CA2124662C - Multi-channel echo cancelling method and a device thereof - Google Patents
Multi-channel echo cancelling method and a device thereofInfo
- Publication number
- CA2124662C CA2124662C CA002124662A CA2124662A CA2124662C CA 2124662 C CA2124662 C CA 2124662C CA 002124662 A CA002124662 A CA 002124662A CA 2124662 A CA2124662 A CA 2124662A CA 2124662 C CA2124662 C CA 2124662C
- Authority
- CA
- Canada
- Prior art keywords
- reception
- channel
- signal
- echo
- reception signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M9/00—Arrangements for interconnection not involving centralised switching
- H04M9/08—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic
- H04M9/082—Two-way loud-speaking telephone systems with means for conditioning the signal, e.g. for suppressing echoes for one or both directions of traffic using echo cancellers
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10K—SOUND-PRODUCING DEVICES; METHODS OR DEVICES FOR PROTECTING AGAINST, OR FOR DAMPING, NOISE OR OTHER ACOUSTIC WAVES IN GENERAL; ACOUSTICS NOT OTHERWISE PROVIDED FOR
- G10K2210/00—Details of active noise control [ANC] covered by G10K11/178 but not provided for in any of its subgroups
- G10K2210/10—Applications
- G10K2210/108—Communication systems, e.g. where useful sound is kept and noise is cancelled
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10K—SOUND-PRODUCING DEVICES; METHODS OR DEVICES FOR PROTECTING AGAINST, OR FOR DAMPING, NOISE OR OTHER ACOUSTIC WAVES IN GENERAL; ACOUSTICS NOT OTHERWISE PROVIDED FOR
- G10K2210/00—Details of active noise control [ANC] covered by G10K11/178 but not provided for in any of its subgroups
- G10K2210/30—Means
- G10K2210/301—Computational
- G10K2210/3012—Algorithms
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10K—SOUND-PRODUCING DEVICES; METHODS OR DEVICES FOR PROTECTING AGAINST, OR FOR DAMPING, NOISE OR OTHER ACOUSTIC WAVES IN GENERAL; ACOUSTICS NOT OTHERWISE PROVIDED FOR
- G10K2210/00—Details of active noise control [ANC] covered by G10K11/178 but not provided for in any of its subgroups
- G10K2210/50—Miscellaneous
- G10K2210/505—Echo cancellation, e.g. multipath-, ghost- or reverberation-cancellation
Abstract
The present invention relates to a multi-channel echo cancelling art enabling proper operation even if using a microphone not generating propagation delay difference between channels.
The present invention comprises: a step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power; a step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of the reception channels based on a reception signal of the reception channel selected in the previous step; and a step of cancelling the echoes from mixed signals in which the echoes are mixed by subtracting the echo replicas from the mixed signals.
The present invention comprises: a step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power; a step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of the reception channels based on a reception signal of the reception channel selected in the previous step; and a step of cancelling the echoes from mixed signals in which the echoes are mixed by subtracting the echo replicas from the mixed signals.
Description
2~24~62 MULTI-C~A~F.T, ECHO CANCELLING METHOD
AND A DEVICE THEREOF
BACKGROUND OF THE INVENTION
The present invention relates to a multi-channel echo cancelling method and a device thereof.
For a multi-channel echo cancelling method and a device thereof for cancelling an echo generated by propagation of reception signals in spatial acoustic paths, in a system having a plurality of reception signals and at least one transmission signal, a cascade connection type and a linear connection type are proposed in CS-89-178 (hereinafter referred to as Paper 1), pp.7-4, No.330, Vol.84 of the technical report of the Institute of Electronics and Communication Engineeirs.
The following problems, however, are pointed out in A5-3 (hereinafter referred to as Paper 2), pp.l44-149, of proceedings of the 6-th digital signal processing symposium, that is, the cascade connection type and the linear connection type have following defects:
(1) its hardware size becomes large because its hardware size is in proportion to square of the number of channels;
(2) convergence of adaptive filters becomes slow when reception signals have strong cross-correlation; and 2s ~3) coefficients of the adaptive filters may not converge .
212~662 ' ~ ~
to optimal values when reception siqnals have strong cross-correlation.
To solve these problems, a multi-channel echo canceller comprising an adaptive filter for each channel S is proposed in A-202 (hereinafter referred to as Paper 3), pp.202, volume 1 of proceedings of the 1991 Institute of Electronics, Iinformation and Communication Engineers spring conference.
Next, according to Paper 3, the case of applying this multi-channel echo canceller to a teleconference system, in which both of the reception signals and the transmission signals have two channels respectively, is explained.
Fig.10 shows a block diagram of voice section of a two-channel teleconference system connecting two teleconference rooms 20 and 21.
An acoustic echo cancelling method in the first teleconference room 20 is explained here for an example.
Let us suppose that the second talker 9 is in the second teleconference room 21. The voice 10l of the second talker 9 passes through the spatial acoustic path is recorded by the third microphone 11l, and is supplied to the second echo canceller 1202.
The voice recorded by the third microphone 111 is sent to the first teleconference room 20 as the first 212~1662 reception si.gnal 1l. Similarly to this, the voice 102 of the second talker 9 passes through the spatial acoustic path and is recorded by the fourth microphone 112, and is supplied to the second echo canceller 1202. The voice recorded by the fourth microphone 112 iS sent to the first teleconference room 20 as the second recepti.on signal 12.
In the first teleconference room 20, the first reception signal 11 is played back by the first 0 loudspeaker 2-, passes through the spatial acoustic path and comes to the first microphone 3, and the first echo 7, is generated at this time. I'he second reception signal 12 is played back by the second loudspeaker 22~ passes through the spatial acoustic path and comes to the first microphone 3, and the second echo 72 is generated at this time. The echoes 71, 72 and the first transmission signal 81 that is the voice of the first talker 6 come to the first microphone 3l are added and become the first mixed signal 41.
Similarly to this, the first reception signal 11 is played back by the first loudspeaker 2" passes through the spatial acoustic path and comes to the second microphone 32 and the third echo 73 is generated at this .. .
time. The second reception signal 12 iS played back by the second loudspeaker 22~ passes through the spatial 21246~2 acoustic path and comes to the second microphone 32 and the fourth echo 74 is generated at this time. The echoes 73~ 74 and the second transmission signal 82 that is the voice of the talker 6 come to the second microphone 32 are added and become the second mixed signal 42.
The first echo canceller 120l is used to remove echoes 7" 72~ 73 and 74 contained in the first and second mixed signals 41 and 42.
The propagation delay difference estimation circuit 110 receives the first reception signal 11 and the second reception signal 12~ estimates a propagation delay difference that is a difference between propagation delays of voices 101 and 102 of the second talker 9 until reaching to the first teleconference room 20 from the second teleconference room 21, and supplies the result to the reception signal selection circuit 111.
The reception signal selection circuit 111, based on :
an estimation result of said propagation delay dif~erence estimation circuit 110, detects a reception signal having shorter propagation delay from two .
reception signals 11 and 12. And it supplies the detection result as a designated reception signal selection information to the selector 105.
The selector 105 receives the first and the second 2s reception signals 11 and 12~ selects the reception signal ~ ~:
:, . ; ::: , :, :.. ,~ ~ . .
' 21246~2 106 that is designated by said reception signal selection information from the two reception signals 1-and 12. And it supplies the result to the first adaptive filter 112l and the second adaptive filter 1122.
The first adaptive filter 1121 receives the reception signal 106 selected by the selector 105, generates the echo replica 1081 corresponding to an echo contained in the first mixed signal 41, and supplies it to the first subtracter 1091.
0 The first s~btracter 109- subtracts the echo replica 108, that is an output of the first adaptive filter 1121 from the first mixed signal 4- and lets the result be the first output signal 5~.
The first adaptive filter 1121 is controlled so as to make the first output signal 51 be minimized.
Similarly to this, the second adaptive filter 1122 receives the reception signal 106 selected by the selector 105, generates the echo replica 1082 corresponding to an echo contained in the second mixed signal 42, and supplies a result to the second subtracter 1092.
The second subtracter 1092 subtracts the echo replica 1082 that is an output of the second adaptive filter 1122 from the second mixed s.ignal 92 and lets the result be 2s the second output signal 52.
2~2~6~2 Here, the second adaptive filte~ 1122 iS controlled so as to make the second output signal 52 be minimized.
A conventional multi-channel echo canceller supposes that the propagation delay between the first reception S signal 11 and the second reception signal 12 changes according to location of the second talker 9 in the second teleconference room 21.
However, there may be a microphone that scarcely generates propagation delay difference although 0 generates power difference according to location of the second talker 9. When such a microphone is used, a conventional multi-channel echo canceler can not select an appropriate reception signal, so that it can not work properly. This has been a problem.
SUM ~ RY OF TFlE lNvl~:h~loN
An object of the present invention is to offer multi-channel echo cancelling methods and devices thereof enabling proper operation even if using a microphone not generating propagation delay difference between channels.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling method comprising: a step of calculating a power of reception signal in each reception channel and selecting a recep~ion channel of a reception signal having the biggest power; a step of generating an echo replica ., . . . .. , '.~. ;' . ' . :! ' - ' .. ..
2~24662 corresponding ~o an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and a step of cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that it is preferable that said step of calculating a power of reception signal in each 0 recep~ion channel and selecting a reception channel of a reception signal having the biggest power calculates a square value of reception signal in each reception channel and selects a reception channel having the biggest square value, or calculates an instantaneous power of reception signal in each reception channel and selects a reception channel having the biggest instantaneous power.
Moreover, it is preferable that said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power calculates a mean value of powers of reception signals in each reception channel and selects a reception channel having the biggest mean value of powers.
Moreover, it is good that said step of calculating a ,, .. .. ,, . ~ ., ,, . . ~
21~662 power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power calculates powers of reception :
signals in each reception channel, sorts the results and selects a reception channel having the biggest median value of powers, or calculates a mode of powers of reception signals in each reception channel and selects a reception channel having the biggest mode of powers. :
Moreover it i<; preferable thAt said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signa:Ls and generates said echo replicas using said adapt:ive filters.
~urther, the objeet mentioned above of the present invention is achieved by a multi-channel echo cancelling method comprising: a step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value; a step of generating an echo replica corresponding to an echo in each mixed 21~6~2 g . .
signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and a step of cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that said step of calculating an absolute value of reception signal in each reception 0 channel and selecting a reception channel of a reception signal having the biggest absolute value calculates a mean value of absolute values of reception signals in each reception channel and selects a reception channel having the biggest mean value of absolute values.
Moreover it is preferable that said step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value calculates absolute values of reception signals in each reception channel, sorts the results and selects a reception channel having the biggest median value of absolute values, or calculates a mode of absolute values of reception signals in each reception channel and selects a reception channel having the biggest mode of absolute values.
Moreover, it is to be noted that said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signals and generates said echo replicas o using said adaptive filters.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling device comprising: a reception signal selection means for calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power; a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel; and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that it is preferable that said 21~662 means for calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power comprises: a plurality of power calculation means, one-S to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel; a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation 0 result from said power calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
Moreover, it is preferable that said power calculation means comprises a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result.
Moreover, it is preferable that said power calculation means comprises: a multiplication means for receiving a reception signal of said reception channel :: :
and squaring this reception signal and outputting a result; and a mean calculation means for storing square values calculated by said multiplication means and ~, 2124~62 calculating a mean power of reception signal from said square values.
Moreover, it is good that said power calculation means comprises: a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result; a register means for storing square values calculated by said multiplication means; and a sorting means for sorting square values stored in said register means and outputting an median value of the sorted square values, or also comprises: a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means; a histogram means for calculating a histogram from square values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
Moreover, it is preferable that said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels, based on said reception signal of the reception channel selected by said reception signal selection means, comprises a 212~2 plurality of adaptive filters one-to-one corresponding to said mixed signals.
Moreover, it is preferable that said adaptive filter is an adaptive transversal filter based on the LMS
algorism or an adaptive transversal filter based on the learning identification method.
Moreover, it is preferable that said means for cancelling said echo from said mixed signal by sub~racting said echo replica from said mixed signal for o each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
The object mentioned above of the present invention ~ :
is achieved by a multi-channel echo cancelling device comprising: a means for calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value; a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel; and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that said means for calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value comprises: a plurality of absolute value calculation means, one-to-one corresponding to each reception signal, for calculating an absolute value of reception signal of 0 said reception channel; a reception channel selection means for selecting a reception channel having the biggest absolute value of reception signal based on the calculation result from said absolute value calculation means; and a selection means for receiving reception lS signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
Moreover, it is preferable that an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel; and a mean calculation means for storing absolute values calculated by said absolute value calculation means and calculating a mean value of absolute values of a reception signal from said absolute ~ . - :.; - - -~
~12~62 values.
Moreover it is good that said absolute value calculation means comprises: an absolute value calculation means for calculating an absolute value of a S reception signal of said reception channel for each reception channel; a register means for storing absolute values calculated by said absolute value calculation means; and a sorting means for sorting absolute values stored in said register means and outputting an median : ' 0 value of the sorted square values, or also comprises: an absolute value calculation means for calculating an absolute value of a reception signal of said reception , '.
channel for each reception channeli a register means for storing absolute values calculated by said absolute value calculation means; a histogram means for calculating a histogram from absolute values stored in said register means; and a maximum value detection means for outputting a representative value of a ~one having maxirnum value in said histograrn.
Moreover, it is good that said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception 2s ,signal selection means comprises a plurality of adaptive 212~i662 filters one-to-one corresponding to said mixed signals.
Moreover, it is preferable that said adaptive filter is an adaptive transversal filter based on the LMS
algorism or an adaptive transversal filter based on the learning identification method.
Moreover, it is preferable that said means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling device lS comprising: a plurality of power calculation means, one-to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel; a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation result from said power calculation means; a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said 2s reception channel selection meansi a plurality of 212466~
adaptive filters, one-to-one corresponding to mixed signals in which echoes generated by propagation of reception signals of said reception channels are mixed, for generating echo replicas corresponding to said echoes based on a reception signal supplied by said selection means; and a plurality of subtraction means, one-to-one corresponding to said mixed signals, for outputting a result by subtracting said echo replicas from said mixed signals.
Moreover, it is good that said power calculation means is substituted by said absolute value calculation means for calculating an absolute value of a reception signal of said reception channel.
The multi-channel echo cancelling method and the device thereof of the present invention can properly operate even if using a microphone not generating a propagation delay difference between channels, because the method and device can select a reception signal, based on a power difference between reception signals.
BRIEF DESCRIPTION ON DR~WINGS
Fig.1 is a block diagram of the first embodiment of the present invention.
Fig.2 is a block diagram of the first embodiment of a power calculation circuit.
Fig.3 is a block diagram of a transversal adaptive 21~62 filter based on the LMS algorism.
Fig.9 is a block diagram of a transversal adaptive filter based on the learning identification method.
Fig.5 is a block diagram of an operation circuit.
Fig.6 is a block diagram of the second embodiment of a power calculation circuit.
Fig.7 is a block diagram of the third embodiment of a power calculation circuit.
Fig.8 is a block diagram of the fourth embodiment of a power calculation circuit.
Fig.9 is a block diagram of the second embodiment of the present invention.
Fig.10 is a block diagram of a conventional multi-channel echo canceller.
DETAILED DESCRIPTION OF THE INVENTION
The embodiment of the present invention is explained in detail, taking an acoustic echo cancelling method for cancelling acoustic echo recorded with microphones after propagatlon of reception signals through spatial acoustic paths from loudspeakers.
Fig.l shows the first embodiment of the present invention that reception signal and transmission signal have M channels, respectively.
~ ~his- rigure, 11, ..., 1M are reception signals of 2s M channels.
21246~2 - 1 9 - ~ .
2" ..., 2M are M units of loudspeakers that outputs corresponding reception signals 11, .. , 1M. ~.
31~ 3M are M units of microphones that records voice of the talker 6.
41~ .. , 4M are mixed signals of M channels. These mixed signals 41, ..., 4M are generated by adding echos and the voice of the talker 6 to be recorded by the microphones 31~ 3M. Where, these echos are generated at the time when the played back reception signals 11, ... , 1~ by the loudspeakers 21~ 2M have come to the microphones 31, ..., 3M through spatial acoustic paths.
100 is a multi-channel echo canceller.
1O11~ 01M are M units of power calculation circuits one-to-one corresponding to reception signals 15 1II . . . ~ 1M~ calculate corresponding reception signal's powers 1O2~ 102M. Now, as the power calculation circuits 101-, ..., 101M have same configuration and operate similarly, only the arbitrary i-th (1 ~ i ~ M) power calculation circuit 1011 is explained.
Fig.2 shows a block diagram of the power càlculation circuit 101~.
This power calculation circuit 1011 is configured so aS to calculate an instantaneous power of the reception signal 11, square the reception signal 1I supplied to the power calculation circuit 1011 by the multiplier 203 and output the result as the power 1021 of the reception signal 1-.
103 is a selection circuit that detects a reception signal having the bi.gqest power based on the reception signal powers 102-, .. , 102M calculated by the power calculation circuits 101l, ..., 101M, outputs the selection information 104.
105 is a selector that receives the M channels' reception s~gnals 1l, ..., 1M, selects and outputs the reception signal 106 having the biggest power based on the selection information 104.
107~, . . . ~ 107M are M units of adaptive filters one-to-one corresponding to the mixed signals 41, . . . ~ 4M~
receive the reception signal 106 having the biggest power and calculate echo replicas 1081~ .. , 108M, respectively.
109~ 109M are M units of subtracters one-to-one corresponding to the mixed signals 4l, ..., 4M~ subtract the echo replicas 1081I ..., 108M from the mixed signals 4~, .. , 4M~ output the results as the output signals 51~ ' . . . ~ 5M~ respectively.
Now, as the adaptive filters 107l~ ..., 107M and the subtracters 1091, ..., 109M have same configuration and operate similarly, only the arbitrary i-th (1 < i ~ M) adaptive filter 107l and the subtracter 109l is explained : . , in detail.
First, the embodiment of the adaptive filter 1071 is explained.
For an exa~ple of an adaptive filter, an adaptive transversal filter based on the LMS algorism is written in the Proceedings of the IEEE, Vol.63, No.12, pp.1692-1716, 1975, USA (hereinafter referred to as Paper 4).
And an adaptive transversal filter based on learning iden~ification method is written in the IEEE
0 transactions on automatic control, Vol. AC-12, No.3, 1967, USA (hereinafter referred to as Paper 5). Using adaptive transversal filter based on the LMS algorism, an operation and an embodiment of the adaptive filter are explained, here.
First, an adaptive transversal filter based on the LMS algorism is explained.
Supposing that the number of taps of the adaptive filter is N, an input signal to the adaptive filter at the time n is x(n), an output signal of the adaptive filter is y(n), an error signal is e(n) and the i-th filter coefficient is w(i,n) (i =0, ..., N-l), an output signal is represented by the equation (1).
N-l y(t~ , W(i, ll)X(rl - i) i~o (1). ' Also, updating of the filter coefficient is given by ; ~
212~662 the following equation.
w(i,n+1)=w(i,n)+~e(n)x(n-i) (2) Above-mentioned processing can be realized by cascade-connecting operation circuits calculating operation for 1 tap.
Next, the j-th operation circuit (j = 1, ..., N) receives the delay element input x~-l(n), the adder input y~--(n) and the filter coefficient update value ~(n), then carries out delay processing represented by the o equation (3).
x~(n)=xJ-.(n-1) (3) Also, it carries out convolution operation .
represented by the equation (4).
x~(n)=y~-~(n)+w(j-l,n)x~-l(n) t4) Further, it carries out coefficient update represented by the equation (5).
w(j-l,n+1)=w(j-l,n)+~(n)x(n-j+1) (5) And it outputs the delay element input x~(n) and the adder output y~(n).
Where, xO(n) and yO(n) are glven as follows.
xO(n)=x(n) (6) yO(n)=0 (7) And the filter coefficient update value ~(n) is defined as below.
~(n)=~e(n) (8) 212~6fi2 Next, the concrete configuration of the adaptive filter 1071 applied by an adaptive transversal filter based on the above-mentioned LMS algorism is explained.
Fig.3 shows a block diagram of the adaptive filter 1071.
This adaptive filter 1071 receives the reception signal 106 and calculates the echo replica 1081. Also, it updates a filter coefficien~ so as to make the error signal 3031 co~responding output signal 51 small.
0 The input signal 106 is supplied to the first operation circuit 304~ as a delay element input.
The first operation circuit 3041 receives the reception signal 106 as a delay element input, also receives the constant 0 that is stored in the first constant register 305 as an adder input, performs delay processing, convolution operation, coefficient updating.
Then, it supplies the delay element output to the delay element input of the second operation circuit 3042 and supplies the adder output to the adder input of the second operation circuit 3042. ~ -The second operation circuit 3042 receives the delay element output of the first operation circuit as a delay element input, also receives the adder output of the first operation circuit as an adder input, performs 2s delay processing, convolution operation, coefficient 21246~2 updating. And, it supplies the delay element output to the delay element input of the third operation circuit 3043 and supplies the adder output to the adder input of the third operation circuit 3043.
Similarly to the second operation circuit 3042, the j-th operation circuit 304~ ~J = 3, ..., N-l) receives the delay element output of the j-l-th operation circuit as a delay element input, also receives the adder output of the j-l-th operation circuit as an adder input, 0 performs delay processing, convolution operation, coefficient updating. And, it supplies the delay element output to the delay element input of the j+1-th operation circuit 304~-, also supplies the adder output to the adder input of the j+l-th operation circuit 304 Similarly, the N-th operation circuit 304N receives the delay element output of the N-l-th operation circuit as a delay element input, also receives the adder output of the N-l-th operation circuit as an adder input, performs delay processing, convolution operation, coefficient updating.
Then, the adder output becomes an output signal of the adaptive filter 1071, that is, the echo replica 1081.
It is to be noted that the delay element output of the N-th operation circuit 304N is not used. Moreover, 2s the multiplier 307 multiplies the error signal 3031 that is an output of the subtracter 109l by the content of the constant register 306, supplies the resulted product to the N units of the operation circuits 304" ..., 304N.
Further, if the above-mentioned filter coefficient S update value ~(n) is the equation (9), it becomes the learning identification method proposed in Paper 5. -~
~(n) = N-l ~e(n) ~,x2(n-i) (9) i~o Fig.4 shows a block diagram of an adaptive filter ~ that is applied by this leaning identification method.
About this adaptive filter, the above-mentioned power calculation circuit 308 and the divider 309 that receives an output of this power calculation circuit 308 lS and an output of the subtracter 307 are added to said :
adaptive filter 107J.
Next, the concrete configuration of the operation clrcuit 304l, ..., 304N is explained.
Fig.5 ls a diagram showing the configuration of the operation circuit 304 of the adaptive filter 1071.
This operation circuit 304 receives the delay element input 321, the adder input 322 and the filter coefficient update value 323, performs delay processing, convolution operation and coefficient updating, and outputs the delay element output 324 and the adder :~
- 21~662 output 325 . : ~ .
First, the delay element input 321 is supplied to the delay element 326 and the first multiplier 327. The delay element 326 outputs a delayed version of the delay ~.
element input 321 as the delay element output 324. ~ ~;
The first multiplier 327 multiplies the delay element input 321 by the content of the coefficient register 331, and supplies the resulted product to the first adder 328.
The first adder 328 sums up the resulted product of the first multiplier 327 and the adder input 322, and outputs the resulted sum as the adder output 325.
The second multiplier 329 multiplies the delay element input 321 by the filter coefficient update value 323, supplies the resulted product to the second adder 330.
The second adder 330 sums up the resulted product of the second multiplier 329 and the content of the coefficient register 331, and outputs the resulted sum to the coefficient register 331.
The coefficient register 331 stores the resulted sum from the second adder 330.
Next, the adaptive filter 1071 configured as above receives the reception signal 106 having the biggest 2s power, calculates the echo replica 1081. The calculated :
21246S~
echo replica is supplied to the subtracter 109,.
The subtracter 109l subtracts the echo replica 108-from the mixed signal 41, outputs the resulted output signal 5l and simultaneously outputs again a part thereof S to the adaptive filter 1071 as the error signal 303l.
By configuring as above, acoustic echoes can be cancelled properly if a microphone not generating propagation delay difference between channels is used.
Next, the second embodiment is explained.
The second embodiment is the same as the first embodiment except for substituting the power calculation circuit in the first embodiment by a power average calculation circuit for calculating a mean power.
It is to be noted that only the arbitrary power lS calculation circuit 500l is explained briefly because the other parts are the same as those of the first embodiment.
Fig.6 shows a block diagram of a power average calculation circult.
This power calculation circuit 500l calculates the mean power of the reception signal 1J. The multiplier 501 squares the reception signal 1l and supplies the result to the average calculation circuit 502.
The average calculation circuit 502 calculates a time average of the calculated products and outputs the 2124662 : ~
result as the mean power 503.
By configuring as above, it is possible to select a reception channel of a reception signal whose power is biggest.
Next, the third embodiment is explained.
The third embodiment is the same as the first embodiment except for substltuting the power calculation circuit in the first embodiment by a power calculation circuit for calculating a medianvalue of powers.
It is to be noted that only the arbitrary power calculation circuit 6001 is explained briefly because the other parts are the same as those of the first embodiment.
Fig.7 shows a block diagram of a power median calculation circuit.
First, the multiplier 601 calculates a power by squaring a reception signal and outputs the result.
The registers 6021 to 602k store k powers in turn.
The sorting Circuit 603 sorts out the stored k powers and the lastly calculated power in turn from large to small or from small to large and outputs the intermediate value of power that is the value of the (k/2)+1-th power as a median value of powers 604. - -~
Note that the sorting circuit 603 outputs the ~k+1)/2-th or ~k+3)/2-th value if k is an odd number.
--' 212~6fi2 ;~ ::
Continuously, the fourth embodiment is explained.
The fourth embodiment is the same as the first embodiment except for substituting the power calculation circuit in the first embodiment by a mode of power calculation circuit for calculating a mode of power.
It is to be noted that only the arbitrary power calculation circuit 7ûOI is explained briefly because the other parts are the same as those of the first embodiment.
Fig.8 shows a block diagram of a mode of power calculation circuit.
First, the multiplier 701 calculates a power by squaring a reception signal and outputs the result. '~
The registers 702, to 702~ stores k powers in turn.
The histogram calculation circuit 703 calculates a histogram based on stored k powers and the lastly calculated a power in turn and outputs the resulted histogram signals 7041, ..., 704~.
The maximum value detection circuit 705 outputs a central value of a zone having maximum value in the histogram as an mode of power 706 based on the histogram signals 704l, ..., 704~.
Ne~t, the fifth embodiment is explained.
Fig.9 shows the fifth embodiment of the present invention in case that the reception signal and the ;~
~124~62 transmission signal are of M-channel, respectively.
121 is the multi-channel echo cancelling device.
The fifth embodiment is the same as the first embodiment except for substituting the power calculation circuits 1011, .. , 101M in the first embodiment by the absolute value calculation circuits 8011~ ..., 8O1M~
The absolute value calculation circuits 8011, ....
801M calculates and outputs an absolute value of the reception signal.
It is to be noted in this case that it is good to use an average of absolute value, median of absolute value and mode of the absolute value, similarly to the above-mentioned second embodiment, the third embodiment and the fourth embodiment. Moreover, for this case the absolute value calculation circuit 801 can be used instead of the multiplier 501, the multiplier 601 cr the multiplier 701.
~: , ...
~,, ,, ",:~
~.': " ''. '~
AND A DEVICE THEREOF
BACKGROUND OF THE INVENTION
The present invention relates to a multi-channel echo cancelling method and a device thereof.
For a multi-channel echo cancelling method and a device thereof for cancelling an echo generated by propagation of reception signals in spatial acoustic paths, in a system having a plurality of reception signals and at least one transmission signal, a cascade connection type and a linear connection type are proposed in CS-89-178 (hereinafter referred to as Paper 1), pp.7-4, No.330, Vol.84 of the technical report of the Institute of Electronics and Communication Engineeirs.
The following problems, however, are pointed out in A5-3 (hereinafter referred to as Paper 2), pp.l44-149, of proceedings of the 6-th digital signal processing symposium, that is, the cascade connection type and the linear connection type have following defects:
(1) its hardware size becomes large because its hardware size is in proportion to square of the number of channels;
(2) convergence of adaptive filters becomes slow when reception signals have strong cross-correlation; and 2s ~3) coefficients of the adaptive filters may not converge .
212~662 ' ~ ~
to optimal values when reception siqnals have strong cross-correlation.
To solve these problems, a multi-channel echo canceller comprising an adaptive filter for each channel S is proposed in A-202 (hereinafter referred to as Paper 3), pp.202, volume 1 of proceedings of the 1991 Institute of Electronics, Iinformation and Communication Engineers spring conference.
Next, according to Paper 3, the case of applying this multi-channel echo canceller to a teleconference system, in which both of the reception signals and the transmission signals have two channels respectively, is explained.
Fig.10 shows a block diagram of voice section of a two-channel teleconference system connecting two teleconference rooms 20 and 21.
An acoustic echo cancelling method in the first teleconference room 20 is explained here for an example.
Let us suppose that the second talker 9 is in the second teleconference room 21. The voice 10l of the second talker 9 passes through the spatial acoustic path is recorded by the third microphone 11l, and is supplied to the second echo canceller 1202.
The voice recorded by the third microphone 111 is sent to the first teleconference room 20 as the first 212~1662 reception si.gnal 1l. Similarly to this, the voice 102 of the second talker 9 passes through the spatial acoustic path and is recorded by the fourth microphone 112, and is supplied to the second echo canceller 1202. The voice recorded by the fourth microphone 112 iS sent to the first teleconference room 20 as the second recepti.on signal 12.
In the first teleconference room 20, the first reception signal 11 is played back by the first 0 loudspeaker 2-, passes through the spatial acoustic path and comes to the first microphone 3, and the first echo 7, is generated at this time. I'he second reception signal 12 is played back by the second loudspeaker 22~ passes through the spatial acoustic path and comes to the first microphone 3, and the second echo 72 is generated at this time. The echoes 71, 72 and the first transmission signal 81 that is the voice of the first talker 6 come to the first microphone 3l are added and become the first mixed signal 41.
Similarly to this, the first reception signal 11 is played back by the first loudspeaker 2" passes through the spatial acoustic path and comes to the second microphone 32 and the third echo 73 is generated at this .. .
time. The second reception signal 12 iS played back by the second loudspeaker 22~ passes through the spatial 21246~2 acoustic path and comes to the second microphone 32 and the fourth echo 74 is generated at this time. The echoes 73~ 74 and the second transmission signal 82 that is the voice of the talker 6 come to the second microphone 32 are added and become the second mixed signal 42.
The first echo canceller 120l is used to remove echoes 7" 72~ 73 and 74 contained in the first and second mixed signals 41 and 42.
The propagation delay difference estimation circuit 110 receives the first reception signal 11 and the second reception signal 12~ estimates a propagation delay difference that is a difference between propagation delays of voices 101 and 102 of the second talker 9 until reaching to the first teleconference room 20 from the second teleconference room 21, and supplies the result to the reception signal selection circuit 111.
The reception signal selection circuit 111, based on :
an estimation result of said propagation delay dif~erence estimation circuit 110, detects a reception signal having shorter propagation delay from two .
reception signals 11 and 12. And it supplies the detection result as a designated reception signal selection information to the selector 105.
The selector 105 receives the first and the second 2s reception signals 11 and 12~ selects the reception signal ~ ~:
:, . ; ::: , :, :.. ,~ ~ . .
' 21246~2 106 that is designated by said reception signal selection information from the two reception signals 1-and 12. And it supplies the result to the first adaptive filter 112l and the second adaptive filter 1122.
The first adaptive filter 1121 receives the reception signal 106 selected by the selector 105, generates the echo replica 1081 corresponding to an echo contained in the first mixed signal 41, and supplies it to the first subtracter 1091.
0 The first s~btracter 109- subtracts the echo replica 108, that is an output of the first adaptive filter 1121 from the first mixed signal 4- and lets the result be the first output signal 5~.
The first adaptive filter 1121 is controlled so as to make the first output signal 51 be minimized.
Similarly to this, the second adaptive filter 1122 receives the reception signal 106 selected by the selector 105, generates the echo replica 1082 corresponding to an echo contained in the second mixed signal 42, and supplies a result to the second subtracter 1092.
The second subtracter 1092 subtracts the echo replica 1082 that is an output of the second adaptive filter 1122 from the second mixed s.ignal 92 and lets the result be 2s the second output signal 52.
2~2~6~2 Here, the second adaptive filte~ 1122 iS controlled so as to make the second output signal 52 be minimized.
A conventional multi-channel echo canceller supposes that the propagation delay between the first reception S signal 11 and the second reception signal 12 changes according to location of the second talker 9 in the second teleconference room 21.
However, there may be a microphone that scarcely generates propagation delay difference although 0 generates power difference according to location of the second talker 9. When such a microphone is used, a conventional multi-channel echo canceler can not select an appropriate reception signal, so that it can not work properly. This has been a problem.
SUM ~ RY OF TFlE lNvl~:h~loN
An object of the present invention is to offer multi-channel echo cancelling methods and devices thereof enabling proper operation even if using a microphone not generating propagation delay difference between channels.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling method comprising: a step of calculating a power of reception signal in each reception channel and selecting a recep~ion channel of a reception signal having the biggest power; a step of generating an echo replica ., . . . .. , '.~. ;' . ' . :! ' - ' .. ..
2~24662 corresponding ~o an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and a step of cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that it is preferable that said step of calculating a power of reception signal in each 0 recep~ion channel and selecting a reception channel of a reception signal having the biggest power calculates a square value of reception signal in each reception channel and selects a reception channel having the biggest square value, or calculates an instantaneous power of reception signal in each reception channel and selects a reception channel having the biggest instantaneous power.
Moreover, it is preferable that said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power calculates a mean value of powers of reception signals in each reception channel and selects a reception channel having the biggest mean value of powers.
Moreover, it is good that said step of calculating a ,, .. .. ,, . ~ ., ,, . . ~
21~662 power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power calculates powers of reception :
signals in each reception channel, sorts the results and selects a reception channel having the biggest median value of powers, or calculates a mode of powers of reception signals in each reception channel and selects a reception channel having the biggest mode of powers. :
Moreover it i<; preferable thAt said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signa:Ls and generates said echo replicas using said adapt:ive filters.
~urther, the objeet mentioned above of the present invention is achieved by a multi-channel echo cancelling method comprising: a step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value; a step of generating an echo replica corresponding to an echo in each mixed 21~6~2 g . .
signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and a step of cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that said step of calculating an absolute value of reception signal in each reception 0 channel and selecting a reception channel of a reception signal having the biggest absolute value calculates a mean value of absolute values of reception signals in each reception channel and selects a reception channel having the biggest mean value of absolute values.
Moreover it is preferable that said step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value calculates absolute values of reception signals in each reception channel, sorts the results and selects a reception channel having the biggest median value of absolute values, or calculates a mode of absolute values of reception signals in each reception channel and selects a reception channel having the biggest mode of absolute values.
Moreover, it is to be noted that said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signals and generates said echo replicas o using said adaptive filters.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling device comprising: a reception signal selection means for calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power; a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel; and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that it is preferable that said 21~662 means for calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power comprises: a plurality of power calculation means, one-S to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel; a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation 0 result from said power calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
Moreover, it is preferable that said power calculation means comprises a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result.
Moreover, it is preferable that said power calculation means comprises: a multiplication means for receiving a reception signal of said reception channel :: :
and squaring this reception signal and outputting a result; and a mean calculation means for storing square values calculated by said multiplication means and ~, 2124~62 calculating a mean power of reception signal from said square values.
Moreover, it is good that said power calculation means comprises: a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result; a register means for storing square values calculated by said multiplication means; and a sorting means for sorting square values stored in said register means and outputting an median value of the sorted square values, or also comprises: a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means; a histogram means for calculating a histogram from square values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
Moreover, it is preferable that said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels, based on said reception signal of the reception channel selected by said reception signal selection means, comprises a 212~2 plurality of adaptive filters one-to-one corresponding to said mixed signals.
Moreover, it is preferable that said adaptive filter is an adaptive transversal filter based on the LMS
algorism or an adaptive transversal filter based on the learning identification method.
Moreover, it is preferable that said means for cancelling said echo from said mixed signal by sub~racting said echo replica from said mixed signal for o each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
The object mentioned above of the present invention ~ :
is achieved by a multi-channel echo cancelling device comprising: a means for calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value; a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel; and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
It is to be noted that said means for calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value comprises: a plurality of absolute value calculation means, one-to-one corresponding to each reception signal, for calculating an absolute value of reception signal of 0 said reception channel; a reception channel selection means for selecting a reception channel having the biggest absolute value of reception signal based on the calculation result from said absolute value calculation means; and a selection means for receiving reception lS signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
Moreover, it is preferable that an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel; and a mean calculation means for storing absolute values calculated by said absolute value calculation means and calculating a mean value of absolute values of a reception signal from said absolute ~ . - :.; - - -~
~12~62 values.
Moreover it is good that said absolute value calculation means comprises: an absolute value calculation means for calculating an absolute value of a S reception signal of said reception channel for each reception channel; a register means for storing absolute values calculated by said absolute value calculation means; and a sorting means for sorting absolute values stored in said register means and outputting an median : ' 0 value of the sorted square values, or also comprises: an absolute value calculation means for calculating an absolute value of a reception signal of said reception , '.
channel for each reception channeli a register means for storing absolute values calculated by said absolute value calculation means; a histogram means for calculating a histogram from absolute values stored in said register means; and a maximum value detection means for outputting a representative value of a ~one having maxirnum value in said histograrn.
Moreover, it is good that said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception 2s ,signal selection means comprises a plurality of adaptive 212~i662 filters one-to-one corresponding to said mixed signals.
Moreover, it is preferable that said adaptive filter is an adaptive transversal filter based on the LMS
algorism or an adaptive transversal filter based on the learning identification method.
Moreover, it is preferable that said means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
The object mentioned above of the present invention is achieved by a multi-channel echo cancelling device lS comprising: a plurality of power calculation means, one-to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel; a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation result from said power calculation means; a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said 2s reception channel selection meansi a plurality of 212466~
adaptive filters, one-to-one corresponding to mixed signals in which echoes generated by propagation of reception signals of said reception channels are mixed, for generating echo replicas corresponding to said echoes based on a reception signal supplied by said selection means; and a plurality of subtraction means, one-to-one corresponding to said mixed signals, for outputting a result by subtracting said echo replicas from said mixed signals.
Moreover, it is good that said power calculation means is substituted by said absolute value calculation means for calculating an absolute value of a reception signal of said reception channel.
The multi-channel echo cancelling method and the device thereof of the present invention can properly operate even if using a microphone not generating a propagation delay difference between channels, because the method and device can select a reception signal, based on a power difference between reception signals.
BRIEF DESCRIPTION ON DR~WINGS
Fig.1 is a block diagram of the first embodiment of the present invention.
Fig.2 is a block diagram of the first embodiment of a power calculation circuit.
Fig.3 is a block diagram of a transversal adaptive 21~62 filter based on the LMS algorism.
Fig.9 is a block diagram of a transversal adaptive filter based on the learning identification method.
Fig.5 is a block diagram of an operation circuit.
Fig.6 is a block diagram of the second embodiment of a power calculation circuit.
Fig.7 is a block diagram of the third embodiment of a power calculation circuit.
Fig.8 is a block diagram of the fourth embodiment of a power calculation circuit.
Fig.9 is a block diagram of the second embodiment of the present invention.
Fig.10 is a block diagram of a conventional multi-channel echo canceller.
DETAILED DESCRIPTION OF THE INVENTION
The embodiment of the present invention is explained in detail, taking an acoustic echo cancelling method for cancelling acoustic echo recorded with microphones after propagatlon of reception signals through spatial acoustic paths from loudspeakers.
Fig.l shows the first embodiment of the present invention that reception signal and transmission signal have M channels, respectively.
~ ~his- rigure, 11, ..., 1M are reception signals of 2s M channels.
21246~2 - 1 9 - ~ .
2" ..., 2M are M units of loudspeakers that outputs corresponding reception signals 11, .. , 1M. ~.
31~ 3M are M units of microphones that records voice of the talker 6.
41~ .. , 4M are mixed signals of M channels. These mixed signals 41, ..., 4M are generated by adding echos and the voice of the talker 6 to be recorded by the microphones 31~ 3M. Where, these echos are generated at the time when the played back reception signals 11, ... , 1~ by the loudspeakers 21~ 2M have come to the microphones 31, ..., 3M through spatial acoustic paths.
100 is a multi-channel echo canceller.
1O11~ 01M are M units of power calculation circuits one-to-one corresponding to reception signals 15 1II . . . ~ 1M~ calculate corresponding reception signal's powers 1O2~ 102M. Now, as the power calculation circuits 101-, ..., 101M have same configuration and operate similarly, only the arbitrary i-th (1 ~ i ~ M) power calculation circuit 1011 is explained.
Fig.2 shows a block diagram of the power càlculation circuit 101~.
This power calculation circuit 1011 is configured so aS to calculate an instantaneous power of the reception signal 11, square the reception signal 1I supplied to the power calculation circuit 1011 by the multiplier 203 and output the result as the power 1021 of the reception signal 1-.
103 is a selection circuit that detects a reception signal having the bi.gqest power based on the reception signal powers 102-, .. , 102M calculated by the power calculation circuits 101l, ..., 101M, outputs the selection information 104.
105 is a selector that receives the M channels' reception s~gnals 1l, ..., 1M, selects and outputs the reception signal 106 having the biggest power based on the selection information 104.
107~, . . . ~ 107M are M units of adaptive filters one-to-one corresponding to the mixed signals 41, . . . ~ 4M~
receive the reception signal 106 having the biggest power and calculate echo replicas 1081~ .. , 108M, respectively.
109~ 109M are M units of subtracters one-to-one corresponding to the mixed signals 4l, ..., 4M~ subtract the echo replicas 1081I ..., 108M from the mixed signals 4~, .. , 4M~ output the results as the output signals 51~ ' . . . ~ 5M~ respectively.
Now, as the adaptive filters 107l~ ..., 107M and the subtracters 1091, ..., 109M have same configuration and operate similarly, only the arbitrary i-th (1 < i ~ M) adaptive filter 107l and the subtracter 109l is explained : . , in detail.
First, the embodiment of the adaptive filter 1071 is explained.
For an exa~ple of an adaptive filter, an adaptive transversal filter based on the LMS algorism is written in the Proceedings of the IEEE, Vol.63, No.12, pp.1692-1716, 1975, USA (hereinafter referred to as Paper 4).
And an adaptive transversal filter based on learning iden~ification method is written in the IEEE
0 transactions on automatic control, Vol. AC-12, No.3, 1967, USA (hereinafter referred to as Paper 5). Using adaptive transversal filter based on the LMS algorism, an operation and an embodiment of the adaptive filter are explained, here.
First, an adaptive transversal filter based on the LMS algorism is explained.
Supposing that the number of taps of the adaptive filter is N, an input signal to the adaptive filter at the time n is x(n), an output signal of the adaptive filter is y(n), an error signal is e(n) and the i-th filter coefficient is w(i,n) (i =0, ..., N-l), an output signal is represented by the equation (1).
N-l y(t~ , W(i, ll)X(rl - i) i~o (1). ' Also, updating of the filter coefficient is given by ; ~
212~662 the following equation.
w(i,n+1)=w(i,n)+~e(n)x(n-i) (2) Above-mentioned processing can be realized by cascade-connecting operation circuits calculating operation for 1 tap.
Next, the j-th operation circuit (j = 1, ..., N) receives the delay element input x~-l(n), the adder input y~--(n) and the filter coefficient update value ~(n), then carries out delay processing represented by the o equation (3).
x~(n)=xJ-.(n-1) (3) Also, it carries out convolution operation .
represented by the equation (4).
x~(n)=y~-~(n)+w(j-l,n)x~-l(n) t4) Further, it carries out coefficient update represented by the equation (5).
w(j-l,n+1)=w(j-l,n)+~(n)x(n-j+1) (5) And it outputs the delay element input x~(n) and the adder output y~(n).
Where, xO(n) and yO(n) are glven as follows.
xO(n)=x(n) (6) yO(n)=0 (7) And the filter coefficient update value ~(n) is defined as below.
~(n)=~e(n) (8) 212~6fi2 Next, the concrete configuration of the adaptive filter 1071 applied by an adaptive transversal filter based on the above-mentioned LMS algorism is explained.
Fig.3 shows a block diagram of the adaptive filter 1071.
This adaptive filter 1071 receives the reception signal 106 and calculates the echo replica 1081. Also, it updates a filter coefficien~ so as to make the error signal 3031 co~responding output signal 51 small.
0 The input signal 106 is supplied to the first operation circuit 304~ as a delay element input.
The first operation circuit 3041 receives the reception signal 106 as a delay element input, also receives the constant 0 that is stored in the first constant register 305 as an adder input, performs delay processing, convolution operation, coefficient updating.
Then, it supplies the delay element output to the delay element input of the second operation circuit 3042 and supplies the adder output to the adder input of the second operation circuit 3042. ~ -The second operation circuit 3042 receives the delay element output of the first operation circuit as a delay element input, also receives the adder output of the first operation circuit as an adder input, performs 2s delay processing, convolution operation, coefficient 21246~2 updating. And, it supplies the delay element output to the delay element input of the third operation circuit 3043 and supplies the adder output to the adder input of the third operation circuit 3043.
Similarly to the second operation circuit 3042, the j-th operation circuit 304~ ~J = 3, ..., N-l) receives the delay element output of the j-l-th operation circuit as a delay element input, also receives the adder output of the j-l-th operation circuit as an adder input, 0 performs delay processing, convolution operation, coefficient updating. And, it supplies the delay element output to the delay element input of the j+1-th operation circuit 304~-, also supplies the adder output to the adder input of the j+l-th operation circuit 304 Similarly, the N-th operation circuit 304N receives the delay element output of the N-l-th operation circuit as a delay element input, also receives the adder output of the N-l-th operation circuit as an adder input, performs delay processing, convolution operation, coefficient updating.
Then, the adder output becomes an output signal of the adaptive filter 1071, that is, the echo replica 1081.
It is to be noted that the delay element output of the N-th operation circuit 304N is not used. Moreover, 2s the multiplier 307 multiplies the error signal 3031 that is an output of the subtracter 109l by the content of the constant register 306, supplies the resulted product to the N units of the operation circuits 304" ..., 304N.
Further, if the above-mentioned filter coefficient S update value ~(n) is the equation (9), it becomes the learning identification method proposed in Paper 5. -~
~(n) = N-l ~e(n) ~,x2(n-i) (9) i~o Fig.4 shows a block diagram of an adaptive filter ~ that is applied by this leaning identification method.
About this adaptive filter, the above-mentioned power calculation circuit 308 and the divider 309 that receives an output of this power calculation circuit 308 lS and an output of the subtracter 307 are added to said :
adaptive filter 107J.
Next, the concrete configuration of the operation clrcuit 304l, ..., 304N is explained.
Fig.5 ls a diagram showing the configuration of the operation circuit 304 of the adaptive filter 1071.
This operation circuit 304 receives the delay element input 321, the adder input 322 and the filter coefficient update value 323, performs delay processing, convolution operation and coefficient updating, and outputs the delay element output 324 and the adder :~
- 21~662 output 325 . : ~ .
First, the delay element input 321 is supplied to the delay element 326 and the first multiplier 327. The delay element 326 outputs a delayed version of the delay ~.
element input 321 as the delay element output 324. ~ ~;
The first multiplier 327 multiplies the delay element input 321 by the content of the coefficient register 331, and supplies the resulted product to the first adder 328.
The first adder 328 sums up the resulted product of the first multiplier 327 and the adder input 322, and outputs the resulted sum as the adder output 325.
The second multiplier 329 multiplies the delay element input 321 by the filter coefficient update value 323, supplies the resulted product to the second adder 330.
The second adder 330 sums up the resulted product of the second multiplier 329 and the content of the coefficient register 331, and outputs the resulted sum to the coefficient register 331.
The coefficient register 331 stores the resulted sum from the second adder 330.
Next, the adaptive filter 1071 configured as above receives the reception signal 106 having the biggest 2s power, calculates the echo replica 1081. The calculated :
21246S~
echo replica is supplied to the subtracter 109,.
The subtracter 109l subtracts the echo replica 108-from the mixed signal 41, outputs the resulted output signal 5l and simultaneously outputs again a part thereof S to the adaptive filter 1071 as the error signal 303l.
By configuring as above, acoustic echoes can be cancelled properly if a microphone not generating propagation delay difference between channels is used.
Next, the second embodiment is explained.
The second embodiment is the same as the first embodiment except for substituting the power calculation circuit in the first embodiment by a power average calculation circuit for calculating a mean power.
It is to be noted that only the arbitrary power lS calculation circuit 500l is explained briefly because the other parts are the same as those of the first embodiment.
Fig.6 shows a block diagram of a power average calculation circult.
This power calculation circuit 500l calculates the mean power of the reception signal 1J. The multiplier 501 squares the reception signal 1l and supplies the result to the average calculation circuit 502.
The average calculation circuit 502 calculates a time average of the calculated products and outputs the 2124662 : ~
result as the mean power 503.
By configuring as above, it is possible to select a reception channel of a reception signal whose power is biggest.
Next, the third embodiment is explained.
The third embodiment is the same as the first embodiment except for substltuting the power calculation circuit in the first embodiment by a power calculation circuit for calculating a medianvalue of powers.
It is to be noted that only the arbitrary power calculation circuit 6001 is explained briefly because the other parts are the same as those of the first embodiment.
Fig.7 shows a block diagram of a power median calculation circuit.
First, the multiplier 601 calculates a power by squaring a reception signal and outputs the result.
The registers 6021 to 602k store k powers in turn.
The sorting Circuit 603 sorts out the stored k powers and the lastly calculated power in turn from large to small or from small to large and outputs the intermediate value of power that is the value of the (k/2)+1-th power as a median value of powers 604. - -~
Note that the sorting circuit 603 outputs the ~k+1)/2-th or ~k+3)/2-th value if k is an odd number.
--' 212~6fi2 ;~ ::
Continuously, the fourth embodiment is explained.
The fourth embodiment is the same as the first embodiment except for substituting the power calculation circuit in the first embodiment by a mode of power calculation circuit for calculating a mode of power.
It is to be noted that only the arbitrary power calculation circuit 7ûOI is explained briefly because the other parts are the same as those of the first embodiment.
Fig.8 shows a block diagram of a mode of power calculation circuit.
First, the multiplier 701 calculates a power by squaring a reception signal and outputs the result. '~
The registers 702, to 702~ stores k powers in turn.
The histogram calculation circuit 703 calculates a histogram based on stored k powers and the lastly calculated a power in turn and outputs the resulted histogram signals 7041, ..., 704~.
The maximum value detection circuit 705 outputs a central value of a zone having maximum value in the histogram as an mode of power 706 based on the histogram signals 704l, ..., 704~.
Ne~t, the fifth embodiment is explained.
Fig.9 shows the fifth embodiment of the present invention in case that the reception signal and the ;~
~124~62 transmission signal are of M-channel, respectively.
121 is the multi-channel echo cancelling device.
The fifth embodiment is the same as the first embodiment except for substituting the power calculation circuits 1011, .. , 101M in the first embodiment by the absolute value calculation circuits 8011~ ..., 8O1M~
The absolute value calculation circuits 8011, ....
801M calculates and outputs an absolute value of the reception signal.
It is to be noted in this case that it is good to use an average of absolute value, median of absolute value and mode of the absolute value, similarly to the above-mentioned second embodiment, the third embodiment and the fourth embodiment. Moreover, for this case the absolute value calculation circuit 801 can be used instead of the multiplier 501, the multiplier 601 cr the multiplier 701.
~: , ...
~,, ,, ",:~
~.': " ''. '~
Claims (33)
1. A multi-channel echo cancelling method comprising the following steps:
calculating a power of reception signal in each of a plurality of reception channel and selecting the reception channel of the reception signal having the highest level of power;
generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
calculating a power of reception signal in each of a plurality of reception channel and selecting the reception channel of the reception signal having the highest level of power;
generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
2. The multi-channel echo cancelling method of claim 1, wherein said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the highest level of power calculates the square of the value of reception signal in each reception channel and selects a reception channel having the largest square value.
3. The multi-channel echo cancelling method of claim 1, wherein said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the highest level of power comprises calculating an instantaneous power of reception signal in each reception channel and selecting a reception channel having the highest level of instantaneous power.
4. The multichannel echo cancelling method of claim 1, wherein said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest power comprises:
calculating a mean value of power of reception signals in each reception channel and selecting a reception channel having the greatest mean value of power.
calculating a mean value of power of reception signals in each reception channel and selecting a reception channel having the greatest mean value of power.
5. The multi-channel echo cancelling method of claim 1, wherein said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the highest level of power comprises calculating the power of reception signals in each reception channel, and selecting a reception channel having the greatest median value of power.
6. The multi-channel echo cancelling method of claim 1, wherein said step of calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the highest level of power comprises: calculating a mode of powers of reception signals in each reception channel and selecting a reception channel having the biggest mode of powers.
7. The multi-channel echo cancelling method of claim 1, wherein said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signals and generates said echo replicas using said adaptive filters.
8. A multi-channel echo cancelling method comprising the following steps:
calculating an absolute value of reception signal in each of a plurality of reception channels and selecting the reception channel of the reception signal having the greatest absolute value;
generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
calculating an absolute value of reception signal in each of a plurality of reception channels and selecting the reception channel of the reception signal having the greatest absolute value;
generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel; and cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
9. The multi-channel echo cancelling method of claim 8, wherein said step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the greatest absolute value comprises: calculating a mean value of absolute values of reception signals in each reception channel and selecting a reception channel having the greatest mean value of absolute values.
10. The multi-channel echo cancelling method of claim 8, wherein said step of calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the greatest absolute value comprises: calculating absolute values of reception signals in each reception channel, sorting the results and selecting reception channel having the largest median value of absolute values.
11. The multi-channel echo cancelling method of claim 8, wherein said step of calculating an absolute value of a reception signal in each reception channel and selecting a reception channel of a reception signal having the biggest absolute value comprises: calculating a mode of absolute values of reception signals in each reception channel and selecting a reception channel having the biggest mode of absolute values.
12. The multi-channel echo cancelling method of claim 8, wherein said step of generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected in the previous step for each reception channel supplies said reception signal of the selected reception channel to adaptive filters that are one-to-one corresponding to said mixed signals and generates said echo replicas using said adaptive filters.
13. A multi-channel echo cancelling device comprising:
a reception signal selection means for calculating a power of reception signal in each of a plurality of reception channels and selecting the reception channel of a reception signal having the highest level of power;
a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel;
and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
a reception signal selection means for calculating a power of reception signal in each of a plurality of reception channels and selecting the reception channel of a reception signal having the highest level of power;
a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel;
and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
14. The multi-channel echo cancelling device of claim 13, wherein said means for calculating a power of reception signal in each reception channel and selecting a reception channel of a reception signal having the highest level of power comprises:
a plurality of power calculation means, one-to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the largest power of reception signal based on the calculation performed by said power calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
a plurality of power calculation means, one-to-one corresponding to each reception signal, for calculating a power of reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the largest power of reception signal based on the calculation performed by said power calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means.
15. The multi-channel echo cancelling device of claim 14, wherein said power calculation means comprises:
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result.
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result.
16. The multi-channel echo cancelling device of claim 14, wherein said power calculation means comprises a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result; and a mean calculation means for storing square values calculated by said multiplication means and calculating a mean power of reception signal from said square values.
17. The multi-channel echo cancelling device of claim 14, wherein said power calculation means comprises:
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means; and a sorting means for sorting square values stored in said register means and outputting a median value of the sorted square values.
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means; and a sorting means for sorting square values stored in said register means and outputting a median value of the sorted square values.
18. The multi-channel echo cancelling device of claim 14, wherein said power calculation means comprises:
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means;
a histogram means for calculating a histogram from square values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
a multiplication means for receiving a reception signal of said reception channel and squaring this reception signal and outputting a result;
a register means for storing square values calculated by said multiplication means;
a histogram means for calculating a histogram from square values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
19. The multi-channel echo cancelling device of claim 13, wherein said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels, based on said reception signal of the reception channel selected by said reception signal selection means, comprises a plurality of adaptive filters one-to-one corresponding to said mixed signals.
20. The multi-channel echo cancelling device of claim 19, wherein said adaptive filter is an adaptive transversal filter based on an LMS algorithm.
21. The multi-channel echo cancelling device of claim 19, wherein said adaptive filter is an adaptive transversal filter based on a learning identification method.
22. The multi-channel echo cancelling device of claim 14, wherein said means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
23. A multi-channel echo cancelling device comprising:
a means for calculating an absolute value of reception signal in each of a plurality of reception channels and selecting a reception channel of a reception signal having the greatest absolute value;
a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel;
and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
a means for calculating an absolute value of reception signal in each of a plurality of reception channels and selecting a reception channel of a reception signal having the greatest absolute value;
a means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means for each reception channel;
and a means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel.
24. The multi-channel echo cancelling device of claim 23, wherein said means for calculating an absolute value of reception signal in each reception channel and selecting a reception channel of a reception signal having the greatest absolute value comprises:
a plurality of absolute value calculation means, one-to-one corresponding to each reception signal, for calculating an absolute value of reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the greatest absolute value of reception signal based on the calculation result from said absolute value calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on a selection performed by said reception channel selection means.
a plurality of absolute value calculation means, one-to-one corresponding to each reception signal, for calculating an absolute value of reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the greatest absolute value of reception signal based on the calculation result from said absolute value calculation means; and a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on a selection performed by said reception channel selection means.
25. The multi-channel echo cancelling device of claim 24, wherein said absolute value calculation means comprises:
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel; and a mean calculation means for storing absolute values calculated by said absolute value calculation means and calculating a mean value of absolute values of a reception signal from said absolute values.
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel; and a mean calculation means for storing absolute values calculated by said absolute value calculation means and calculating a mean value of absolute values of a reception signal from said absolute values.
26. The multi-channel echo cancelling device of claim 24, wherein said absolute value calculation means comprises:
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel;
a register means for storing absolute values calculated by said absolute value calculation means; and a sorting means for sorting absolute values stored in said register means and outputting a median value of the sorted square values.
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel;
a register means for storing absolute values calculated by said absolute value calculation means; and a sorting means for sorting absolute values stored in said register means and outputting a median value of the sorted square values.
27. The multi-channel echo cancelling device of claim 23, wherein said absolute value calculation means comprises:
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel;
a register means for storing absolute values calculated by said absolute value calculation means;
a histogram means for calculating a histogram from absolute values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
an absolute value calculation means for calculating an absolute value of a reception signal of said reception channel for each reception channel;
a register means for storing absolute values calculated by said absolute value calculation means;
a histogram means for calculating a histogram from absolute values stored in said register means; and a maximum value detection means for outputting a representative value of a zone having maximum value in said histogram.
28. The multi-channel echo cancelling device of claim 23, wherein said means for generating an echo replica corresponding to an echo in each mixed signal generated by propagation of reception signals of said reception channels based on said reception signal of the reception channel selected by said reception signal selection means comprises a plurality of adaptive filters one-to-one corresponding to said mixed signals.
29. The multi-channel echo cancelling device of claim 28, wherein said adaptive filter is an adaptive transversal filter based on an LMS algorithm.
30. The multi-channel echo cancelling device of claim 28, wherein said adaptive filter is an adaptive transversal filter based on a learning identification method.
31. The multi-channel echo cancelling device of claim 23, wherein said means for cancelling said echo from said mixed signal by subtracting said echo replica from said mixed signal for each reception channel comprises a plurality of subtraction means, one-to-one corresponding to said mixed signals, for subtracting said echo replica from said mixed signal.
32. A multi-channel echo cancelling device comprising:
a plurality of power calculation means, one-to-one corresponding to each of a plurality of reception channels carrying a respective reception signal, for calculating the power of said reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation result from said power calculation means;
a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means;
a plurality of adaptive filters, one-to-one corresponding to mixed signals in which echoes generated by propagation of reception signals of said reception channels are mixed, for generating echo replicas corresponding to said echoes based on a reception signal supplied by said selection means; and a plurality of subtraction means, one-to-one corresponding to said mixed signals, for outputting a result by subtracting said echo replicas from said mixed signals.
a plurality of power calculation means, one-to-one corresponding to each of a plurality of reception channels carrying a respective reception signal, for calculating the power of said reception signal of said reception channel;
a reception channel selection means for selecting a reception channel having the biggest power of reception signal based on the calculation result from said power calculation means;
a selection means for receiving reception signals of reception channels and outputting a reception signal of a selected reception channel based on the selected result by said reception channel selection means;
a plurality of adaptive filters, one-to-one corresponding to mixed signals in which echoes generated by propagation of reception signals of said reception channels are mixed, for generating echo replicas corresponding to said echoes based on a reception signal supplied by said selection means; and a plurality of subtraction means, one-to-one corresponding to said mixed signals, for outputting a result by subtracting said echo replicas from said mixed signals.
33. The multi-channel echo cancelling device of claim 32, wherein said power calculation means is substituted by said absolute value calculation means for calculating an absolute value of a reception signal of said reception channel.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP5-167261 | 1993-05-31 | ||
| JP5167261A JPH084243B2 (en) | 1993-05-31 | 1993-05-31 | Method and apparatus for removing multi-channel echo |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2124662A1 CA2124662A1 (en) | 1994-12-01 |
| CA2124662C true CA2124662C (en) | 1999-03-23 |
Family
ID=15846462
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002124662A Expired - Fee Related CA2124662C (en) | 1993-05-31 | 1994-05-30 | Multi-channel echo cancelling method and a device thereof |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US5513265A (en) |
| EP (1) | EP0627825B1 (en) |
| JP (1) | JPH084243B2 (en) |
| CA (1) | CA2124662C (en) |
| DE (1) | DE69433000T2 (en) |
Families Citing this family (43)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3397269B2 (en) * | 1994-10-26 | 2003-04-14 | 日本電信電話株式会社 | Multi-channel echo cancellation method |
| US5602928A (en) * | 1995-01-05 | 1997-02-11 | Digisonix, Inc. | Multi-channel communication system |
| DE19517469A1 (en) * | 1995-05-12 | 1996-11-14 | Sel Alcatel Ag | Hands-free procedure for a multi-channel transmission system |
| US5761318A (en) * | 1995-09-26 | 1998-06-02 | Nippon Telegraph And Telephone Corporation | Method and apparatus for multi-channel acoustic echo cancellation |
| FR2739736B1 (en) * | 1995-10-05 | 1997-12-05 | Jean Laroche | PRE-ECHO OR POST-ECHO REDUCTION METHOD AFFECTING AUDIO RECORDINGS |
| US5923749A (en) * | 1995-12-06 | 1999-07-13 | Telefonaktiebolaget Lm Ericsson | Method and system for eliminating acoustic echos in a digital telecommunication system |
| US5774562A (en) * | 1996-03-25 | 1998-06-30 | Nippon Telegraph And Telephone Corp. | Method and apparatus for dereverberation |
| JPH09266456A (en) * | 1996-03-28 | 1997-10-07 | Mitsubishi Electric Corp | Echo canceler |
| TW432855B (en) * | 1996-04-25 | 2001-05-01 | Mitsubishi Electric Corp | Echo eliminator |
| US5796819A (en) * | 1996-07-24 | 1998-08-18 | Ericsson Inc. | Echo canceller for non-linear circuits |
| JP2950260B2 (en) * | 1996-11-22 | 1999-09-20 | 日本電気株式会社 | Noise suppression transmitter |
| JP3175622B2 (en) * | 1997-03-03 | 2001-06-11 | ヤマハ株式会社 | Performance sound field control device |
| FR2762467B1 (en) * | 1997-04-16 | 1999-07-02 | France Telecom | MULTI-CHANNEL ACOUSTIC ECHO CANCELING METHOD AND MULTI-CHANNEL ACOUSTIC ECHO CANCELER |
| US6430295B1 (en) | 1997-07-11 | 2002-08-06 | Telefonaktiebolaget Lm Ericsson (Publ) | Methods and apparatus for measuring signal level and delay at multiple sensors |
| SE512903C2 (en) * | 1997-10-29 | 2000-06-05 | Telia Ab | Method and apparatus for stereoacoustic echo cancellation |
| US6023514A (en) * | 1997-12-22 | 2000-02-08 | Strandberg; Malcolm W. P. | System and method for factoring a merged wave field into independent components |
| DE19801389A1 (en) * | 1998-01-16 | 1999-07-22 | Cit Alcatel | Echo cancellation method with adaptive FIR filters |
| DE19801390A1 (en) | 1998-01-16 | 1999-07-22 | Cit Alcatel | Device and method for echo cancellation with adaptive FIR filters |
| DE19815942A1 (en) | 1998-04-09 | 1999-10-14 | Cit Alcatel | Multi-channel Echo Canceller with Compander |
| AU2302401A (en) * | 1999-12-09 | 2001-06-18 | Frederick Johannes Bruwer | Speech distribution system |
| JP2002009675A (en) * | 2000-06-26 | 2002-01-11 | Nec Corp | Method and device for canceling echo for multiplex channel |
| JP3972921B2 (en) | 2004-05-11 | 2007-09-05 | ソニー株式会社 | Voice collecting device and echo cancellation processing method |
| US9554207B2 (en) | 2015-04-30 | 2017-01-24 | Shure Acquisition Holdings, Inc. | Offset cartridge microphones |
| US9565493B2 (en) | 2015-04-30 | 2017-02-07 | Shure Acquisition Holdings, Inc. | Array microphone system and method of assembling the same |
| US9653060B1 (en) * | 2016-02-09 | 2017-05-16 | Amazon Technologies, Inc. | Hybrid reference signal for acoustic echo cancellation |
| US9659555B1 (en) * | 2016-02-09 | 2017-05-23 | Amazon Technologies, Inc. | Multichannel acoustic echo cancellation |
| US10367948B2 (en) | 2017-01-13 | 2019-07-30 | Shure Acquisition Holdings, Inc. | Post-mixing acoustic echo cancellation systems and methods |
| US10522167B1 (en) * | 2018-02-13 | 2019-12-31 | Amazon Techonlogies, Inc. | Multichannel noise cancellation using deep neural network masking |
| WO2019198557A1 (en) * | 2018-04-09 | 2019-10-17 | ソニー株式会社 | Signal processing device, signal processing method, and signal processing program |
| WO2019231632A1 (en) | 2018-06-01 | 2019-12-05 | Shure Acquisition Holdings, Inc. | Pattern-forming microphone array |
| US11297423B2 (en) | 2018-06-15 | 2022-04-05 | Shure Acquisition Holdings, Inc. | Endfire linear array microphone |
| EP3854108A1 (en) | 2018-09-20 | 2021-07-28 | Shure Acquisition Holdings, Inc. | Adjustable lobe shape for array microphones |
| EP3667662B1 (en) * | 2018-12-12 | 2022-08-10 | Panasonic Intellectual Property Corporation of America | Acoustic echo cancellation device, acoustic echo cancellation method and acoustic echo cancellation program |
| US11438691B2 (en) | 2019-03-21 | 2022-09-06 | Shure Acquisition Holdings, Inc. | Auto focus, auto focus within regions, and auto placement of beamformed microphone lobes with inhibition functionality |
| EP3942842A1 (en) | 2019-03-21 | 2022-01-26 | Shure Acquisition Holdings, Inc. | Housings and associated design features for ceiling array microphones |
| US11558693B2 (en) | 2019-03-21 | 2023-01-17 | Shure Acquisition Holdings, Inc. | Auto focus, auto focus within regions, and auto placement of beamformed microphone lobes with inhibition and voice activity detection functionality |
| CN114051738A (en) | 2019-05-23 | 2022-02-15 | 舒尔获得控股公司 | Steerable speaker array, system and method thereof |
| TW202105369A (en) | 2019-05-31 | 2021-02-01 | 美商舒爾獲得控股公司 | Low latency automixer integrated with voice and noise activity detection |
| CN114467312A (en) | 2019-08-23 | 2022-05-10 | 舒尔获得控股公司 | Two-dimensional microphone array with improved directivity |
| US11552611B2 (en) | 2020-02-07 | 2023-01-10 | Shure Acquisition Holdings, Inc. | System and method for automatic adjustment of reference gain |
| USD944776S1 (en) | 2020-05-05 | 2022-03-01 | Shure Acquisition Holdings, Inc. | Audio device |
| WO2021243368A2 (en) | 2020-05-29 | 2021-12-02 | Shure Acquisition Holdings, Inc. | Transducer steering and configuration systems and methods using a local positioning system |
| CN116918351A (en) | 2021-01-28 | 2023-10-20 | 舒尔获得控股公司 | Hybrid Audio Beamforming System |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS6412728A (en) * | 1987-07-07 | 1989-01-17 | Fujitsu Ltd | Echo canceller |
| US5305309A (en) * | 1989-12-06 | 1994-04-19 | Fujitsu Limited | Echo canceller |
| JP2792252B2 (en) * | 1991-03-14 | 1998-09-03 | 日本電気株式会社 | Method and apparatus for removing multi-channel echo |
| JP2792311B2 (en) * | 1992-01-31 | 1998-09-03 | 日本電気株式会社 | Method and apparatus for removing multi-channel echo |
| JP2508574B2 (en) * | 1992-11-10 | 1996-06-19 | 日本電気株式会社 | Multi-channel eco-removal device |
-
1993
- 1993-05-31 JP JP5167261A patent/JPH084243B2/en not_active Expired - Fee Related
-
1994
- 1994-05-30 EP EP94108319A patent/EP0627825B1/en not_active Expired - Lifetime
- 1994-05-30 CA CA002124662A patent/CA2124662C/en not_active Expired - Fee Related
- 1994-05-30 DE DE69433000T patent/DE69433000T2/en not_active Expired - Fee Related
- 1994-05-31 US US08/251,236 patent/US5513265A/en not_active Expired - Lifetime
Also Published As
| Publication number | Publication date |
|---|---|
| DE69433000D1 (en) | 2003-09-11 |
| JPH06343051A (en) | 1994-12-13 |
| EP0627825B1 (en) | 2003-08-06 |
| CA2124662A1 (en) | 1994-12-01 |
| EP0627825A3 (en) | 1998-04-01 |
| EP0627825A2 (en) | 1994-12-07 |
| US5513265A (en) | 1996-04-30 |
| DE69433000T2 (en) | 2004-01-29 |
| JPH084243B2 (en) | 1996-01-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2124662C (en) | Multi-channel echo cancelling method and a device thereof | |
| US5323459A (en) | Multi-channel echo canceler | |
| JP2792252B2 (en) | Method and apparatus for removing multi-channel echo | |
| Benesty et al. | Advances in network and acoustic echo cancellation | |
| US5343522A (en) | Adaptive sparse echo canceller using a sub-rate filter for active tap selection | |
| CA2051147C (en) | Echo canceller using impulse response estimating method | |
| US5727073A (en) | Noise cancelling method and noise canceller with variable step size based on SNR | |
| JPH05218906A (en) | Method and device for eliminating multi-channel echo | |
| US6404886B1 (en) | Method and apparatus for echo cancelling with multiple microphones | |
| EP0760575A2 (en) | Echo canceller with adaptive and non-adaptive filters | |
| US6493448B1 (en) | Process for echo suppression with adaptive fir filters | |
| EP0760574A2 (en) | Echo canceller system with shared coefficient memory | |
| JPH06318885A (en) | Unknown system identifying method/device using band division adaptive filter | |
| JPH09116615A (en) | Echo suppressor | |
| EP1911258A1 (en) | Method and system for low delay echo cancelling | |
| WO2004015961A2 (en) | Estimating bulk delay in a telephone system | |
| US7062041B2 (en) | Device and method for carrying out multichannel acoustic echo cancellation with a variable number of channels | |
| JP3180739B2 (en) | Method and apparatus for identifying unknown system by adaptive filter | |
| CA1219649A (en) | Echo cancelling system comprising cascade-connected local echo cancellers operable independently of one another | |
| Kar et al. | An improved order estimation of MSF for stereophonic acoustic echo cancellation | |
| Kuo et al. | Adaptive acoustic echo cancellation microphone | |
| JP3152825B2 (en) | Acoustic echo canceller | |
| JPS62157450A (en) | Telephone set | |
| JPH02278926A (en) | Fir lattice hybrid echo canceller | |
| JPH07212278A (en) | Echo erasing device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKLA | Lapsed |