suppressor used in a loud-speaking communication system for making a loud-speaking call using a microphone and a loudspeaker. An adaptive filter that adaptively identifies the impulse response of the return path and estimates the echo component of the return path from the input signal to the return path, and subtracts the echo component estimated by the adaptive filter from the output signal from the return path. The subtractor and the adaptive filter are provided before or after the adaptive filter, and the feedback gain of the feedback path when the echo component has a predetermined minimum value at which the reflection between the microphone and the speaker becomes a non-reflection state is set to a predetermined value. An echo suppressor, comprising: an amplifier having a gain obtained by adding a margin gain. 2. The echo suppressor according to claim 1, wherein said amplifier comprises a variable amplifier having a variable gain, and the gain is set from outside before the start of the loudspeaker call. 3. An echo suppression amount estimator for comparing the input signal to the subtractor and the output signal from the subtractor to estimate the amount of suppression of the echo component, wherein the amplifier comprises a variable amplifier having a variable gain; 2. The apparatus according to claim 1, further comprising: an optimum gain extracting unit configured to extract an optimum gain in which a suppression amount during a call estimated by the echo suppression amount estimating unit is maximum in a finite gain section and set the optimum gain in an amplifier. Echo suppression device. 4. The echo according to claim 1, further comprising a speech band out-of-band removal filter for removing a frequency component other than the frequency band of the voice of the loudspeaker speech from the input signal to the subtractor. Suppression device.