Electrical audio signal processing systems and devices – Dereverberators
Reexamination Certificate
1999-11-22
2002-07-23
Isen, Forester W. (Department: 2747)
Electrical audio signal processing systems and devices
Dereverberators
C381S083000, C381S093000, C379S406010
Reexamination Certificate
active
06424720
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a method and a device for reducing multi-channel acoustic echo and restoring auditory perspective or “adapting sound to space”. It finds a particularly important application in systems for digitally transmitting sound signals on several transmission channels between a local site and one or more remote sites, each site having several microphones and several loudspeakers. Such systems are referred to as “multi-channel” systems with regard to the sound take (the sound is picked up by several microphones), transmission (on several channels) and sound restoration (by means of several loudspeakers).
By way of example, although this is not restrictive, the invention will be described in an application involving video conference rooms, both on the basis of transmission between two rooms, in what is referred to as a “point to point” configuration (where the image from the remote room is displayed on a screen or several viewpoints of the remote room are displayed on several screens) and with regard to transmission within a network comprising more than two rooms in what is referred to as a “multi-point” configuration (where the images of the remote rooms are displayed on several screens).
In a room having N microphones and M loudspeakers, there are N×M acoustic echo paths. On the one hand, conventional systems for eliminating acoustic echo do not allow echo to be controlled at a reasonable cost in such a context. On the other, connecting each microphone in a local room to a loudspeaker of a remote room for transmission purposes in order to obtain the best possible distribution of the sound would multiply the number of transmission channels, making the transmission cost prohibitive in terms of commercial development.
Apart from the conventional systems for eliminating acoustic echo, systems for reducing acoustic echo are known, based on variations in the level of the sound signals. A major disadvantage generally encountered in such systems is the constraint placed on interactivity, i.e. a reduction in reception quality, in particular a considerable variation in sound level on reception in what are referred to as “double speech” situations, i.e. where, in a given local room, there is an effective local sound signal present simultaneously with an effective sound signal from the remote room.
SUMMARY OF THE INVENTION
The objective of the present invention is to provide a method and a device for reducing multi-channel acoustic echo and adapting sound to space, which will allow the echo to be reduced whilst maintaining interactivity and which, although using a relatively small number of transmission channels, will guarantee restoration of the auditory perspective. In its application to video-conferencing, the present invention will enable the remote meeting to be conducted as a natural communication situation.
To this end, the invention specifically proposes a method of reducing acoustic echo and adapting sound to space in a system for digitally transmitting sound signals on P transmission channels between a local site and at least one remote site, each having N microphones and M loudspeakers, N, M, P being integers, it being possible for the values of N and M to differ depending on the sites, whereby :
(a) a cumulative distribution function is computed for each microphone signal x
i
(n) from the local site, i being an integer ranging between 1 and N and n denoting the time rank of the samples, and a cumulative distribution function is computed for each loudspeaker signal z
j
(n) from the local site, j being an integer ranging between 1 and M (the concept of cumulative distribution function will be defined below); then, for every i, 1≦i≦N,
(b) a first attenuation factor G
mic
(i,n) is computed for the microphone signal x
i
(n) from the local site on the basis of a ratio between the cumulative distribution functions obtained previously;
(c) the first attenuation factor G
mic
(i,n) is adjusted so as to obtain a second attenuation factor G′
mic
(i,n) defined as follows:
G′
mic
(
i,n
)=
S
1
(
G
mic
(
i,n
))
where S
1
(G
mic
(i,n))=s if G
mic
(i,n)≦s,
S
1
(G
mic
(i,n))=G
mic
(i,n) if s<G
mic
(i,n)<1 and
S
1
(G
mic
(i,n))=1 if G
mic
(i,n)≧1,
s being a predetermined minimum threshold which is strictly less than 1;
(d) on the basis of the cumulative distribution functions of microphone and loudspeaker signals computed previously, it is determined
whether the microphone signal x
i
(n) is an echo signal only or a signal coming solely from the local site in the case of a first situation or
if the microphone signal x
i
(n) contains components from the local site and other components from the remote site in the case of a second situation;
(e) a third attenuation factor G″
mic
(i,n) is computed which, in the first situation, is equal to the second attenuation factor G′
mic
(i,n) and in said second situation is equal to the second attenuation factor G′
mic
(i,n) but in which the minimum threshold s in the computation used to obtain it is increased by a predetermined value;
(f) a fourth attenuation factor &Ggr;(i,n) is computed on the basis of a ratio between the cumulative distribution functions of microphone signals;
(g) the fourth attenuation factor &Ggr;(i,n) is adjusted in order to obtain a fifth attenuation factor &Ggr;′(i,n) defined as follows:
&Ggr;′(i,n)=
S
2
(&Ggr;(
i,n
))
where S
2
(&Ggr;(i,n))=s′ if &Ggr;(i,n)≦s′,
S
2
(&Ggr;(i,n) )=&Ggr;(i,n) if s′<&Ggr;(i,n)<1 and
S
2
(&Ggr;(i,n))=1 if &Ggr;(i,n)≧1,
s′ being a predetermined minimum threshold strictly less than 1;
(h) the product of the third and fifth attenuation factors G″
mic
(i,n) and r′(i,n) obtained previously is computed so as to obtain a global attenuation factor G*
mic
(i,n) defined by:
G*
mic
(
i,n
)=
G″
mic
(
i,n
). &Ggr;(
i,n
)
(i) the global attenuation factor G*
mic
(i,n) is adjusted so as to obtain a weighting factor &bgr;
i
(n) defined as follows:
&bgr;
i
(
n
)=
S
4
(
G*
mic
(
i,n
))
where S
4
(G*
mic
(i,n))=s″ if G*
mic
(i,n)≦s″ and
S
4
(G*
mic
(i,n))=G*
mic
(i,n) if s″<G*
mic
(i,n)≦1,
s″ being a predetermined minimum threshold strictly less than 1;
(j) a signal y
k
(n) is transmitted on each transmission channel, k being an integer between 1 and P, in the form of a linear combination of the weighted microphone signals x
i
(n), defined as follows:
y
k
(
n
)
=
∑
i
=
1
N
α
k
,
i
(
n
)
·
β
i
(
n
)
·
x
i
(
n
)
where &agr;
k.i
(n) denotes the predetermined real coding coefficients and &bgr;
i
(n) denotes the weighting factors obtained previously; then, for every integer j, 1≦j≦M:
(k) a sixth attenuation factor G
HP
(j,n) is computed for the loudspeaker signal z
j
(n) from the remote site on the basis of cumulative distribution functions calculated for each transmitted signal y
k
(n) from the local site;
(l) the sixth attenuation factor G
HP
(j,n) is adjusted so as to obtain a weighting factor &lgr;
j
(n) defined as follows:
&lgr;
j
(
n
)=
S
3
(
G
HP
(
j,n
))
where S
3
(G
HP
(j,f))=* if G
HP
(j,n)≦s*,
S
3
(G
HP
(j,n))=(G
HP
(j,n) if s*<G
HP
(j,n)<1 and
S
3
(G
HP
(j,n))=1 if G
HP
(j,n)≧1,
s* being a predetermined minimum threshold strictly less than 1;
(m) the loudspeaker signal z
j
(n) of the remote site is determined on the basis of a linear combination of the weighted transmitted signals y
k
(n), defined as follows:
z
j
(
n
)
=
λ
j
(
n
)
·
∑
k
=
1
P
γ
j
.
k
(
n
)
·
y
k
(
n
)
where &lgr;
j.k
(n) denotes the predetermined real decoding coefficients and where &lgr;
j
(n) denotes the weighting factors obtained previously; and
(n) the loudspeaker signal z
j
(n) thus obtained is emitted on the j-
th
loudspeaker of the remote site.
The operations outlined above are performed in a s
Emerit Marc
Mahieux Yannick
Saliou Alain
Thomas Jean-Philippe
France Telecom
Grier Laura A.
Isen Forester W.
Marshall Gerstein & Borun
LandOfFree
Method and device for reducing 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 device for reducing multi-channel acoustic echo..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for reducing multi-channel acoustic echo... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2834733