JPH09130306A - Loud speaking device and echo canceller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To constitute the whole of the device with a low cost by cancelling the echo by an adaptive filter and a non-adaptive filter in the case of the occurrence of the variance of the echo path characteristic. SOLUTION: A connection control means 2 connects a busy channel selected by a reception detection means 1 to a first signal path 3. When an interrupt occurs, the interrupt channel is connected to a second signal path 4. An adaptive filter(ADF) 5 estimates the echo passing round from the reception channel to the transmission channel while learning to generate a pseudo echo. When there are two busy reception channels, a non-adaptive filter(NADF) 6 loads a filter coefficient from the outside and estimates the passing-round echo to generate a pseudo echo. A subtractor 8 subtracts the pseuco echo signal from a transmission signal. A reception filter 9 performs operation to output a sound signal to right and left speakers 10 and 11, and an acoustic image is orientated in accordance with the position of a speaker on the picture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a loudspeaker communication device having a two-speaker and one-microphone structure used in, for example, a multipoint communication conference system, and an echo canceller used in this device.

[0002]

2. Description of the Related Art In recent years, personal computers (PCs) and workstations (PCs) have been used as a service in which people in remote areas share images, voices, documents, etc., via a network to build a close cooperative work environment. WS)
A multi-point communication conference system using multimedia terminals such as is attracting attention.

In this multipoint communication conference system, people at remote locations can talk while looking at each other's faces on a monitor screen as if they were facing each other. In addition, one document or application can be shared on the display of each terminal for meetings and collaborative work.

The terminal of the conventional multipoint communication conference system has, for example, a structure in which a microphone and a speaker are additionally provided in WS or the like. In this case, the images of a plurality of speakers are displayed on the monitor of each WS, and When one of them speaks, the voice spoken to the other party is collected by a microphone and output from a speaker of a remote terminal via a communication line to hold a video conference.

At this time, the conference participants (listeners) of the respective terminals are
Who is talking is judged from the movement of the speaker's mouth on the monitor and the sound generated from the speaker.

In this multipoint communication conference system,
In order to recognize the speaker from the voice or to smoothly proceed with the conference, it is effective to localize the sound image to a different position for each speaker using a plurality of speakers.

In this case, by controlling the level and delay time of the sound output from each speaker on the listener side, the sound image can be localized at the position corresponding to the speaker on the screen.

At present, there is a strong demand for a hands-free voice call due to the convenience of this communication conference system, and a so-called echo canceling technique (echo canceling technique) that suppresses echo feeling and howling that are generated when a speaker output voice wraps around a microphone is generated. Canceller) is devised. A pseudo stereo loudspeaker is a combination of the echo canceller and the pseudo stereo voice generation technology.

In this pseudo-stereo loudspeaker system, as shown in FIG. 6, the transfer function (hereinafter referred to as a sound image localization function) G _Ri (z) of the input monaural audio signal is processed by the sound image localization processing units 111 and 112. , G _Li (z), and outputs from the speakers 113 and 114 connected to each. At this time, before the sound image localization processing units 111 and 112,
Alternatively, the audio signal obtained by the echo canceller 115 from the subsequent stage and the microphone 116 generated from the speakers 113 and 114
The echo signal is canceled by subtracting the sound signal collected in step (1) with the subtractor 117.

In this case, a plurality of speakers 113, 114
When the sound is output from the microphone, each sound is heard by the listener and also collected by the microphone 116. This microphone 116
In this case, since each voice is input from different directions, there are a plurality of echo paths, and the processing amount of the echo canceller 115 increases to suppress the plurality of echoes, and the residual echo increases when the speaker changes. Occurs.

A conventional pseudo-stereo loudspeaker system is shown in FIG.
As shown in FIG. 5, the adaptive filter (ADF) 121 has an adaptive filter (ADF) 121 that learns with the monaural audio signals in the preceding stages of the sound image localization processing units 111 and 112 and the signal subtracted by the subtractor 117. Learns the acoustic characteristics of the echo path, so that the echo when the sound output from each of the speakers 113 and 114 goes into the microphone 116 is erased. In this case, the adaptive filter (ADF)
The single 121 has almost the same amount of processing and memory capacity as in monaural.

At this time, if the acoustic characteristic of the total echo path seen from the echo canceller is H _i (z), the acoustic characteristic H _i (z) can be expressed by the following equation (10).

[0013] _{H i (z) = G Ri} (z) H R (z) + G Li (z) H L (z) ...... (10) from the equation (10), the sound image localization function G _Ri (z) , G _Li (z)
It can be seen that, etc. are included in the acoustic characteristic H _i (z) of the echo path.

Transfer function H between the left and right speakers and microphones
_R (z) and _HL (z) are constant as long as the acoustic characteristics are stationary, but the sound image localization functions G _Ri (z) and G _Li (z) change each time the speaker changes.

By the way, in the multipoint communication conference system, the number of participants in the conference is large, so that the speaker frequently changes.

However, in the pseudo-stereo loudspeaker system shown in FIG. 7, when there is a change of speaker, that is, when the sound image localization function changes, the echo path characteristic H _i is changed.
(z) fluctuates and the echo cannot be canceled sufficiently.

Therefore, as shown in FIG. 8, a plurality of adaptive filters (ADF) 122,
A linear combination type using 123 and an adder 124 for adding them can be considered. For this linear combination type,
After the sound image localization calculation, echoes are canceled by one adaptive filter (ADF) 122, 123 for each echo path from the left and right speakers 113, 114 to the microphone 116, so the echo path characteristics change even if the sound image localization function changes. Instead, if the adaptive filters (ADF) 122 and 123 are once converged by the received voice from any one of the plurality of points, the amount of echo cancellation even if the speaker changes, unless the echo path changes thereafter. No longer occurs. However, in this case, twice the amount of calculation and memory capacity are required as compared with the case where only one adaptive filter (ADF) is used, which causes a cost increase.

[0018]

As described above, the conventional pseudo-stereo loudspeaker system causes problems such as an increase in residual echo, an increase in the amount of memory, and an increase in the amount of calculation when the speaker is changed, as compared with the monaural sound system. .

Further, in the above-described pseudo-stereo loudspeaker, the echo canceller is provided for a single speaker, so that two people at two different points can have a discussion as in a multipoint communication conference. The situation where two different speakers talk at the same time, such as a situation where a person is fighting or a situation where another conference participant interrupts and asks a question during the explanation by one person, is not taken into consideration. In this case, the echo is canceled. Absent.

Therefore, the configuration of the above-described pseudo-stereo loudspeaker system, that is, a plurality of adaptive filters (ADF), is provided so that the echo can be canceled even at the time of the two-party simultaneous call.
If is applied as is, the calculation amount and memory amount are 2
There is a problem in that the cost is doubled and the cost of the entire apparatus is significantly increased.

The present invention has been made in order to solve such a problem and is capable of canceling an echo when a speaker is changed or a simultaneous call is made in multipoint communication, and is compared with a conventional one. Therefore, it is an object of the present invention to provide a loudspeaker communication device and an echo canceller that can configure the entire device at a relatively low cost without significantly increasing the calculation amount.

[0022]

In order to achieve the above-mentioned object, a loudspeaker communication apparatus according to the invention of claim 1 has a plurality of receiving channels for receiving receiving signals from a plurality of points,
A reception detecting means for detecting a reception signal received on each of the reception channels, and a reception channel on which the reception signal is received according to the number of receptions of the reception signal detected by the reception detection means. Connection control means connected to any one of the signal path of, a plurality of speakers for localizing the sound image of the received voice to a different position for each voice,
The sound image localization means for controlling the outputs of the plurality of speakers with respect to the respective received signals input through the first and second signal paths to localize the sound image to the positions corresponding to the respective received signals, and the voice of the speaker. A microphone for collecting sound is connected to the first signal path, and an echo path that wraps around the transmission channel side from the reception channel through the first signal path, the sound image localization means, the speaker, and the microphone is estimated, and a pseudo echo is generated. An adaptive filter to be generated, and an echo path information storage unit in which filter coefficients corresponding to the echo path estimated by the adaptive filter are stored,
Loaded from the echo path information storage means is a filter coefficient corresponding to an echo path which is connected to the second signal path and wraps around the transmission channel side from the reception channel through the second signal path, the sound image localization means, the speaker and the microphone. A non-adaptive filter that estimates an echo that wraps around to the transmission channel side through the second signal path and generates a pseudo echo, and subtracts each pseudo echo generated by the non-adaptive filter and the adaptive filter from the echo signal of the transmission channel. It is equipped with a container.

In the case of the invention described in claims 1 and 5, when there are two receiving channels among the plurality of receiving channels, an adaptive filter and a non-adaptive filter having a smaller processing amount than this are provided. Each echo is estimated by using it to generate a pseudo echo, and a subtracter cancels the subtraction echo from the echo signal on the transmission channel side. The non-adaptive filter used here is
Since the processing of sequentially learning / updating the filter coefficient is not performed, the amount of calculation is smaller than that of the adaptive filter and the entire apparatus can be configured at a relatively low cost.

A loudspeaker communication apparatus according to a second aspect of the present invention is the loudspeaker communication apparatus according to the first aspect, wherein the sound image localization means produces a sound image at a corresponding position with respect to a received signal input through the first signal path. A first signal calculation unit that performs a function calculation for each speaker for localizing, and a function calculation for each speaker that localizes a sound image at a corresponding position with respect to a received signal input through the second signal path. It is characterized in that it is composed of a second signal operation unit for performing operation and a plurality of operation results for the same speaker calculated by the first and second signal operation units.

In the case of the present invention as defined in claim 2, since the sound image localization means is provided with the first signal calculation section, the second signal calculation section and the respective adders, there are two speakers. Even if it becomes, the sound image can be localized for each.

A loudspeaker communication apparatus according to a third aspect of the present invention is the loudspeaker communication apparatus according to the first aspect, wherein the echo information storage means stores acoustic characteristic information corresponding to each sound image localization position as a filter coefficient. It is characterized by having a memory of.

According to the third aspect of the present invention, when the sound image localization changes, the data of the filter coefficient depending on the occasion can be loaded from each memory.

According to a fourth aspect of the present invention, there is provided the loudspeaker communication apparatus according to the first aspect, wherein the echo information storage means has a unit in which the transfer function estimated by the adaptive filter is in a stationary section. First storage means for storing a plurality of past echo path information including acoustic characteristics of the sound image localization means, and past plurality of echo path information including acoustic characteristics of the sound image localization means stored by the first storage means A means for obtaining echo path information excluding the acoustic characteristics of the sound image localization means and a number of echo paths extending from the speaker to the microphone for storing echo path information excluding the obtained acoustic characteristics of the sound image localization means Second storage means and a method for obtaining echo path information including the following acoustic characteristics of the sound image localization means based on the echo path information of the second storage means. It is characterized by comprising and.

In the case of the invention described in claim 4, the estimated transfer functions H _i-1 '(z) and H _i ' (z) are the first as the plurality of past echo path information including the acoustic characteristics of the sound image localization means. Of the estimated transfer functions H _i-1 '(z),
From H _i '(z) to the transfer function between speaker and microphone H _R (z)
, H _L (z) are calculated and stored in the second storage means. Then, the transfer function H _{i + 1} '(z) of the adaptive filter 5 in the (i + 1) th section is obtained from the transfer functions H _R (z) and H _L (z).

An echo canceller according to a fifth aspect of the present invention includes a reception detecting means for detecting reception signals received by a plurality of reception channels, and the reception according to the number of receptions of the reception signals detected by the reception detecting means. Connection control means for connecting the receiving channel from which a signal is received to either one of the first or second signal paths, and the first signal path, which is connected to the first signal path, from the reception channel to the speaker And an adaptive filter for estimating an echo path sneaking to the transmission channel side through a microphone to generate a pseudo echo, an echo path information accumulating means for accumulating filter coefficients corresponding to the echo path estimated by the adaptive filter, and the second signal path. And a filter coefficient corresponding to an echo path that wraps around from the second signal path to the transmission channel. A non-adaptive filter for estimating an echo path which is loaded from the multipath information storage means and circumvents the transmission channel through the second signal path to generate a pseudo echo, and each pseudo echo generated by the non-adaptive filter and the adaptive filter. And a subtractor for subtracting from the echo signal.

[0031]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a pseudo stereo loudspeaker system according to the first embodiment of the present invention.

In the figure, reference numeral 1 is a reception detection means,
The lines of the respective reception channels A1 to E1 from different points A to E, etc. are connected, and the reception channel in which the reception signal is received is detected therein. Reference numeral 2 denotes a connection control means, which connects the currently receiving channel detected by the reception detection means 1 to the first signal path 3. Also, when a certain channel is receiving a call and another channel receives a new received signal (interrupt), the interrupt channel is connected to the second signal path 4. 5 is an adaptive filter (A
DF), for example with about 200 taps. The adaptive filter 5 learns and estimates the echo that wraps around from the reception channel to the transmission channel, and generates a pseudo echo. A non-adaptive filter 6 is, for example, a finite impulse response filter (FIR filter). Although this non-adaptive filter 6 has the same number of taps as the adaptive filter 5, it is not a type of filter that sequentially learns and updates the filter coefficient like the adaptive filter 5, and therefore the data processing amount is equivalent to that number. 1 is enough. This non-adaptive filter 6 loads a filter coefficient (characteristic information of the echo path including the receiving filter 9) from another when the number of receiving channels being received becomes two, and estimates a wraparound echo using the filter coefficient. Then generate a pseudo echo. Reference numeral 7 denotes a filter coefficient storage means having a plurality of memories, the filter coefficients converged by the adaptive filter (ADF) 5 are saved in each memory, and the filter coefficients are adapted to the adaptive filter (ADF) 5 or non-adaptive. Filter (NA
DF) 6 and so on. A subtracter 8 subtracts the pseudo echo signal from the transmission signal. Reference numeral 9 is a reception filter as a sound image localization processing means, and the left and right transfer functions G
Speaker 1 for the right side by calculating _Ri (z) and G _Li (z)
0, a voice signal is output to the left speaker 11, and a sound image is localized according to the position of the speaker on the screen. 1
Reference numeral 2 is a microphone for collecting voices when an operator (conference participant) of this apparatus speaks to another. In this pseudo stereo loudspeaker,
Adaptive filter (ADF) 5 and non-adaptive filter (NA
The echo canceller is composed of the DF) 6, the filter coefficient accumulating means 7, the subtracter 8 and the like.

Next, the structure of the reception filter 9 will be described with reference to FIG.

As shown in the figure, the reception filter 9 is a filter 21 (having a transfer function of G1 _Ri (z)) for generating a right-side voice and a left-side voice for the reception signal from the first signal path 3. A filter 22 (having a transfer function of G1 _Li (z)), a filter 23 (having a transfer function of G2 _Ri (z)) for generating a right-side voice with respect to a received signal from the second signal path 4, and a left-side voice. A filter 24 (having a transfer function of G 2 _Li (z)), an adder 25 for adding the calculation results of the right voices and outputting to the speaker 10, an adder 26 for adding the calculation results of the left voices and outputting to the speaker 11, and the like. Have

Next, the configuration of the filter coefficient storage means 7 will be described in detail with reference to FIG.

As shown in the figure, the filter coefficient storage means 7 stores the estimated transfer functions HA '(z) to HE' (z) which are the filter coefficients of the pseudo echo corresponding to the receiving channels A1 to E1. Has a plurality of memories 31 to 35.

Next, the relationship between the filter coefficient storage means 7 and the adaptive filter 5 will be described with reference to FIG.

As shown in the figure, for example, the memory 3 for the estimated transfer function HA '(z) in the filter coefficient storage means 7 is provided.
Between 1 and the adaptive filter 5, the received signal X (data X0 to X99) from the signal path 3 is accumulated at addresses 0 to 99 of the coefficient memory (not shown) in the adaptive filter 5,
Also, from the address 100 of the coefficient memory in the adaptive filter 5,
The filter coefficients h0 to h99 are accumulated at the addresses, the convolution operation is performed between them, and the operation result is output as Y. Then, the data at address 200 in memory 31 is loaded into address 100 in the coefficient memory in adaptive filter 5, the data at address 201 is loaded in address 101, or conversely the data at address 100 in coefficient memory in adaptive filter 5 is loaded. Is saved in the memory at address 200, and so on.

That is, in the case of receiving from the point A, the data is loaded from the memory 31 to the coefficient memory in the adaptive filter 5, the filter coefficient is converged (learning) by the voice from the point A, and the partner A When switching to a location other than the point, the data in the coefficient memory in the adaptive filter 5 is stored in the memory 3
Save to 1 and load the data from the memory corresponding to the new point to the coefficient memory in the adaptive filter 5. The learning of the adaptive filter 5 is continued using this coefficient.

Next, the operation of this pseudo-stereo loudspeaker will be described.

In the case of this pseudo-stereo loudspeaker, when a reception signal from, for example, point A is input to the reception channel A1, the reception detection means 1 detects it and the connection control means 2 detects the first signal path. 3 to the first signal path 3
The reception signal is input to the reception filter 9 and the adaptive filter 5 through. In the reception filter 9, it is possible to know from which position the sound image should be localized from the reception channel A1 detected by the reception detecting means 1. Therefore, the transfer function is calculated to localize the sound image at that position, and each speaker 10,
The gain to 11 is controlled. And each speaker 10, 1
The sound output from 1 is the microphone 12 as an echo signal.
Is collected and transmitted to the subtractor 8.

On the other hand, in the adaptive filter 5, a corresponding filter coefficient is loaded from the filter coefficient accumulating means 7 on the basis of the input received signal, the echo sneaking from the receiving channel A1 to the transmitting channel 13 is estimated, and a pseudo echo is generated. It is generated and output to the subtractor 8.

Then, in the subtracter 8, the pseudo echo and the collected echo are subtracted, and the difference signal is fed back to the adaptive filter 5, and the adaptive filter 5 eliminates the difference signal, that is, echo. When the voice input to the signal path 3 becomes a voice other than the channel A1 by learning to converge, it is saved in the filter coefficient storage means 7 as a new filter coefficient.

Here, for example, a received signal from the point B may be received on the channel B1 while receiving a call on the channel A1.

In this case, when the reception signal from the point B is received by the channel B1, the reception detection means 1 detects it and the connection control means 2 connects the channel B1 to the second signal path 4, The reception signal is input to the reception filter 9 and the non-adaptive filter 6 through the second signal path 4.

The reception filter 9 performs a function operation for sound image localization so that the sound image is localized at a predetermined position of the reception channel B1 detected by the reception detection means 1, and two output signals are added to each of the left and right sides. Then, the outputs of the speakers 10 and 11 are controlled. The sound output from each of the speakers 10 and 11 is collected by the microphone 12 as an echo and transmitted to the subtractor 8.

In this case, although a filter coefficient memory for the number of points is required, it is possible to cancel the echo with a processing amount equivalent to that when a monaural echo canceller is used. A communication device can be configured.

In addition to the above, the received signal may be received on the channel B1 immediately after the reception of the channel A1 is completed. In this case, the connection control means 2 connects the channel B1 to the vacant first signal path 3 after the reception of the call.

Then, the adaptive filter 5 on the first signal path 3 temporarily stores the filter coefficient converged by the received signal from the point A in the memory 3 for the point A in the filter coefficient storage means 7.
After saving to 1, the filter coefficient is loaded from the B point memory 32 instead, and thereafter, learning is performed by the received signal from the B point.

That is, even when the received signal is received on the channel B1, if the adaptive filter 5 is vacant, a pseudo echo is generated by the adaptive filter 5, so that the echo can be reliably canceled.

As described above, according to the pseudo-stereo loudspeaker system of this embodiment, as the core components of the echo canceller, the adaptive filter 5 whose filter coefficient is sequentially updated to the optimum value and the filtering operation by the loaded filter coefficient are performed. By using it in combination with the non-adaptive filter 6 that performs the above, it is possible to cancel the echo when two parties talk at the same time among many channels A1 to E1 at a time, and a plurality of speakers talk while alternating. A terminal used for a multipoint communication conference system, that is, a pseudo stereo loudspeaker can be constructed at low cost.

Conventionally, when there were two sound image localization positions, a single echo canceller could not suppress the echo.
Low cost non-adaptive filter 6 in addition to adaptive filter 5
By providing the (auxiliary echo canceller), the echo can be canceled by the respective filters 5 and 6 for the two-party simultaneous call section. However, since the filter coefficient of the auxiliary echo canceller is only loaded from the filter coefficient accumulating means 7, it is not possible to follow the acoustic characteristic variation of the echo path in a timely manner, but if there is no significant change in the acoustic characteristic between the speaker and the microphone, the echo is canceled. It can be suppressed sufficiently. Further, since the sound image localization control by the reception filter 9 of the pseudo-stereo loudspeaker according to this embodiment is a gain control that can be easily calculated, the pseudo-stereo sound can be reproduced with a processing amount equivalent to that of a monaural echo canceller.

Next, a modified example of the filter coefficient storage means 7 will be described with reference to FIG.

In the case of the filter coefficient storage means 7, the number of memories is required for the conference participants. For this reason, if the number of participants in the conference increases, the memory will be increased accordingly and the cost will increase.

Therefore, it is conceivable to modify the filter coefficient storage means in order to suppress the cost increase.

In this case, as shown in the figure, in the filter coefficient accumulating means 50, the sound localization function is stationary.
'a memory 52 for storing (z), likewise i-th estimated transfer function H _i of the interval' (i-1) -th estimated transfer function of the interval the adaptive filter 5 H _i-1 stores (z) Memory 51,
These estimated transfer function _{H i-1 '(z)} , H i' (z), the transfer between the speaker microphone function H _R (z), obtains the H _L (z) (decomposed), these speaker microphone Transfer function H between
_A decomposition synthesis processing unit 53 for obtaining an initial value of the transfer function H _{i + 1} '(z) of the adaptive filter 5 in the (i + 1) th interval from _R (z) and _HL (z) (synthesis processing); Two memories 54 for storing the speaker-microphone transfer functions H _R (z) and H _L (z),
55. An echo canceller having this coefficient accumulating means 50 is called an estimated transfer function decomposition / synthesis type echo canceller.

In this case, the estimated transfer functions H _i-1 '(z) and H _i ' (z) of the past two sections estimated by the adaptive filter 5 are stored in the memories 51 and 52, and the speaker transfer function The inter-microphone transfer functions H _R (z) and H _L (z) are obtained and the memory 54,
Since it is stored in 55, the echo canceller can be constructed relatively inexpensively.

The method of calculating the estimated transfer function by the estimated transfer function decomposition / synthesis type echo canceller will be described below.

First, consider the i-th section in which the unit in which the sound image localization function is stationary is a unit.

When the transfer functions of the past two sections (i section and i-1 section) estimated by the adaptive filter 5 are estimated transfer functions H _i-1 '(z) and H _i ' (z), the echo path is Assuming that the estimation is stationary and accurate, this estimated acoustic characteristic, that is, the estimated transfer function uses the speaker-microphone transfer functions H _R (z) and H _L (z) independent of the sound image localization function. Can be expressed as

H _i '(z) = G _Ri (z) H _R (z) + G _Li (z) H _L (z) H _i-1 ' (z) = G _Ri-1 (z) H _R (z ) + G _Li-1 (z) _HL (z) (1) Echo cancellers with multiple echo paths by decomposing and combining transfer functions are used to obtain multiple echo cancellers using this relationship. The transfer function between the speaker and the microphone is calculated based on the transfer function in the past.

[0063] _{H R (z) = D (} z) -1 (G Li-1 (z) H R (z) -G Li (z) H L (z)) H L (z) = D (z) _{^{-1 (-G Ri-1 (z}} ) H R (z) + G Ri (z) H L (z)) where _{D (z) = G Ri (} z) G Li-1 (z) -G Li (z ) G _Ri-1 (z) ………… (2) And when a new sound image localization function is obtained in the (i-1) section, H _{i + 1} '(z) = G _{Ri + 1} ( _{z) H R (z) +} G Li + 1 (z) H L (z) ............ formula (3) becomes the operation, it is possible to obtain a filter initial value of the echo canceller of the i + 1 interval. By using these plural filter coefficients in each filter, it is possible to realize an echo canceller in which the echo cancellation amount does not deteriorate due to the change of the speaker.

Generally, there are delay control and gain control as the sound image localization control method, but here, the case of gain control which is relatively easy to calculate will be described.

In this case, since the sound image localization function is G _Ri (z) = g _Ri , G _Li (z) = g _Li (4), H _R (z) and H _L (z) are , Is calculated by the following formula.

[0066] _{H R (z) = 1 /} D (g Li-1 H R (z) -g Li H L (z)) H L (z) = 1 / D (-g Ri-1 H R (z ) + G _{Ri HL} (z)) where D = g _Ri g _Li-1 −g _Li g _Ri-1 (5) When this estimated transfer function decomposition / synthesis type echo canceller is used, You can expect a great effect.

That is, in the case of the filter coefficient accumulating means 7 shown in FIG. 3, although the echo canceling amount is deteriorated until the initial learning at the start of the communication with the other party who receives the signal for the first time, the filter coefficient accumulating means 7 In the case of 50, if the learning from two points (that is, two steady states) is performed even if the call from all points of the other party is not completed, the deterioration of echo cancellation due to the initial learning does not occur thereafter. Even if the echo path between the speaker and microphone changes,
If the two points are learned again, then the echo cancellation amount will not be deteriorated due to the speaker change, as in the above.

In the loudspeaker system of the above embodiment, the adaptive filter 5, the non-adaptive filter 6 and the receiving filter 9 are used.
Since there are two signal paths to (1st signal path 3 and 2nd signal path 4), it is not possible to cancel echoes for three or more types of sound image localization functions at the same time.

Therefore, it is also possible to add a non-adaptive filter such as a fixed FIR filter to a call from three or more different points within a range where the processing amount can be allowed.

That is, a case where a plurality of second signal paths 4 and non-adaptive filters 6 are provided can be considered.

By adding a digital filter with a small processing amount in this way, it is possible to optimally cope with the frequency of simultaneous calls in an actual application and the cost.

Further, the present invention is not limited to the TV conference system in which the image of the other speaker can be confirmed on the monitor, but the speaker is not displayed on the screen, and the shared document or figure is displayed on the screen. It can also be applied to the teleconference and the system without the monitor screen, that is, the audio conference.

In the above embodiment, the simplest two-speaker, one-microphone type has been described, but the present invention can be applied to three or more speakers and a plurality of input-side microphones.

[0074]

As described above, according to the present invention, 2
In a loudspeaker communication device having a speaker and a microphone, when a voice from a speaker among a plurality of conference participants is reproduced in pseudo stereo, simultaneous talk between two parties and change of speakers occur, which may cause echo path characteristics. If fluctuations occur, the echo can be canceled by the adaptive filter and the non-adaptive filter.

Further, the non-adaptive filter for the second speaker does not automatically update the filter coefficient unlike the adaptive filter, so that the amount of calculation is not significantly increased as compared with the conventional one, so that the entire apparatus is It can be constructed at a relatively low cost.

[Brief description of the drawings]

FIG. 1 is a diagram showing an embodiment of a pseudo stereo voice communication device according to the present invention.

FIG. 2 is a diagram showing a detailed configuration of a reception filter of this pseudo-stereo loudspeaker.

FIG. 3 is a diagram showing a detailed configuration of a filter coefficient accumulating unit of this pseudo-stereo loudspeaker.

FIG. 4 is a diagram showing a relationship between a filter coefficient accumulating unit and an adaptive filter in FIG.

5 is a diagram showing a modification of the filter coefficient accumulating unit in FIG.

FIG. 6 is a diagram showing a conventional pseudo-stereo loudspeaker system.

FIG. 7 is a diagram showing an example in which a monaural echo canceller is applied to a conventional pseudo stereo loudspeaker.

FIG. 8 is a diagram showing an example in which a linear combination type echo canceller is applied to a conventional pseudo stereo loudspeaker.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Reception detection means, 2 ... Connection control means, 3 ... 1st signal path, 4 ... 2nd signal path, 5 ... Adaptive filter (AD
F), 6 ... Non-adaptive filter (NADF), 7, 50 ... Filter coefficient accumulation control means, 8 ... Subtractor, 9 ... Reception filter, 31-35 ... Memory.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04M 3/56 G10K 11/16 H

Claims (5)

[Claims] 1. A plurality of reception channels for receiving reception signals from a plurality of points, a reception detection unit for detecting reception signals received on the individual reception channels, and a reception signal detected by the reception detection unit. Connection control means for connecting the reception channel from which the reception signal is received to one of the first and second signal paths according to the number of receptions of the reception signal, and the sound image of the reception voice is localized at a different position for each voice. A plurality of speakers for controlling the output of the plurality of speakers with respect to each received signal input through the first and second signal paths, and a sound image localization means for localizing a sound image at a position corresponding to each received signal. A microphone for collecting the voice of the speaker; and a microphone connected to the first signal path from the reception channel to the first signal path, the sound image localization means, and a speaker. An adaptive filter for estimating an echo path sneaking to the transmission channel side through a speaker and a microphone and generating a pseudo echo; an echo path information accumulating means for accumulating a filter coefficient corresponding to the echo path estimated by the adaptive filter; A filter coefficient, which is connected to a signal path and corresponds to an echo path sneaking from the reception channel to the transmission channel side through the second signal path, the sound image localization means, the speaker and the microphone, is loaded from the echo path information storage means and the second A non-adaptive filter for estimating an echo sneaking to the transmission channel side through a signal path and generating a pseudo echo, and a subtractor for subtracting each pseudo echo generated by the non-adaptive filter and the adaptive filter from the echo signal of the transmission channel. Loud communication characterized by Apparatus. 2. The loudspeaker apparatus according to claim 1, wherein the sound image localization means localizes a sound image at a corresponding position with respect to a received signal input through the first signal path. A first signal calculation unit for performing a function calculation, a second signal calculation unit for performing a function calculation for each speaker for localizing a sound image at a corresponding position with respect to the reception signal input through the second signal path, A loudspeaker communication device comprising: a plurality of adders that add the calculation results for the same speaker calculated by the first and second signal calculation units. 3. The loudspeaker communication apparatus according to claim 1, wherein the echo path information storage means includes a plurality of memories for storing acoustic characteristic information corresponding to each sound image localization position as a filter coefficient. Loudspeaker. 4. The loudspeaker communication apparatus according to claim 1, wherein the echo path information accumulating unit includes acoustic characteristics of the sound image localization unit in a unit in which the transfer function estimated by the adaptive filter is a unit. First storage means for storing a plurality of echo path information, and a plurality of past echo path information including acoustic characteristics of the sound image localization means stored by the first storage means, excluding acoustic characteristics of the sound image localization means Means for obtaining echo path information, second storage means provided for the number of echo paths from the speaker to the microphone, and storing the echo path information excluding the obtained acoustic characteristics of the sound image localization means, and the second storage means. And a means for obtaining echo path information including the following acoustic characteristics of the sound image localization means based on the echo path information of the storage means. Hands-free communication device according to claim 1. 5. A reception detecting unit for detecting reception signals received on a plurality of reception channels, and a reception channel for receiving the reception signals according to the number of receptions of the reception signals detected by the reception detection unit. Connection control means connected to either one of the first signal path and the second signal path; and an echo path connected to the first signal path and sneaking from the reception channel to the transmission channel side through the first signal path, the speaker and the microphone. An adaptive filter for estimating a noise and generating a pseudo echo, an echo path information storage means for storing a filter coefficient corresponding to an echo path estimated by the adaptive filter, and a second signal connected to the second signal path. The filter coefficient corresponding to the echo path sneaking from the path to the transmission channel is loaded from the echo path information storage means, and A non-adaptive filter for estimating an echo path sneaking into the transmission channel through the two signal paths and generating a pseudo echo; and a subtractor for subtracting each pseudo echo generated by the non-adaptive filter and the adaptive filter from the echo signal of the transmission channel. An echo canceller characterized by being provided.