Method for processing speech signal data with reverberation filtering
View Patent ↗Method for processing speech signal data. A speech signal is divided into frames. Each frame is characterized by a frame number T representing a unique interval of time. Each speech signal is characterized by a power spectrum with respect to frame T and frequency band ω. A speech segment and a reverberation segment of the speech signal is determined. L filter coefficients W(k) (k=1, 2, . . . , L) respectively corresponding to L frames immediately preceding frame T are computed such that the L filter coefficients minimize a function Φ that is a linear combination of sum of squares of a residual speech power in the reverberation segment and a sum of squares of a subtracted speech power in the speech segment. The computed L filter coefficients are stored within storage media of the computing apparatus.
1. A method for processing speech signal data of at least one speech signal through use of a computing apparatus, the time domain of each speech signal divided into a plurality of frames, each frame characterized by a frame number T representing a unique interval of time, each speech signal characterized by a power spectrum with respect to frame T and frequency band ω of a plurality of frequency bands into which a frequency range of each speech signal has been divided, said method comprising:
computing a speech segment of a first speech signal, said speech segment consisting of a first set of frames of the plurality of frames of the first signal;
determining a reverberation segment of the first speech signal, said reverberation segment consisting of a second set of frames of the plurality of frames of the first signal;
computing with the computing apparatus L filter coefficients W(k) (k=1, 2, . . . , L) respectively corresponding to L frames immediately preceding frame T such that the L filter coefficients minimize a function Φ in accordance with a set of equations for Φ consisting of:
Φ
=
G
Tail
·
ϕ
Tail
+
G
Speech
·
ϕ
Speech
ϕ
Tail
=
∑
T
∈
Tail
∑
ω
{
X
ω
(
T
)
-
∑
k
=
1
L
W
(
k
)
·
X
ω
(
T
-
k
)
}
2
ϕ
Speech
=
∑
T
∈
Speech
∑
ω
{
∑
l
=
1
L
W
(
l
)
·
X
ω
(
T
-
l
)
}
2
wherein X ω (T) denotes a power spectrum of the first speech signal, wherein G Tail and G Speech are weighting coefficients, wherein the frames T in the summation over T ε Speech encompass the first set of frames in the speech segment, wherein the frames T in the summation over T ε Tail encompass the second set of frames in the reverberation segment, and wherein the frequency bands in the summation over ω encompass the plurality of frequency bands; and
storing the computed L filter coefficients within storage media of the computing apparatus.
2. The method of claim 1 , wherein said computing the reverberation segment comprises:
computing speech tracks S(T) and P(T); and
assigning to the reverberation segment those frames of the plurality of frames of the first speech signal that satisfy P(T)−S(T)>γ, wherein γ denotes a specified threshold value, and wherein said computing speech tracks S(T) and P(T) are performed in accordance with the equations of:
energy
(
T
)
=
10.0
*
log
10
(
1
N
∑
i
=
1
N
x
[
i
]
2
)
P
(
T
)
=
10
C
1
*
energy
(
T
)
Q
(
T
)
=
(
1
-
α
l
)
*
Q
(
T
-
1
)
+
α
l
*
P
(
T
)
α
l
=
C
2
*
C
3
*
Q
(
T
-
1
)
2
P
(
T
)
2
S
(
T
)
=
(
1
-
α
h
)
*
S
(
T
-
1
)
+
α
h
*
P
(
T
)
α
h
=
C
3
*
P
(
T
)
2
Q
(
T
-
1
)
2
wherein x[i] is a measure of the amplitude of an observed speech signal pulse coded modulation (PCM) data value i in frame T, wherein N is a total number of PCM data values in the frame T, and wherein C1, C2, and C3 are specified constants.
3. The method of claim 1 , wherein the method further comprises computing G Tail and G Speech according to the equations of:
G
Tail
=
{
1
N
Tail
∑
T
∈
Tail
∑
ω
{
X
ω
(
T
)
}
}
-
2
G
Speech
=
{
1
N
Speech
∑
T
∈
Speech
∑
ω
{
X
ω
(
T
)
}
}
-
2
wherein N Tail is the total number of frames in the trailing reverberation segment (T ε Tail), and wherein N Speech is the total number of frames in the speech segment (T ε Speech).
4. The method of claim 1 , wherein said computing the L filter coefficients comprises:
computing a matrix A;
computing a vector C; and
computing a vector C according to B=A −1 ·C,
wherein
C
=
A
·
B
A
=
[
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
1
)
·
X
ω
(
T
-
1
)
⋯
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
L
)
·
X
ω
(
T
-
1
)
⋮
⋰
⋮
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
1
)
·
X
ω
(
T
-
L
)
⋯
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
L
)
·
X
ω
(
T
-
L
)
]
B
=
[
W
(
1
)
⋮
W
(
L
)
]
C
=
[
G
Tail
·
∑
T
∈
Tail
∑
ω
X
ω
(
T
)
·
X
ω
(
T
-
1
)
⋮
G
Tail
·
∑
T
∈
Tail
∑
ω
X
ω
(
T
)
·
X
ω
(
T
-
L
)
]
5. The method of claim 1 , wherein the method further comprises:
computing a dereverberated power spectrum D′ ω (T) according to:
D
ω
′
(
T
)
=
X
ω
′
(
T
)
-
∑
k
=
1
L
W
′
(
k
)
·
X
ω
′
(
T
-
k
)
wherein X′ ω (T) is a power spectrum of a second speech signal for frame number T of frequency band ω, wherein if the computed W(k) is nonnegative for k=1, 2, . . . , L then W′(k)=W(k), and wherein if the computed W(k) is negative for at least one k of k=1, 2, . . . L then setting W′(k)=0 for the values of k at which the computed W(k) is negative and calculating W′(k) via a repetitive relaxation procedure for the values of k at which the computed W(k) is nonnegative; and
storing the computed D′ ω (T) within the storage media of the computing apparatus.
6. The method of claim 5 , wherein the computed W(k) is nonnegative for k=1, 2, . . . , L.
7. The method of claim 5 , wherein the computed W(k) is negative for at least one k of k=1, 2, . . . L.
8. The method of claim 5 , wherein the method further comprises:
determining a noise segment consisting of N Noise frames of the plurality of frames of the first signal, wherein the N Noise frames are not comprised by either the speech segment or the reverberation segment;
computing a noise spectrum U ω of the first speech signal via
U
ω
=
1
N
Noise
∑
T
∈
Noise
X
ω
(
T
)
wherein the frames T in the summation over T ε Noise encompass the N Noise frames in the noise segment;
if D′ ω (T)≧βU ω such that β is a specified constant, then setting a dereverberated power spectrum Z ω (T)=D′ ω (T) otherwise setting Z ω (T)=βU ω ; and
storing Z ω (T) within the storage media of the computing apparatus.
9. The method of claim 5 , wherein the second speech signal consists of the first speech.
10. The method of claim 5 , wherein the second speech signal occurs after the first speech signal has ended.
11. The method of claim 5 , wherein the second speech signal consists of the first speech signal and X′ ω (T) consists of X ω (T), and wherein the method further comprises after said computing D′ ω (T):
receiving a plurality of additional sets of speech signal frames;
cumulatively adding each additional set of speech signal frames to the frames of the first speech signal to generate a corresponding power spectrum X″ ω (T) for each additional set of speech signal frames; and
after generating the power spectrum X″ ω (T) for each additional set of speech signal frames: computing updated L filter coefficients W″(k) (k=1, 2, . . . , L) corresponding to power spectrum X″ ω (T) in accordance with the set of equations for Φ in which X″ ω (T) replaces X ω (T) and W″(k) replaces W(k); and computing an updated dereverberated power spectrum D″ ω (T) according to:
D
ω
″
(
T
)
=
X
ω
″
(
T
)
-
∑
k
=
1
L
W
″
(
k
)
·
X
ω
″
(
T
-
k
)
.
12. The method of claim 11 , wherein each additional set of speech signal frames consists of one additional speech signal frame.
13. A method for processing speech signal data of at least one speech signal through use of a computing apparatus, the time domain of each speech signal divided into a plurality of frames, each frame characterized by a frame number T representing a unique interval of time, each speech signal characterized by a power spectrum with respect to frame T and frequency band ω of a plurality of frequency bands into which a frequency range of each speech signal has been divided, said method comprising:
determining a speech segment of a first speech signal, said speech segment consisting of a first set of frames of the plurality of frames of the first signal;
computing a reverberation segment of the first speech signal, said reverberation segment consisting of a second set of frames of the plurality of frames of the first signal;
computing with the computing apparatus L filter coefficients W(k) (k=1, 2, . . . , L) respectively corresponding to L frames immediately preceding frame T such that the L filter coefficients minimize a function Φ in accordance with a set of equations for Φ consisting of:
Φ
=
G
Tail
·
ϕ
Tail
+
G
Speech
·
ϕ
Speech
ϕ
Tail
=
∑
T
∈
Tail
∑
ω
{
X
ω
(
T
)
-
∑
k
=
1
L
W
(
k
)
·
X
ω
(
T
-
k
)
}
2
ϕ
Speech
=
∑
T
∈
Speech
∑
ω
{
∑
l
=
1
L
W
(
l
)
·
X
ω
(
T
-
l
)
}
2
wherein X ω (T) denotes a power spectrum of the first speech signal, wherein G Tail and G Speech are weighting coefficients, wherein the frames T in the summation over TεSpeech encompass the first set of frames in the speech segment, wherein the frames T in the summation over TεTail encompass the second set of frames in the reverberation segment, and wherein the frequency bands in the summation over ω encompass the plurality of frequency bands; and
storing the computed L filter coefficients within storage media of the computing apparatus.
wherein said computing the reverberation segment comprises:
computing speech tracks S(T) and P(T); and
assigning to the reverberation segment those frames of the plurality of frames of the first speech signal that satisfy P(T)−S(T)>γ, wherein γ denotes a specified threshold value, and wherein said computing speech tracks S(T) and P(T) are performed in accordance with the equations of:
energy
(
T
)
=
10.0
*
log
10
(
1
N
∑
i
=
1
N
x
[
i
]
2
)
P
(
T
)
=
10
C
1
*
energy
(
T
)
Q
(
T
)
=
(
1
-
α
l
)
*
Q
(
T
-
1
)
+
α
l
*
P
(
T
)
α
l
=
C
2
*
C
3
*
Q
(
T
-
1
)
2
P
(
T
)
2
S
(
T
)
=
(
1
-
α
h
)
*
S
(
T
-
1
)
+
α
h
*
P
(
T
)
α
h
=
C
3
*
P
(
T
)
2
Q
(
T
-
1
)
2
wherein x[i] is a measure of the amplitude of an observed speech signal pulse coded modulation (PCM) data value i in frame T, wherein N is a total number of PCM data values in the frame T, and wherein C1, C2, and C3 are specified constants;
wherein the method further comprises computing G Tail and G Speech according to the equations of:
G
Tail
=
{
1
N
Tail
∑
T
∈
Tail
∑
ω
{
X
ω
(
T
)
}
}
-
2
G
Speech
=
{
1
N
Speech
∑
T
∈
Speech
∑
ω
{
X
ω
(
T
)
}
}
-
2
wherein N Tail is the total number of frames in the trailing reverberation segment (TεTail), and wherein N Speech is the total number of frames in the speech segment (TεSpeech);
wherein said computing the L filter coefficients comprises:
computing a matrix A;
computing a vector C; and
computing a vector C according to B=A −1 ·C,
wherein
C
=
A
·
B
A
=
[
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
1
)
·
X
ω
(
T
-
1
)
…
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
L
)
·
X
ω
(
T
-
1
)
⋮
⋱
⋮
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speach
∑
ω
X
ω
(
T
-
1
)
·
X
ω
(
T
-
L
)
…
G
Tail
or
Speech
·
∑
T
∈
Tail
or
Speech
∑
ω
X
ω
(
T
-
L
)
·
X
ω
(
T
-
L
)
]
B
=
[
W
(
1
)
⋮
W
(
L
)
]
C
=
[
G
Tail
·
∑
T
∈
Tail
∑
ω
X
ω
(
T
)
·
X
ω
(
T
-
1
)
⋮
G
Tail
·
∑
T
∈
Tail
∑
ω
X
ω
(
T
)
·
X
ω
(
T
-
L
)
]
wherein the method further comprises:
computing a dereverberated power spectrum D′ ω (T) according to:
D
ω
′
(
T
)
=
X
ω
′
(
T
)
-
∑
k
=
1
L
W
′
(
k
)
·
X
ω
′
(
T
-
k
)
wherein X′ ω (T) is a power spectrum of a second speech signal for frame number T of frequency band ω, wherein if the computed W(k) is nonnegative for k=1, 2, . . . , L then W′(k)=W(k), and wherein if the computed W(k) is negative for at least one k of k=1, 2, . . . L then setting W′(k)=0 for the values of k at which the computed W(k) is negative and calculating W′(k) via a repetitive relaxation procedure for the values of k at which the computed W(k) is nonnegative; and
storing the computed D′ ω (T) within the storage media of the computing apparatus;
wherein the computed W(k) is nonnegative for k=1, 2, . . . , L;
wherein the method further comprises:
determining a noise segment consisting of N Noise frames of the plurality of frames of the first signal, wherein the N Noise frames are not comprised by either the speech segment or the reverberation segment;
computing a noise spectrum U ω of the first speech signal via
U
ω
=
1
N
Noise
∑
T
∈
Noise
X
ω
(
T
)
wherein the frames T in the summation over TεNoise encompass the N Noise frames in the noise segment;
if D′ ω (T)≧βU ω such that β is a specified constant, then setting a dereverberated power spectrum Z ω (T)=D′ ω (T) otherwise setting Z ω (T)=βU ω ; and
storing Z ω (T) within the storage media of the computing apparatus;
wherein the second speech signal consists of the first speech signal and X′ ω (T) consists of X ω (T), and wherein the method further comprises after said computing D′ ω (T):
receiving a plurality of additional sets of speech signal frames;
cumulatatively adding each additional set of speech signal frames to the frames of the first speech signal to generate a corresponding power spectrum X″ ω (T) for each additional set of speech signal frames; and
after generating the power spectrum X″ ω (T) for each additional set of speech signal frames: computing updated L filter coefficients W″(k) (k=1, 2, . . . , L) corresponding to power spectrum X″ ω (T) in accordance with the set of equations for Φ in which X″ ω (T) replaces X ω (T) and W″ (k) replaces W(k); and computing an updated dereverberated power spectrum D″ ω (T) according to:
D
ω
″
(
T
)
=
X
ω
″
(
T
)
-
∑
k
=
1
L
W
″
(
k
)
·
X
ω
″
(
T
-
k
)
.