METHOD AND APPARATUS FOR EXTENDING THE BANDWIDTH OF A SPEECH SIGNAL

• US 20070067163A1
• Filed: 09/01/2006
• Published: 03/22/2007
• Est. Priority Date: 09/02/2005
• Status: Active Grant

• Current Assignee: Apple Inc.
• Original Assignee: Nortel Networks Limited (Nortel Networks Corporation)
• Inventors: Rabipour, Rafi, Qian, Yasheng, Kabal, Peter
• Granted Patent: US 7,734,462 B2
• Time in Patent Office: Days
• US Class Current: 704/207
• CPC Class Codes: G10L 21/038 using band spreading techniques
• CPC Class Codes: G10L 25/18 the extracted parameters being spectral information of each sub-band

Abstract

A bandwidth extension module, and an associated method and computer-readable medium, suitable for use in artificially extending the bandwidth of a lowband speech signal. The bandwidth extension module comprises a band-pass filter configured to produce a band-pass signal from the lowband speech signal; at least one carrier frequency modulator, each carrier frequency modulator configured to pitch-synchronously modulate the band-pass signal about a respective carrier frequency, the at least one carrier frequency modulator collectively producing a highband speech signal component; a synthesis filter configured to determine a highband speech signal based on the highband speech signal component; and a summation module configured to combine the lowband speech signal with the highband speech signal to obtain a bandwidth-extended speech signal.

Claims

1. A method of artificially extending the bandwidth of a lowband speech signal, comprising:

band-pass filtering the lowband speech signal to obtain a band-pass signal;

pitch-synchronously modulating said band-pass signal about at least one carrier frequency to obtain a highband speech signal component;

determining a highband speech signal based on said highband speech signal component;

combining said lowband speech signal with said highband speech signal to obtain a bandwidth-extended speech signal.

2. The method defined in claim 1, further comprising:

detecting a pitch of said lowband speech signal.

3. The method defined in claim 2, further comprising:

using a pitch estimation module to detect said pitch.

4. The method defined in claim 2, wherein said step of band-pass filtering comprises utilizing a band-pass filter having a passband.

5. The method defined in claim 4, further comprising:

determining each of the at least one said carrier frequency on the basis of (i) said pitch and (ii) said passband of said band-pass filter.

6. The method defined in claim 5, wherein the at least one carrier frequency includes a plurality of carrier frequencies.

7. The method defined in claim 6, wherein pitch-synchronously modulating said band-pass signal about the at least one carrier frequency to obtain said highband speech signal component comprises pitch-synchronously modulating said band-pass signal about each of said carrier frequencies in said plurality of carrier frequencies, and combining the results to obtain said highband speech signal component.

8. The method defined in claim 7, wherein said plurality of carrier frequencies includes three carrier frequencies.

9. The method defined in claim 6, wherein each of said plurality of carrier frequencies is the sum of a respective nominal carrier frequency and a respective correction factor.

10. The method defined in claim 9, wherein said passband of said band-pass filter is between approximately 3000 Hz and approximately 4000 Hz.

11. The method defined in claim 10, wherein a first said nominal carrier frequency is approximately 4500 Hz, and wherein a second said nominal carrier frequency is approximately 5500 Hz.

12. The method defined in claim 11, wherein a third said nominal carrier frequency is approximately 6500 Hz.

13. The method defined in claim 1, further comprising:

prior to said pitch-synchronously modulating, inverse filtering said band-pass signal to flatten a spectrum of said band-pass signal.

14. The method defined in claim 1, wherein said highband speech signal component comprises an excitation signal.

15. The method defined in claim 14, further comprising:

multiplying said excitation signal by an excitation gain to obtain a scaled excitation signal.

16. The method defined in claim 15, further comprising:

determining said excitation gain based on said pitch and on a set of lowband linear spectral frequencies.

17. The method defined in claim 15, wherein said determining a highband speech signal based on said highband speech signal component comprises synthesizing said highband speech signal based on said scaled excitation signal and a set of highband linear spectral frequencies.

18. The method defined in claim 17, further comprising:

determining said highband linear spectral frequencies based on said pitch and on a set of lowband linear spectral frequencies.

19. The method defined in claim 18, further comprising:

determining said lowband linear spectral frequencies based on said lowband speech signal.

20. The method defined in claim 19, further comprising:

prior to said pitch-synchronously modulating, inverse filtering said band-pass signal to compensate for amplitude variations in a spectrum of said band-pass signal, said amplitude variations being characterized by said lowband linear spectral frequencies.

21. The method defined in claim 20, wherein said combining said lowband speech signal with said highband speech signal to obtain a bandwidth-extended speech signal comprises combining said highband speech signal with a delayed version of said lowband speech signal to obtain said bandwidth-extended speech signal.

22. The method defined in claim 1, further comprising:

pre-filtering an original speech signal to obtain said lowband speech signal, said pre-filtering causing partial extension of a frequency spectrum of said original speech signal into an intermediate frequency band.

23. The method defined in claim 22, wherein said pre-filtering comprises upsampling, low-pass filtering and spectral shaping.

24. The method defined in claim 23, wherein said intermediate frequency band extends from approximately 3400 Hz to approximately 4000 Hz.

25. The method defined in claim 22, wherein said original speech signal has no component above 3400 Hz that is not significantly attenuated and wherein said lowband speech signal has no component above 4000 Hz that is not significantly attenuated.

26. The method defined in claim 1, further comprising:

classifying said lowband speech signal as belonging to a strong harmonic mode, an unvoiced mode or a mixed mode.

27. The method defined in claim 26, wherein pitch-synchronously modulating said band-pass signal about at least one carrier frequency to obtain said highband speech signal is only performed in response to said lowband speech signal being classified as belonging to said strong harmonic mode.

28. The method defined in claim 27, further comprising multiplying an output of a noise generator with an output of an envelope operator applied to said band-pass signal to obtain said highband speech signal component in response to said lowband speech signal being classified as belonging to said unvoiced mode or said mixed mode.

29. A bandwidth extension module suitable for use in artificially extending the bandwidth of a lowband speech signal, comprising:

means for band-pass filtering the lowband speech signal to obtain a band-pass signal;

means for pitch-synchronously modulating said band-pass signal about at least one carrier frequency to obtain a highband speech signal component;

means for determining a highband speech signal based on said highband speech signal component;

means for combining said lowband speech signal with said highband speech signal to obtain a bandwidth-extended speech signal.

30. A computer-readable medium comprising computer-readable program code which, when interpreted by a computing apparatus, causes the computing apparatus to execute a method of artificially extending the bandwidth of a lowband speech signal, the computer-readable program code comprising:

first computer-readable program code for causing the computing apparatus to obtain a band-pass signal by band-pass filtering the lowband speech signal;

second computer-readable program code for causing the computing apparatus to obtain a highband speech signal component by pitch-synchronously modulating said band-pass signal about at least one carrier frequency;

third computer-readable program code for causing the computing apparatus to determine a highband speech signal based on said highband speech signal component;

fourth computer-readable program code for causing the computing apparatus to obtain a bandwidth-extended speech signal by combining said lowband speech signal with said highband speech signal.

31. A bandwidth extension module suitable for use in artificially extending the bandwidth of a lowband speech signal, comprising:

a band-pass filter configured to produce a band-pass signal from the lowband speech signal;

at least one carrier frequency modulator, each said carrier frequency modulator configured to pitch-synchronously modulate said band-pass signal about a respective carrier frequency, the at least one carrier frequency modulator collectively producing a highband speech signal component;

a synthesis filter configured to determine a highband speech signal based on said highband speech signal component;

a summation module configured to combine said lowband speech signal with said highband speech signal to obtain a bandwidth-extended speech signal.

32. The bandwidth extension module defined in claim 31, implemented at one of (i) a central office;

(ii) a mobile switching center; and

(iii) digital switching equipment.

33. The bandwidth extension module defined in claim 31, implemented in an adapter for a wideband-capable telephony device.

34. The bandwidth extension module defined in claim 31, integrated with a wideband-capable telephony device.

35. The bandwidth extension module defined in claim 31, further comprising:

a pitch estimation module configured to detect a pitch of said lowband speech signal.

36. The bandwidth extension module defined in claim 35, wherein said band-pass filter has a passband, the bandwidth extension module further comprising:

a carrier frequency generator configured to determine each respective carrier frequency on the basis of (i) said pitch and (ii) said passband of said band-pass filter.

37. The bandwidth extension module defined in claim 36, wherein the at least one carrier frequency modulator includes a plurality of carrier frequency modulators.

38. The bandwidth extension module defined in claim 37, wherein each respective carrier frequency is the sum of a respective nominal carrier frequency and a respective correction factor.

39. The bandwidth extension module defined in claim 38, wherein said passband of said band-pass filter is between approximately 3000 Hz and approximately 4000 Hz.

40. The bandwidth extension module defined in claim 39, wherein a first respective nominal carrier frequency is approximately 4500 Hz, and wherein a second respective nominal carrier frequency is approximately 5500 Hz.

41. The bandwidth extension module defined in claim 40, wherein a third respective nominal carrier frequency is approximately 6500 Hz.

42. The bandwidth extension module defined in claim 31, further comprising:

an inverse filter connected between the band-pass filter and the at least one carrier frequency modulator, said inverse filter configured to flatten a spectrum of said band-pass signal.

43. The bandwidth extension module defined in claim 31, wherein said highband speech signal component comprises an excitation signal and wherein said bandwidth extension module further comprises:

a functional element configured to multiply said excitation signal by an excitation gain to obtain a scaled excitation signal, said excitation gain being determined based on said pitch and on a set of lowband linear spectral frequencies.

44. The bandwidth extension module defined in claim 43, wherein to determine said highband speech signal based on said highband speech signal component, said synthesis utilizes said scaled excitation signal and a set of highband linear spectral frequencies, said highband linear spectral frequencies being determined based on said pitch and on a set of lowband linear spectral frequencies.

45. The bandwidth extension module defined in claim 44, further comprising:

an estimation module configured to determine said highband linear spectral frequencies based on said pitch and on a set of lowband linear spectral frequencies.

46. The bandwidth extension module defined in claim 45, further comprising:

an estimation module configured to determine said lowband linear spectral frequencies based on said lowband speech signal.

47. The bandwidth extension module defined in claim 46, further comprising:

an inverse filter connected between the band-pass filter and the at least one carrier frequency modulator, said inverse filter configured to compensate for amplitude variations in a spectrum of said band-pass signal, said amplitude variations being characterized by said lowband linear spectral frequencies.

48. The bandwidth extension module defined in claim 47, further comprising:

a delay element configured to delay said lowband speech signal prior to combining by the summation module.

49. The bandwidth extension module defined in claim 31, further comprising:

a pre-emphasis module configured to process an original speech signal to obtain said lowband speech signal, thereby to cause partial extension of a frequency spectrum of said original speech signal into an intermediate frequency band.

50. The bandwidth extension module defined in claim 49, wherein said pre-emphasis module comprises an upsampler, a low-pass filter and a spectral shaping filter.

51. The bandwidth extension module defined in claim 50, wherein said intermediate frequency band extends from approximately 3400 Hz to approximately 4000 Hz.

52. The bandwidth extension module defined in claim 49, wherein said original speech signal has no component above 3400 Hz that is not significantly attenuated and wherein said lowband speech signal has no component above 4000 Hz that is not significantly attenuated.

53. The bandwidth extension module defined in claim 31, further comprising:

a classifier configured to classify said lowband speech signal as belonging to a strong harmonic mode, an unvoiced mode or a mixed mode;

a selector connected to said classifier, and configured to allow said highband speech signal component to be produced from the at least one carrier frequency modulator only in response to said lowband speech signal being classified as belonging to said strong harmonic mode.

54. The bandwidth extension module defined in claim 53, further comprising:

a noise generator producing an output;

an envelope operator processing said band-pass signal to produce an output;

said selector further configured to cause said highband speech signal component to be produced by multiplication of the output of the noise generator with the output of the envelope operator in response to said lowband speech signal being classified as belonging to said unvoiced mode or said mixed mode.

55. An excitation signal generator, comprising:

a bandpass filter configured to produce a band-pass signal from the lowband speech signal;

a modulator bank comprising a plurality of carrier frequency modulators, each of said carrier frequency modulators configured to frequency shift the band-pass signal to a respective carrier frequency associated with the respective carrier frequency modulator, thereby to produce a respective one of a plurality of modulated signals;

a summation module configured to combine the modulated signals into an excitation signal for use in generating a highband speech signal that complements the lowband speech signal in a highband frequency range;

the carrier frequency associated with a given one of the carrier frequency modulators being selected based on a pitch of the lowband speech signal to ensure pitch-synchronicity between the bandpass signal and the respective modulated signal produced by the given one of the carrier frequency modulators.

56. The excitation signal generator defined in claim 55, further comprising:

an inverse filter connected between the band-pass filter and the modulator bank, said inverse filter configured to flatten a spectrum of said band-pass signal.

57. The excitation signal generator defined in claim 56, wherein said bandwidth extension module is configured to receive a detected pitch of said lowband speech signal, wherein said band-pass filter has a passband, the bandwidth extension module further comprising:

a carrier frequency generator configured to determine each respective carrier frequency on the basis of (i) said pitch and (ii) said passband of said band-pass filter.

58. The excitation signal generator defined in claim 57, wherein each respective carrier frequency is the sum of a respective nominal carrier frequency and a respective correction factor.

59. The excitation signal generator defined in claim 58, wherein said passband of said band-pass filter is between approximately 3000 Hz and approximately 4000 Hz.

60. The excitation signal generator defined in claim 59, wherein a first respective nominal carrier frequency is approximately 4500 Hz, and wherein a second respective nominal carrier frequency is approximately 5500 Hz.

61. The excitation signal generator defined in claim 60, wherein a third respective nominal carrier frequency is approximately 6500 Hz.

62. The excitation signal generator defined in claim 55, further comprising:

an inverse filter connected between the band-pass filter and the modulator bank, said inverse filter configured to flatten a spectrum of said band-pass signal.

63. The excitation signal generator defined in claim 55, further comprising:

a pre-emphasis module configured to process an original speech signal to obtain said lowband speech signal, thereby to cause partial extension of a frequency spectrum of said original speech signal into an intermediate frequency band.

64. The excitation signal generator defined in claim 63, wherein said pre-emphasis module comprises an upsampler, a low-pass filter and a spectral shaping filter.

65. The excitation signal generator defined in claim 64, wherein said intermediate frequency band extends from approximately 3400 Hz to approximately 4000 Hz.

66. The excitation signal generator defined in claim 63, wherein said original speech signal has no component above 3400 Hz that is not significantly attenuated and wherein said lowband speech signal has no component above 4000 Hz that is not significantly attenuated.

67. The excitation signal generator defined in claim 55, further comprising:

a classifier configured to classify said lowband speech signal as belonging to a strong harmonic mode, an unvoiced mode or a mixed mode;

a selector connected to said classifier, and configured to allow said excitation signal to be produced from the modulated signals only in response to said lowband speech signal being classified as belonging to said strong harmonic mode.

68. The excitation signal generator defined in claim 67, further comprising

a noise generator producing an output;

an envelope operator processing said band-pass signal to produce an output;

said selector further configured to cause said excitation signal to be produced by multiplication of the output of the noise generator with the output of the envelope operator in response to said lowband speech signal being classified as belonging to said unvoiced mode or said mixed mode.

69. A bandwidth extension module, comprising:

an input for receiving a first speech signal having first frequency content in a first frequency range;

a processing entity; and

an output for producing a second speech signal having second frequency content in a second frequency range that includes an additional frequency range outside the first frequency range;

wherein when the first frequency content contains harmonics in the first frequency range obeying a harmonic relationship, said processing entity is configured to cause the second frequency content to contain harmonics in the first frequency range and in the additional frequency range that collectively obey said harmonic relationship.

70. The bandwidth extension module defined in claim 69, wherein said processing entity comprises:

a band-pass filter configured to produce a band-pass signal from the first speech signal;

at least one carrier frequency modulator, each said carrier frequency modulator configured to pitch-synchronously modulate said band-pass signal about a respective carrier frequency, the at least one carrier frequency modulator collectively producing a highband speech signal component;

a synthesis filter configured to determine a highband speech signal based on said highband speech signal component;

a summation module configured to combine said first speech signal with said highband speech signal to obtain said second speech signal.