Electrical audio signal processing systems and devices – Dereverberators
Reexamination Certificate
1999-03-02
2003-04-22
Lee, Ping (Department: 2644)
Electrical audio signal processing systems and devices
Dereverberators
C379S406080, C379S406100
Reexamination Certificate
active
06553122
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a multi-channel acoustic echo cancellation method and apparatus which cancel acoustic echoes that cause howling and give rise to psychoacoustical problems in a teleconference system provided with a multi-receive-channel system, and a recording medium that has recorded thereon a program for implementing the multi-channel acoustic echo cancellation.
In recent years, various forms of telecommunication have emerged with the widespread proliferation of digital networks, such as ISDN, LAN and the Internet, and with the development of high efficiency speech and image coding techniques. In a TV conference or desktop teleconference in which each participant can talk while looking at the other party through utilization of a wide screen television, or a personal computer or workstation placed at the participant's seat, a hands-free telecommunication system is often employed which allows two or more persons to participate in the conversation with ease and provides a more realistic teleconferencing environment. However, this system that uses loudspeakers and microphones inevitably suffers from echoes and howling. To overcome this problem, acoustic echo canceller techniques are indispensable.
In such situations as mentioned above, acoustic echo cancellers are now in wide use, but they are mostly for one-channel audio use and cancel an acoustic echo over only one channel from one loudspeaker to one microphone. On the other hand, stereo is common in many TV broadcast programs and music media, and there is also a strong demand for the realization of a multi-channel hands-free telecommunication system. To meet this requirement, it is necessary to implement a multi-channel acoustic echo canceller which permits cancellation of acoustic echoes from two or more loudspeakers (channels) to a microphone. In recent years, technical problems and solutions thereto have been investigated very actively toward the realization of such a multi-channel echo canceller.
Conventionally, such a configuration as depicted in
FIG. 1
is used to cancel acoustic echoes in a teleconference system which is composed of a receive system of N (N≧2) channels and a send system of M (M≧2) channels. That is, N-channel echo cancellers
22
1
,
22
2
, . . . ,
22
M
, which constitute an echo cancellation part
22
, are connected between received signal terminals
11
1
,
11
2
, . . . ,
11
N
of all the N receive channels and each of the M send channels, respectively. Received signals from the received signal terminals
11
1
,
11
2
, . . . ,
11
N
of the respective receive channels are applied to loudspeakers
12
1
,
12
2
, . . . ,
12
N
, from which they are radiated as acoustic signals. The illustrated echo canceller system cancels acoustic echoes which are produced when the acoustic signals are picked up by microphones
16
1
,
16
2
, . . . ,
16
M
after propagating over echo paths
15
nm
(where 1≦n≦N and 1≦m≦M).
The N-channel echo cancellers
22
1
,
22
2
, . . . ,
22
M
have the same configuration, which is such as depicted in FIG.
2
. This configuration is described as being applied to a two-channel system in B. Widow and S. D. Stearns, “Adaptive signal processing,” Prince-Hall, Inc., pp.198-200 (1985). In the configuration of
FIG. 2
, received signals x
1
(k), x
2
(k), . . . , x
N
(k) are input into adaptive filters
221
1
,
221
2
, . . . ,
221
N
each of which form N estimated echo paths. The outputs from the adaptive filters
221
1
,
221
2
, . . . ,
221
N
are added together by an adder
222
, by which an echo replica y′
m
(k) is generated. The difference between the echo replica y′
m
(k) and the picked-up output signal (echo) y
m
(k) from the microphone
16
m
is detected by a subtractor
223
. An error signal (a residual echo) e
m
(k) provided from the subtractor
223
is fed back to the adaptive filters
221
1
through
221
N
. The error signal and the received signals x
1
(k) to x
N
(k) are used to determine filter coefficient vectors, for example, by an NLMS algorithm, and the adaptive filters
221
1
to
221
N
are controlled adaptively.
Incidentally, though not shown in
FIG. 1
, the echo cancellers
22
1
through
22
M
aim to prevent that when an acoustic signal z
m
(k) originally intended to send is input into the microphones
16
1
to
16
M
and sent out through the echo cancellers
22
1
to
22
M
, a reproduced sound from each loudspeaker, picked up by each microphone, is sent out as an echo y
m
(k) together with the signal z
m
(k). In other words, the error signal e
m
(k) provided as the result of echo cancellation contains the signal z
m
(k) that ought to be sent. However, the present invention is directed toward the cancellation of an echo signal y
m
(k) which is produced when the acoustic signals radiated from the loudspeakers are picked up by the microphone
16
m;
hence, no particular mention will be made herein to the signal z
m
(k) that ought to be sent.
When cross-correlation among the received signals x
1
(k) to x
N
(k) is low, the adaptive filters
221
1
,
221
2
, . . . ,
221
N
estimate the corresponding echo paths with relatively high accuracy, thus producing echo replicas that accurately simulate the acoustic echoes to be cancelled. In actual teleconferences, however, speech of one speaker is sent over multiple channels from the far end in many cases, and the received signals are so highly cross-correlated that the convergence speeds and accuracies of the adaptive filters are both degraded, often resulting in failure to provide intended echo cancellation capabilities. As a solution to this problem, there is proposed in U.S. Pat. No. 5,661,813 a scheme which reduces or changes the cross-correlation of the received signals by pre-processing them prior to their input into the N-channel echo cancellers
22
1
,
22
2
, . . . ,
22
M
.
The configuration disclosed in the above U.S. patent is such as depicted in
FIG. 3
, in which a pre-processing part
30
equipped with the above-mentioned function is placed between the received signal terminals
11
1
to
11
N
and the loudspeakers
12
1
to
12
N
and the N-channel echo cancellers
22
1
to
22
M
. In
FIG. 4
there is shown an example of the configuration of the pre-processing part
30
. The received signals from the received signal terminals
11
1
to
11
N
and additive signals, generated in additive signal generating parts
301
1
,
301
2
, . . . ,
301
N
, are added by adders
302
1
,
302
2
, . . . ,
302
N
, from which processed signals x
1
′(k), x
2
′(k), . . . , x
N
′(k) are provided, respectively. In the generation of the additive signals, received signal information may be used or may not be used. By increasing the magnitudes of the additive signals, the convergence characteristics of the adaptive filters
221
1
,
221
2
, . . . ,
221
N
can be improved. A similar scheme is disclosed in U.S. Pat. No. 5,828,756. Many of pre-processing syste
m
s already proposed, for example, in U.S. Pat. No. 5,661,813 and J. Benesty, D. R. Morgan, and M. M. Sondi, “A Better Understanding and an Improved Solution to the Problems of Stereophonic Acoustic Echo Cancellation,” Proc. ICASSP97, vol. 1, pp.303-306 (1997), can be implemented by mathematical modifications with the configuration depicted in FIG.
4
. For example, even in a pre-processing part which pre-processes each of N-channel signals x
i
(k) (where i=1, 2, . . . , N) at a discrete time k by using a processing function f
i
(where i=1, 2, . . . , N) and outputs a processed signal x
i
′(k) (where i=1, 2, . . . , N) in the following form:
x
i
′(
k
)=
f
i
[x
i
(
k
)] (1)
Eq. (1) can be modified as follows:
x
i
′(
k
)=
x
i
(
k
)+(
f
i
[x
i
(
k
)]−
x
i
(
k
)) (2)
Therefore, the additive signal f
i
[x
i
(k)]−x
i
(k) can be regarded as a signal obtained by pre-processing the original signal x
i
(k).
With a view to improving the convergence characteristics o
Haneda Yoichi
Kaneda Yutaka
Makino Shoji
Shimauchi Suehiro
Lee Ping
Nippon Telegraph and Telephone Corporation
LandOfFree
Method and apparatus for multi-channel acoustic echo... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for multi-channel acoustic echo..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for multi-channel acoustic echo... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3027169