Method of bandwidth extension for narrow-band speech

• US 6,988,066 B2
• Filed: 10/04/2001
• Issued: 01/17/2006
• Est. Priority Date: 10/04/2001
• Status: Expired due to Fees

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: AT&T Corporation (AT&T, Inc.)
• Inventors: Malah, David
• Primary Examiner(s): Ometz, David L.
• Assistant Examiner(s): JACKSON, JAKIEDA R
• Application Number:
  US09/970,743
• Publication Number:
  US 20030093278A1
• Time in Patent Office: 1,566 Days
• Field of Search: 704/219, 704/223, 704/220, 704/200
• US Class Current: 704/219
• CPC Class Codes: G10L 21/038 using band spreading techniques

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 M_nb area coefficients from the partial correlation coefficient, and extracting M_wb area coefficients using interpolation. Wideband parcors are computed from the M_wb 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 M_nb area coefficients are converted to log-area coefficients for the purpose of extracting, through shifted-interpolation, M_wb log-area coefficients. The M_wb log-area coefficients are then converted to M_wb area coefficients before generating the wideband parcors.

Claims

1. A method of producing a wideband signal from a narrowband signal, the method comprising:

computing M_nb area coefficients from the narrowband signal;

interpolating the M_nb area coefficients into M_wb area coefficients;

generating a highband signal using the M_wb area coefficients; and

combining the highband signal with the narrowband signal interpolated to the highband sampling rate to form the wideband signal.

2. The method of claim 1, wherein computing M_nb area coefficients further comprises computing M_nb area coefficient using the following equation:

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},1,$

where A₁ corresponds to a cross-section at the lips, A_Mnb+1 correspond to cross-sections of the vocal tract at the glottis opening and r_i are reflection coefficients.

3. The method of claim 1, wherein interpolating the M_nb area coefficients into M_wb area coefficients further comprises interpolating using a linear first order polynomial interpolation scheme.

4. The method of claim 1, wherein interpolating the M_nb area coefficients further comprises interpolating using a cubic spline interpolation scheme.

5. The method of claim 1, wherein interpolating the M_nb area coefficients further comprises interpolating using a fractal interpolation scheme.

6. The method of claim 1, further comprising:

insuring that the interpolated M_wb area coefficients are positive; and

setting A_Mwb+1^wb to a finite positive fixed value.

7. The method of claim 1, wherein interpolating the M_nb area coefficients further comprises interpolating by a factor of 2, with a ¼

sampling interval shift.

8. A method of bandwidth extension of a narrowband signal, the method comprising:

computing M_nb log-area coefficients from the narrowband signal;

interpolating the M_nb log-area coefficients into M_wb log-area coefficients;

generating a highband signal using the interpolated M_wb log-area coefficients; and

combining the highband signal with the narrowband signal interpolated to the highband sampling rate to generate a wideband signal.

9. The method of claim 8, wherein computing M_nb log-area coefficients further comprises computing M_nb area coefficients using the equation below and computing their logarithmic values:

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},1,$

where A₁ corresponds to a cross-section at the lips, A_Mnb+1 correspond to cross-sections of the vocal tract at the glottis opening and r_i are reflection coefficients.

10. The method of claim 8, wherein interpolating the M_nb log-area coefficients further comprises interpolating using a linear first order polynomial interpolation scheme.

11. The method of claim 8, wherein interpolating the M_nb log-area coefficients further comprises interpolating using a cubic spline interpolation scheme.

12. The method of claim 8, wherein interpolating the M_nb log-area coefficients further comprises interpolating using a fractal interpolation scheme.

13. The method of claim 8, wherein interpolating the M_nb log-area coefficients further comprises interpolating by a factor of 2, with a ¼

sample shift.

14. A method of extending the bandwidth of a narrowband signal, a preprocessing of the narrowband signal producing narrowband partial correlation coefficients (parcors), the method comprising:

(1) computing M_nb area coefficients from the narrowband parcors;

(2) computing M_nb log-area coefficients from the M_nb area coefficients;

(3) obtaining M_wb log-area coefficients from the M_nb log-area coefficients;

(4) computing M_wb area coefficients from the M_wb log-area coefficients;

(5) computing wideband parcors from the M_wb area coefficients;

(6) generating a highband signal using the wideband parcors; and

(7) combining the highband signal with the narrowband signal interpolated to the highband sampling rate.

15. The method of extending the bandwidth of a narrowband signal of claim 14, wherein obtaining M_wb log-area coefficients further comprises obtaining M_nb times two log-area coefficients using interpolation.

16. A method of producing a wideband signal from a narrowband signal, the method comprising:

(1) computing narrowband linear predictive coefficients (LPCs) from the narrowband signal;

(2) computing narrowband parcors r_i associated with the narrowband LPCs;

(3) computing M_nb area coefficients A_i^nb, i=1, 2, . . . , M_nb using the following;

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},1,$

where A₁ corresponds to a cross-section at lips, A_Mnb+1 and corresponds to a cross-section of a vocal tract at a glottis opening;

(4) extracting M_wb area coefficients from the M_nb area coefficients using interpolation;

(5) computing wideband parcors using the M_wb area coefficients according to the following;

$r_i^{\mathrm{wb}}=\frac{A_i^{\mathrm{wb}}-A_{i+1}^{\mathrm{wb}}}{A_i^{\mathrm{wb}}+A_{i+1}^{\mathrm{wb}}},i=1,2,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},M_{\mathrm{wb}};<br><br>$

(6) computing wideband LPCs a_i^wb, i=1, 2, . . . , M_wb, from the wideband parcors; and

(7) synthesizing a wideband signal y_wb using the wideband LPCs and an excitation signal.

17. The method of producing a wideband signal from a narrowband signal of claim 16, the method further comprising:

(8) highpass filtering the wideband signal y_wb to generate a highband signal; and

(9) combining the highband signal with the narrowband signal interpolated to the wideband sampling rate to produce a wideband signal Ŝ

_wb.

18. The method of producing a wideband signal from a narrowband signal of claim 16, wherein extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation further comprises interpolating by a factor of 4 followed by a single sample shift and decimating by a factor of 2.

19. The method of producing a wideband signal from a narrowband signal of claim 16, the method further comprising:

(8) generating the excitation signal from a narrowband prediction residual signal using fullwave rectification.

20. The method of producing a wideband signal from a narrowband signal of claim 16, wherein M_wb equals two times M_nb.

21. The method of producing a wideband signal from a narrowband signal of claim 16, wherein extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation further comprises interpolating by a factor of 2 with a ¼

sample shift.

22. The method of producing a wideband signal from a narrowband signal of claim 16, wherein extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation further comprises using a first order linear shifted-interpolation.

23. The method of producing a wideband signal from a narrowband signal of claim 16, wherein extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation further comprises using cubic-spline interpolation.

24. The method of producing a wideband signal from a narrowband signal of claim 16, wherein extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation further comprises using fractal interpolation.

25. A method of extending the bandwidth of a narrowband signal, the method comprising:

(1) computing narrowband linear predictive coefficients (LPCs) from the narrowband signal;

(2) computing narrowband parcors associated with the narrowband LPCs;

(3) computing M_nb area coefficients using the narrowband parcors;

(4) extracting M_wb area coefficients from the M_nb area coefficients using shifted-interpolation;

(5) converting the M_wb area coefficients into wideband LPCs; and

(6) synthesizing a wideband signal y_wb using the wideband LPCs and an excitation signal.

26. The method of extending the bandwidth of a narrowband signal of claim 25, the method further comprising:

(7) highpass filtering the wideband signal y_wb to produce a highband signal; and

(8) combining the highband signal with the narrowband signal interpolated to the wideband sampling rate to produce a wideband signal Ŝ

_wb.

27. The method of extending the bandwidth of a narrowband signal of claim 25, wherein the step of converting the M_wb area coefficients into wideband LPCs further comprising computing wideband parcors from the M_wb area coefficients and using step-down back-recursion to compute the wideband LPCs.

28. The method of extending the bandwidth of a narrowband signal of claim 25, the method further comprising computing the excitation signal from a narrowband prediction residual signal.

29. The method of extending the bandwidth of a narrowband signal of claim 25, wherein the higher band of the excitation signal is highpass filtered white noise.

30. A method of extending the bandwidth of a narrowband signal, the method comprising:

(1) computing narrowband linear predictive coefficients (LPCs) from the narrowband signal;

(2) computing M_nb area coefficients using the narrowband LPCs;

(3) extracting M_wb area coefficients from the M_nb area coefficients using interpolation;

(4) converting the M_wb area coefficients into wideband LPCs; and

(5) synthesizing a wideband signal y_wb using the wideband LPCs and highpass filtered white noise in the higher band of an excitation signal and a linear prediction residual signal in the lower band of the excitation signal.

31. The method of extending the bandwidth of a narrowband signal of claim 30, wherein computing the excitation signal from a narrowband prediction residual signal further comprises inverse filtering the narrowband signal.

32. A method of producing a wideband signal from a narrowband signal, the method comprising:

(1) producing a wideband excitation signal from the narrowband signal;

(2) computing partial correlation coefficients r_i (parcors) from the narrowband signal;

(3) computing M_nb area coefficients according to the following equation;

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},1,$

where A₁ corresponds to the cross-section at lips and A_Mnb+1 corresponds to the cross-section at a glottis opening;

(4) extracting M_wb area coefficients from the M_nb area coefficients using interpolation;

(5) computing wideband parcors r_i^wb from the interpolated M_wb area coefficients according to the following;

$r_i^{\mathrm{wb}}=\frac{A_i^{\mathrm{wb}}-A_{i+1}^{\mathrm{wb}}}{A_i^{\mathrm{wb}}+A_{i+1}^{\mathrm{wb}}},i=1,2,\mathrm{\dots <br><br>}<br><br>\text{\hspace{1em}<br><br>},M_{\mathrm{wb}};<br><br>$

(6) computing wideband linear predictive coefficients (LPCs) a_i^wb from the wideband parcors r_i^wb;

(7) synthesizing a wideband signal y_wb from the wideband LPCs a_i^wb and the wideband excitation signal;

(8) highpass filtering the wideband signal y_wb to produce a highband signal; and

(9) generating a wideband signal Ŝ

_wb by summing the highband signal and the narrowband signal interpolated to the wideband sampling rate.

33. The method of producing a wideband signal from a narrowband signal of claim 32, wherein producing the wideband excitation signal from the narrowband signal further comprises:

performing linear prediction on the narrowband signal to find a_i^wb LP coefficients;

interpolating the narrowband signal to produce an upsampled narrowband signal;

producing a narrowband residual signal {tilde over (r)}_nb by inverse filtering the upsampled interpolated narrowband signal using a transfer function associated with the a_i^wb LP coefficients; and

generating the wideband excitation signal from the narrowband residual signal {tilde over (r)}_nb.

34. A method of producing a wideband signal from a narrowband signal, the method receiving data associated with a narrowband signal, the method comprising:

(1) computing M_nb area coefficients using the narrowband data;

(2) extracting M_wb area coefficients from the M_nb area coefficients using interpolation; and

(3) synthesizing a wideband signal y_wb using wideband coefficients processed from data associated with the M_nb area coefficients and an excitation signal.

35. The method of producing a wideband signal from a narrowband signal of claim 34, the method further comprising:

(4) highpass filtering the wideband signal y_wb to form a highband signal; and

(5) generating a wideband signal Ŝ

_wb by summing the highband signal and the narrowband signal interpolated to the wideband sampling rate.

36. A method of producing a wideband signal from a narrowband signal, the method comprising:

(1) computing M_nb area coefficients from the narrowband signal;

(2) computing M_nb log-area coefficients from the M_nb area coefficients;

(3) interpolating the M_nb log-area coefficients into M_wb log-area coefficients;

(4) converting the M_wb log-area coefficients into M_wb area coefficients; and

(5) synthesizing a wideband signal y_wb using the M_wb area coefficients and an excitation signal.

37. The method of producing a wideband signal from a narrowband signal of claim 36, the method further comprising:

(6) highpass filtering the wideband signal y_wb to produce a highband signal; and

(7) combining the highband signal with the narrowband signal interpolated to the wideband sampling rate to generate a wideband signal Ŝ

_wb.

38. The method of claim 36, wherein computing M_nb area coefficients further comprises computing M_nb area coefficients using the following equation:

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>\text{\hspace{1em}<br><br>}<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>},1,$

where A₁ corresponds to a cross-section at the lips, A_Mnb+1 corresponds to a cross-section at the glottis opening and r_i are reflection coefficients.

39. The method of claim 36, wherein interpolating the M_nb log-area coefficients into M_wb log-area coefficients further comprises interpolating using a linear first order polynomial interpolation scheme.

40. The method of claim 36, wherein interpolating the M_nb log-area coefficients further comprises interpolating using a cubic spline interpolation scheme.

41. The method of claim 36, wherein interpolating the M_nb log-area coefficients further comprises interpolating using a fractal interpolation scheme.

42. The method of claim 36, wherein interpolating the M_nb log-area coefficients further comprises interpolating by a factor of 2, with a ¼

sample shift.

43. The method of claim 36, wherein interpolating the M_nb log-area coefficients further comprises interpolating by a factor of 4 followed by a single sample shift and decimating by a factor of 2.

44. A method of generating a wideband signal from a narrowband signal, the method comprising:

(1) producing a wideband excitation signal from the narrowband signal;

(2) computing partial correlation coefficients r_i (parcors) from the narrowband signal;

(3) computing M_nb area coefficients according to the following equation;

$A_i=\frac{1+r_i}{1-r_i}<br><br>A_{i+1};<br><br>\text{\hspace{1em}<br><br>}<br><br>i=M_{\mathrm{nb}},M_{\mathrm{nb}}-1,\mathrm{\dots <br><br>},1,$

where A₁ corresponds to the cross-section at lips and A_Mnb+1 corresponds to the cross-section at a glottis opening;

(4) computing M_nb log-area coefficients by applying a log operator to the M_nb area coefficients;

(5) extracting M_wb log-area coefficients from the M_nb log-area coefficients using shifted-interpolation;

(6) converting the M_wb log-area coefficients into M_wb area coefficients;

(7) computing wideband parcors r_i^wb from the M_wb area coefficients according to the following;

$r_i^{\mathrm{wb}}=\frac{A_i^{\mathrm{wb}}-A_{i+1}^{\mathrm{wb}}}{A_i^{\mathrm{wb}}+A_{i+1}^{\mathrm{wb}}},\text{\hspace{1em}<br><br>}<br><br>i=1,2,\mathrm{\dots <br><br>},M_{\mathrm{wb}};<br><br>$

(8) computing wideband linear predictive coefficients (LPCs) a_i^wb from the wideband parcors r_i^wb; and

(9) synthesizing a wideband signal y_wb from the wideband LPCs a_i^wb and the wideband excitation signal.

45. The method of generating an output wideband signal from a narrowband signal of claim 44, the method further comprising:

(10) highpass filtering the wideband signal y_wb to generate a highband signal S_hb; and

(11) generating a wideband signal Ŝ

_wb by summing the highband signal S_hb and the narrowband signal interpolated to the wideband sampling rate.

46. The method of generating a wideband signal from a narrowband signal of claim 44, wherein producing a wideband excitation signal from the narrowband signal further comprises:

performing linear prediction on the narrowband signal to find a_i^wb LP coefficients;

interpolating the narrowband signal to produce an upsampled interpolated narrowband signal;

producing a narrowband residual signal {tilde over (r)}_nb by inverse filtering the upsampled interpolated narrowband signal using a transfer function associated with the a_i^wb LP coefficients; and

generating a wideband excitation signal from the narrowband residual signal {tilde over (r)}_nb.

47. A method of producing a wideband signal from a narrowband signal, the method comprising:

computing M_nb area coefficients from the narrowband signal;

interpolating the M_nb area coefficients into M_wb area coefficients; and

generating the wideband signal using the M_wb area coefficients.

48. The method of generating a wideband signal from a narrowband signal of claim 47, wherein interpolating the M_nb area coefficients further comprises interpolating by a factor of 4 followed by a single sampling interval shift and decimating by a factor of 2.

49. A method of producing a wideband signal from a narrowband signal, the method comprising:

computing M_nb log-area coefficients by applying a log operator to M_nb area coefficients generated from the narrowband signal;

extracting M_wb log-area coefficients from the M_nb log-area coefficients using interpolation; and

generating a wideband signal using M_wb area coefficients generated from the M_wb log-area coefficients.

50. The method of generating a wideband signal from a narrowband signal of claim 49, wherein extracting the M_nb log-area coefficients using interpolation further comprises interpolating by a factor of 4 followed by a single sampling interval shift and decimating by a factor of 2.