Parameter determination device, method, program and recording medium for determining a parameter indicating a characteristic of sound signal
A parameter determination device includes: a spectral envelope estimating portion performing estimation of a spectral envelope using a parameter η 0 specified in a predetermined method, regarding the η 0 -th power of absolute values of a frequency domain sample sequence corresponding to a time-series signal as a power spectrum on the assumption that the parameter η 0 and a parameter η are positive numbers; a whitened spectral sequence generating portion obtaining a whitened spectral sequence which is a sequence obtained by dividing the frequency domain sample sequence by the spectral envelope; and a parameter acquiring portion determining such a parameter η that generalized Gaussian distribution with the parameter η as a shape parameter approximates a histogram of the whitened spectral sequence.
1. A parameter determination device comprising:
processing circuitry configured to:
receive a time domain sound signal as a time-series signal;
perform estimation of a spectral envelope using a parameter η 0 specified in a predetermined method, regarding the η 0 -th power of absolute values of a frequency domain sample sequence corresponding to the time-series signal as a power spectrum on the assumption that the parameter η 0 and a parameter η are positive numbers;
obtain a whitened spectral sequence which is a sequence obtained by dividing the frequency domain sample sequence by the spectral envelope; and
determine, as a parameter representing the characteristic of the time-series signal, such a parameter η that generalized Gaussian distribution with the parameter η as a shape parameter approximates a histogram of the whitened spectral sequence.
2. The parameter determination device according to claim 1 , wherein processes of the processing circuitry in which the parameter η determined by the processing circuitry is caused to be the parameter η 0 specified in the predetermined method, are further performed once or more times.
3. The parameter determination device according to claim 1 or 2 , wherein the processing circuitry determines the parameter η based on a value of m q1 , which is a q1-th order moment of the whitened spectral sequence, and a value of m q2 , which is a q2-th order moment of the whitened spectral sequence, on the assumption that q1 and q2 are predetermined different positive integers.
4. The parameter determination device according to claim 3 , wherein, on the assumption that N is a predetermined positive number, that k=0,1, . . . , N−1 is satisfied, that each value of the whitened spectral sequence is indicated by X W (k), that Γ is a gamma function, and that F −1 is an inverse function of a function F, η determined by the parameter acquiring portion is defined by the following expression:
[
Expression
15
]
η
=
F
′
-
1
(
m
q
1
(
m
q
2
)
q
1
/
q
2
)
F
′
(
η
)
=
Γ
(
(
q
1
+
1
)
/
η
)
(
Γ
(
1
/
η
)
)
1
-
q
1
/
q
2
(
Γ
(
(
q
2
+
1
)
/
η
)
)
q
1
/
q
2
m
q
1
=
1
N
∑
k
=
0
N
-
1
X
W
(
k
)
q
1
,
m
q
2
=
1
N
∑
k
=
0
N
-
1
X
W
(
k
)
q
2
.
5. The parameter determination device according to claim 4 , wherein the processing circuitry calculates m q1 /((m q2 ) q1/q2 ) based on the whitened spectral sequence and, by referring to a plurality of different pairs of η and F(η) corresponding to η prepared in advance, obtains η corresponding to F(η) closest to the calculated m q1 /((m q2 ) q1/q2 ).
6. The parameter determination device according to claim 4 , wherein, on the assumption that an approximate curve function of the inverse function F −1 is ˜F −1 , the processing circuitry calculates m q1 /((m q2 ) q1/q2 ) based on the whitened spectral sequence and determines η by calculating an output value when the calculated m q1 /((m q2 ) q1/q2 ) is inputted to the approximate curve function ˜F −1 .
7. The parameter determination device according to claim 1 , wherein
the parameter determination device determines the parameter η for each frame; and
the parameter η 0 specified in the predetermined method for a frame for which the parameter η is to be determined currently is η determined for a frame before the frame for which the parameter η is to be determined currently.
8. A non-transitory computer-readable recording medium in which a program for causing a computer to function as each portion of the parameter determination device according to claim 1 is recorded.
9. The parameter determination method, implemented by a parameter determination device, comprising:
receiving, by processing circuitry of the parameter determination device, a time domain sound signal as a time-series signal;
a spectral envelope estimating step, by the processing circuitry, of performing estimation of a spectral envelope using a parameter η 0 specified in a predetermined method, regarding the η 0 -th power of absolute values of a frequency domain sample sequence corresponding to the time-series signal as a power spectrum on the assumption that the parameter η 0 and a parameter η are positive numbers;
a whitened spectral sequence generating step, by the processing circuitry, of obtaining a whitened spectral sequence which is a sequence obtained by dividing the frequency domain sample sequence by the spectral envelope; and
a parameter acquiring step, by the processing circuitry, of determining, as a parameter representing the characteristic of the time-series signal, such a parameter η that generalized Gaussian distribution with the parameter η as a shape parameter approximates a histogram of the whitened spectral sequence.
10. The parameter determination method according to claim 9 , wherein processes of the spectral envelope estimating step, the whitened spectral sequence generating step and the parameter acquiring step, in which the parameter η determined by the parameter acquiring portion is caused to be the parameter η 0 specified in the predetermined method, are further performed once or more times.
11. The parameter determination method according to claim 9 or 10 , wherein the parameter acquiring step includes determining the parameter η based on a value of m q1 , which is a q1-th order moment of the whitened spectral sequence, and a value of m q2 , which is a q2-th order moment of the whitened spectral sequence, on the assumption that q1 and q2 are predetermined different positive integers.
12. The parameter determination method according to claim 11 , wherein, on the assumption that N is a predetermined positive number, that k=0,1, . . . , N−1 is satisfied, that each value of the whitened spectral sequence is indicated by X W (k), that Γ is a gamma function, and that F −1 is an inverse function of a function F, η determined by the parameter acquiring portion is defined by the following expression:
[
Expression
16
]
η
=
F
′
-
1
(
m
q
1
(
m
q
2
)
q
1
/
q
2
)
F
′
(
η
)
=
Γ
(
(
q
1
+
1
)
/
η
)
(
Γ
(
1
/
η
)
)
1
-
q
1
/
q
2
(
Γ
(
(
q
2
+
1
)
/
η
)
q
1
/
q
2
)
m
q
1
=
1
N
∑
k
=
0
N
-
1
X
W
(
k
)
q
1
,
m
q
2
=
1
N
∑
k
=
0
N
-
1
X
W
(
k
)
q
2
.
13. The parameter determination method according to claim 12 , wherein the parameter acquiring step includes calculating m q1 /((m q2 ) q1/q2 ) based on the whitened spectral sequence and, by referring to a plurality of different pairs of η and F(η) corresponding to η prepared in advance, obtains η corresponding to F(η) closest to the calculated m q1 /((m q2 ) q1/q2 ).
14. The parameter determination method according to claim 12 , wherein, on the assumption that an approximate curve function of the inverse function F −1 is ˜F −1 , the parameter acquiring step includes calculating m q1 /((m q2 ) q1/q2 ) based on the whitened spectral sequence and determines η by calculating an output value when the calculated m q1 /((m g2 ) q1/q2 ) is inputted to the approximate curve function ˜F − 1.
15. The parameter determination method according to claim 9 , wherein
the parameter determination method determines the parameter η for each frame; and
the parameter η 0 specified in the predetermined method for a frame for which the parameter η is to be determined currently is η determined for a frame before the frame for which the parameter η is to be determined currently.