US 6,895,375 B2 · System for bandwidth extension of Narrow-band speech

Abstract

A system and method are disclosed for extending the bandwidth of a narrowband signal such as a speech signal. The method applies a parametric approach to bandwidth extension but does not require training. The parametric representation relates to a discrete acoustic tube model (DATM). The method comprises computing narrowband linear predictive coefficients (LPCs) from a received narrowband speech signal, computing narrowband partial correlation coefficients (parcors) using recursion, computing Mnb area coefficients from the partial correlation coefficient, and extracting Mwb area coefficients using interpolation. Wideband parcors are computed from the Mwb area coefficients and wideband LPCs are computed from the wideband parcors. The method further comprises synthesizing a wideband signal using the wideband LPCs and a wideband excitation signal, highpass filtering the synthesized wideband signal to produce a highband signal, and combining the highband signal with the original narrowband signal to generate a wideband signal. In a preferred variation of the invention, the Mnb area coefficients are converted to log-area coefficients for the purpose of extracting, through shifted-interpolation, Mwb log-area coefficients. The Mwb log-area coefficients are then converted to Mwb area coefficients before generating the wideband parcors.

Claims94

Dependent claims

A system for generating a wideband signal from a narrowband signal, the system comprising:

a linear predictive coefficient module that computes narrowband coefficients;
an area coefficient module that computes area coefficients using the narrowband coefficients;
an area shifted-interpolation module the performs a shifted-interpolation of the area coefficients;
and
a module that transforms the shift-interpolated area coefficients into wideband linear predictive coefficients used for generating a wideband signal ywb.

2 more elements

A system for generating a wideband signal from a narrowband signal, the system comprising:

(1) a module for processing the narrowband signal comprising:
(a) a signal interpolation module producing an interpolated narrowband signal;
(b) an inverse filter that filters the interpolated narrowband signal;
and
(c) a nonlinear operation module that generates an excitation signal from the filtered interpolated narrowband signal;
(2) a module for producing wideband coefficients comprising:
(a) a linear predictive analysis module that produces Mnb narrowband coefficients associated with the narrowband signal;
(b) an area parameter module that computes area parameters using the Mnb narrowband coefficients;
(c) a shifted-interpolation module that computes shift-interpolated area parameters from the area parameters;
and
(d) a module that computes Mwb wideband coefficients from the shift-interpolated area parameters;
and
(3) a synthesis module that receives the Mwb wideband coefficients and the excitation signal to synthesize a wideband signal ywb.

10 more elements

A system for generating a wideband signal from a narrowband signal, the system comprising:

(1) a narrowband signal module that produces an interpolated narrowband signal at a wideband sampling rate and produces a wideband excitation signal from the narrowband signal;
and
(2) a wideband signal module that generates a wideband signal ywb according to a method comprising:
(a) computing partial correlation coefficients ri (parcors) from the narrowband signal;
(b) computing Mnb area coefficients according to the following equation:
A
i
=
1
+
r
i
1
-
r
i
⁢
A
i
+
1
;
⁢
i
=
M
nb
,
M
nb
-
1
,
…
,
1
,
where A1 corresponds to the cross-section at lips and AMnb+corresponds to the cross-section at a glottis opening;
(c) computing Mnb log-area coefficients by applying a natural-log operator to the Mnb area coefficients;
(d) extracting Mwb log-area coefficients from the Mnb log-area coefficients using shifted-interpolation;
(e) converting the Mwb log-area coefficients into Mwb area coefficients;
(f) computing wideband parcors riwb from the Mwb area coefficients according to the following:
r
i
wb
=
A
i
wb
-
A
i
+
1
wb
A
i
wb
+
A
i
+
1
wb
,
⁢
i
=
1
,
2
,
…
,
M
wb
;
(g) computing wideband linear predictive coefficients (LPCs) aiwb from the wideband parcors riwb;
and
(h) synthesizing a wideband signal ywb from the wideband LPCs aiwb and the wideband excitation signal;
(i) highpass filtering the wideband signal ywb to form a highband signal Shb;
and
(j) generating a wideband signal ŝwb by summing the highband signal Shb and the interpolated narrowband signal.

80 more elements

A computer-readable storage medium storing a method for controlling a computer device to produce a wideband signal from a narrowband signal, the method comprising:

(1) generating a wideband excitation signal from the narrowband signal;
(2) computing Mnb area coefficients from the narrowband signal;
(3) extracting Mwb area coefficients from the Mnb area coefficients using interpolation;
(4) computing wideband linear predictive coefficients (LPCs) using the Mwb area coefficients;
and
(5) synthesizing a wideband signal ywb from the wideband LPCs aiwb and the wideband excitation signal.

3 more elements

A computer-readable storage medium storing instructions for controlling a computer device to produce a wideband signal from a narrowband signal, the instructions controlling the computer device to perform the steps of:

(1) producing a wideband excitation signal from the narrowband signal;
(2) computing partial correlation coefficients ri (parcors) from the narrowband signal;
(3) computing Mnb area coefficients according to the following equation:
A
i
=
1
+
r
i
1
-
r
i
⁢
A
i
+
1
;
⁢
i
=
M
nb
,
M
nb
-
1
,
…
,
1
,
where A1 corresponds to the cross-section at lips and AMnb+1 corresponds to the cross-section at a glottis opening;
(4) computing Mnb log-area coefficients by applying a natural-log operator to the Mnb area coefficients;
(5) extracting Mwb log-area coefficients from the Mnb log-area coefficients using shifted-interpolation;
(6) converting the Mwb log-area coefficients into Mwb area coefficients;
(7) computing wideband parcors riwb from the Mwb area coefficients according to the following:
r
i
wb
=
A
i
wb
-
A
i
+
1
wb
A
i
wb
+
A
i
+
1
wb
,
⁢
i
=
1
,
2
,
…
,
M
wb
;
(8) computing wideband linear predictive coefficients (LPCs) aiwb from the wideband parcors riwb;
and
(9) synthesizing a wideband signal ywb from the wideband LPCs aiwb and the wideband excitation signal.

75 more elements

A computer-readable storage medium storing a method for controlling a computer device to produce a wideband signal from a narrowband signal, the method comprising:

(1) computing partial correlation coefficients ri (parcors) from the narrowband signal;
(2) computing Mnb area coefficients according to the following equation:
A
i
=
1
+
r
i
1
-
r
i
⁢
A
i
+
1
;
⁢
i
=
M
nb
,
M
nb
-
1
,
…
,
1
,
where A1 corresponds to the cross-section at lips and AMnb+1 corresponds to the cross-section at a glottis opening;
(3) computing Mnb log-area coefficients;
(4) extracting Mwb log-area coefficients from the Mnb log-area coefficients using shifted-interpolation;
(5) computing Mwb area coefficients from the Mnb log-area coefficients;
(6) computing wideband parcors riwb from the Mwb area coefficients according to the following:
r
i
wb
=
A
i
wb
-
A
i
+
1
wb
A
i
wb
+
A
i
+
1
wb
,
⁢
i
=
1
,
2
,
…
,
M
wb
;
(7) computing wideband linear predictive coefficients (LPCs) aiwb from the wideband parcors riwb;
and
(8) synthesizing a wideband signal ywb from the wideband LPCs and an excitation signal.

74 more elements

Dependencies are inferred from the claim text.

System for bandwidth extension of Narrow-band speech

Life of this patent

System for bandwidth extension of Narrow-band speech

Life of this patent

Abstract

Claims94