Linear predictive analysis apparatus, method, program and recording medium

Patent metadata

Field	Value
Publication number	US-10134420-B2
Application number	US-201815889775-A
Country	US
Kind code	B2
Filing date	Feb 6, 2018
Priority date	Jan 24, 2014
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

An autocorrelation calculating part calculates autocorrelation Ro(i) from an input signal. A predictive coefficient calculating part performs linear predictive analysis using modified autocorrelation R′o(i) obtained by multiplying the autocorrelation Ro(i) by a coefficient wo(i). Here, it is assumed that a case where, for at least part of each order i, the coefficient wo(i) corresponding to each order i monotonically increases as a value having negative correlation with a fundamental frequency of an input signal in a current frame or a past frame increases and a case where the coefficient wo(i) monotonically decreases as a value having positive correlation with a pitch gain in a current frame or a past frame increases, are included.

First claim

What is claimed is:

1. A linear predictive analysis method for obtaining a coefficient which can be converted into a linear predictive coefficient corresponding to an input time series signal for each frame which is a predetermined time interval, the linear predictive analysis method comprising:
an autocorrelation calculating step of calculating autocorrelation R o (i) between an input time series signal X o (n) of a current frame and an input time series signal X o (n−i) i sample before the input time series signal X o (n) or an input time series signal X o (n+i) i sample after the input time series signal X o (n) for each of at least i=0, 1, . . . , P max ; and
a predictive coefficient calculating step of obtaining a coefficient which can be converted into linear predictive coefficients from the first-order to the P max -order using modified autocorrelation R′ o (i) obtained by multiplying the autocorrelation R o (i) by a coefficient for each corresponding i,
wherein the linear predictive analysis method further comprises a coefficient determining step of acquiring the coefficient from one coefficient table among coefficient tables t0, t1 and t2 using a period, an estimate value of the period, a quantization value of the period or a value having negative correlation with a fundamental frequency based on an input time series signal in the current frame or a past frame and a value having positive correlation with intensity of periodicity or a pitch gain assuming that a coefficient w t0 (i) is stored in the coefficient table t0, a coefficient w t1 (i) is stored in the coefficient table t1, and a coefficient w t2 (i) is stored in the coefficient table t2,
for at least part of i other than i=0, w t0 (i)<w t1 (i)≤w t2 (i), for at least part of each i among other i other than i=0, w t0 (i)≤w t1 (i)<w t2 (i), and for the remaining each i other than i=0, w t0 (i)≤w t1 (i)≤w t2 (i),
according to the period, the estimate value of the period, the quantization value of the period or the value having negative correlation with the fundamental frequency and the value having positive correlation with the intensity of periodicity or the pitch gain, (1) when the period is short and the pitch gain is large, a coefficient is acquired from the coefficient table t0 in the coefficient determining step, (9) when the period is long and the pitch gain is small, a coefficient is acquired from the coefficient table t2 in the coefficient determining step, (2) when the period is short and the pitch gain is medium, (3) when the period is short and the pitch gain is small, (4) when the period is medium and the pitch gain is large, (5) when the period is medium and the pitch gain is medium, (6) when the period is medium and the pitch gain is small, (7) when the period is long and the pitch gain is large, and (8) when the period is long and the pitch gain is medium, a coefficient is acquired from any of the coefficient tables t0, t1 and t2 in the coefficient determining step,
in at least one of (2), (3), (4), (5), (6), (7) and (8), a coefficient is acquired from the coefficient table t1 in the coefficient determining step, and
assuming that an identification number of a coefficient table tj k from which a coefficient is acquired in the coefficient determining step in the case of (k) where k=1, 2, . . . , 9 is j k , j 1 ≤j 2 ≤j 3 , j 4 ≤j 5 ≤j 6 , j 7 ≤j 8 ≤j 9 , j 1 ≤j 4 ≤j 7 , j 2 ≤j 5 ≤j 8 , and j 3 ≤j 6 ≤j 9 .

2. A linear predictive analysis method for obtaining a coefficient which can be converted into a linear predictive coefficient corresponding to an input time series signal for each frame which is a predetermined time interval, the linear predictive analysis method comprising:
an autocorrelation calculating step of calculating autocorrelation R o (i) between an input time series signal X o (n) of a current frame and an input time series signal X o (n−i) i sample before the input time series signal X o (n) or an input time series signal X o (n+i) i sample after the input time series signal X o (n) for each of at least i=0, 1, . . . , P max ; and
a predictive coefficient calculating step of obtaining a coefficient which can be converted into linear predictive coefficients from the first-order to the P max -order using modified autocorrelation R′ o (i) obtained by multiplying the autocorrelation R o (i) by a coefficient for each corresponding i,
wherein the linear predictive analysis method further comprises a coefficient determining step of acquiring the coefficient from one coefficient table among coefficient tables t0, t1 and t2 using a value having positive correlation with a fundamental frequency based on an input time series signal in the current frame or a past frame and a value having positive correlation with intensity of periodicity or a pitch gain assuming that a coefficient w t0 (i) is stored in the coefficient table t0, a coefficient w t1 (i) is stored in the coefficient table t1, and a coefficient w t2 (i) is stored in the coefficient table t2,
for at least part of i other than i=0, w t0 (i)<w t1 (i)≤w t2 (i), for at least part of each i among other i other than i=0, w t0 (i)≤w t1 (i)<w t2 (i), and for the remaining each i other than i=0, w t0 (i)≤w t1 (i)≤w t2 (i),
according to the value having positive correlation with the fundamental frequency and the value having positive correlation with the intensity of periodicity or the pitch gain, (1) when the fundamental frequency is high and the pitch gain is large, a coefficient is acquired from the coefficient table t0 in the coefficient determining step, (9) when the fundamental frequency is low and the pitch gain is small, a coefficient is acquired from the coefficient table t2 in the coefficient determining step, (2) when the fundamental frequency is high and the pitch gain is medium, (3) when the fundamental frequency is high and the pitch gain is small, (4) when the fundamental frequency is medium and the pitch gain is large, (5) when the fundamental frequency is medium and the pitch gain is medium, (6) when the fundamental frequency is medium and the pitch gain is small, (7) when the fundamental frequency is low and the pitch gain is large, and (8) when the fundamental frequency is low and the pitch gain is medium, a coefficient is acquired from any of the coefficient tables t0, t1 and t2 in the coefficient determining step,
in at least one of (2), (3), (4), (5), (6), (7) and (8), a coefficient is acquired from the coefficient table t1 in the coefficient determining step, and
assuming that an identification number of a coefficient table tj k from which a coefficient is acquired in the coefficient determining step in the case of (k) where k=1, 2, . . . , 9 is j k , j 1 ≤j 2 ≤j 3 , j 4 ≤j 5 ≤j 6 , j 7 ≤j 8 ≤j 9 , j 1 ≤j 4 ≤j 7 , j 2 ≤j 5 ≤j 8 , and j 3 ≤j 6 ≤j 9 .

3. A linear predictive analysis apparatus which obtains a coefficient which can be converted into a linear predictive coefficient corresponding to an input time series signal for each frame which is a predetermined time interval, the linear predictive analysis apparatus comprising:
processing circuitry configured to
calculate autocorrelation R o (i) between an input time series signal X o (n) of a current frame and an input time series signal X o (n−i) i sample before the input time series signal X o (n) or an input time series signal X o (n+i) i sample after the input time series signal X o (n) for each of at least i=0, 1, . . . , P max ; and
obtain a coefficient which can be converted into linear predictive coefficients from the first-order to the P max -order using modified autocorrelation R′ o (i) obtained by multiplying the autocorrelation R o (i) by a coefficient for each corresponding i,
wherein the processing circuitry further configured to acquire the coefficient from one coefficient table among coefficient tables t0, t1 and t2 using a period, an estima

Assignees

• Nippon Telegraph & Telephone

Inventors

• Kamamoto Yutaka
• Moriya Takehiro
• Harada Noboru

Classifications

• G10L19/06 Primary

  Determination or coding of the spectral characteristics, e.g. of the short-term prediction coefficients · CPC title

• G10L25/12 Primary

  the extracted parameters being prediction coefficients · CPC title

• G10L25/21

  the extracted parameters being power information · CPC title

• G10L25/06

  the extracted parameters being correlation coefficients · CPC title

• G10L25/90

  Pitch determination of speech signals · CPC title