System of encoding and decoding speech signals

• US 6,604,070 B1
• Filed: 09/15/2000
• Issued: 08/05/2003
• Est. Priority Date: 09/22/1999
• Status: Expired due to Term

• Current Assignee: MACOM Technology Solutions Holdings, Inc.
• Original Assignee: Conexant Systems Incorporated (Synaptics Incorporated)
• Inventors: Thyssen, Jes, Shlomot, Eyal, Gao, Yang, Benyassine, Adil, Su, Huan-yu
• Primary Examiner(s): To, Doris H.
• Assistant Examiner(s): Nolan, Daniel A.
• Application Number:
  US09/663,734
• Time in Patent Office: 1,054 Days
• Field of Search: 704/222, 704/230, 704/229, 704/219, 704/220, 704/228, 704/223, 704/500, 704/503, 370/336
• US Class Current: 704/222
• CPC Class Codes: G10L 19/00 Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis in musical instruments G10H
• CPC Class Codes: G10L 19/167 Audio streaming, i.e. formatting and decoding of an encoded audio signal representation into a data stream for transmission or storage purposes
• CPC Class Codes: G10L 19/24 Variable rate codecs, e.g. for generating different qualities using a scalable representation such as hierarchical encoding or layered encoding
• CPC Class Codes: H03G 3/00 Gain control in amplifiers or frequency changers

Abstract

A speech compression system capable of encoding a speech signal into a bitstream for subsequent decoding to generate synthesized speech is disclosed. The speech compression system optimizes the bandwidth consumed by the bitstream by balancing the desired average bit rate with the perceptual quality of the reconstructed speech. The speech compression system comprises a full-rate codec, a half-rate codec, a quarter-rate codec and an eighth-rate codec. The codecs are selectively activated based on a rate selection. In addition, the full and half-rate codecs are selectively activated based on a type classification. Each codec is selectively activated to encode and decode the speech signals at different bit rates emphasizing different aspects of the speech signal to enhance overall quality of the synthesized speech.

Claims

1. A speech compression system for processing a speech signal, the speech compression system comprising:

an encoding system operable to select a rate for encoding the speech signal as a function of characterization of the speech signal,

the encoding system operable to select a type classification, comprising one of a first type and a second type,

the encoding system operable to encode the speech signal as a function of the rate and the type classification when a first rate is selected,

the encoding system operable to encode the speech signal as a function of the rate, regardless of the type classification, when a second rate is selected; and

the encoding system operable to encode an adaptive codebook gain and a fixed codebook gain when the first rate is selected and the type classification is the second type, where the adaptive codebook gain is encoded exclusive of the fixed codebook gain.

2. The speech compression system of claim 1, where the encoding system is operable to jointly encode an adaptive codebook gain and a fixed codebook gain when the first rate is selected and the type classification is the first type.

3. The speech compression system of claim 2, where the jointly encoded adaptive codebook gain and the fixed codebook gain are encoded for each of a plurality of subframes of the speech signal.

4. The speech compression system of claim 2, where the adaptive codebook gain and the fixed codebook gain are representative of a frame of the speech signal.

5. The speech compression system of claim 1, where the encoded speech signal comprises a first portion and a second portion when the first rate is selected, the first portion representative of parameters of the speech signal on a frame basis and the second portion representative of parameters of the speech signal on a subframe basis.

6. The speech compression system of claim 5, where the encoding system comprises an initial frame processing module and an excitation-processing module, the initial frame processing module operable to generate the first portion of the encoded speech signal and the excitation-processing module operable to generate the second portion of the encoded speech signal when the type classification of the frame is the first type.

7. The speech compression system of claim 5, where the encoding system comprises an initial frame processing module and an excitation-processing module, the initial frame processing module operable to generate part of the first portion of the encoded speech signal and the excitation-processing module operable to generate part of the first portion of the encoded speech signal and the second portion of the encoded speech signal when the type classification of the frame is the second type.

8. The speech compression system of claim 1, where the first rate is one of 8.5 kilobits per second and 4.0 kilobits per second.

9. The speech compression system of claim 1, where the second rate is one of 2.0 kilobits per second and 0.8 kilobits per second.

10. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

a first codec operable to encode and decode a first frame of the speech signal, the first codec activated as a function of a selected bit rate, the first codec operable to encode and decode as a function of a type classification, the type classification comprising one of a first type and a second type; and

a second codec operable to encode and decode a second frame of the speech signal, the second codec activated as a function of a selected bit rate to encode and decode the second frame regardless of the type classification;

where the first codec is operable to encode an adaptive codebook gain and a fixed codebook gain when the first rate is selected and the type classification is the second type, and where the adaptive codebook gain is encoded exclusive of the fixed codebook gain.

11. The variable rate speech compression system of claim 10, further comprising an initial frame processing module operable to characterize the first frame and the second frame, and the initial frame processing module further operable to select the bit rate for each frame as a function of the characterization.

12. The variable rate speech compression system of claim 11, where the first frame and the second frame are each characterized as one of a plurality of classes, the classes comprising silence/background noise, stationary noise-like unvoiced speech, non-stationary unvoiced speech, onset speech, non-stationary voiced speech and stationary voiced speech.

13. The variable rate speech compression system of claim 11, where the initial frame processing module is operable to select the bit rate as a function of a mode signal, the mode signal comprising a premium mode, a standard mode and an economy mode.

14. The variable rate speech compression system of claim 10, further comprising an initial frame processing module operable to characterize the first frame as one of a plurality of classes, and the initial frame processing module further operable to perform the type classification as a function of the characterization.

15. The variable rate speech compression system of claim 10, where the first type is a frame characterized as at least one of silence/background noise, stationary noise-like unvoiced speech, non-stationary unvoiced speech, onset speech and non-stationary voiced speech.

16. The variable rate speech compression system of claim 10, where the second type is a frame characterized as stationary voiced speech.

17. The variable rate speech compression system of claim 10, where the first codec is operable to encode an adaptive codebook contribution representative of each of a plurality of subframes of the first frame when the type classification is the first type.

18. The variable rate speech compression system of claim 10, where the first codec is operable to encode an adaptive codebook contribution representative of the first frame when the type classification is the second type.

19. The variable rate speech compression system of claim 10, where the first codec comprises one of a full-rate codec and a half-rate codec.

20. The variable rate speech compression system of claim 10, where the second codec comprises one of a quarter-rate codec and an eighth-rate codec.

21. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

an encoding system operable to extract and encode parameters of a frame of the speech signal, the encoding system comprising;

a frame processing module operable to encode parameters representative of the frame as a function of a rate selection comprising one of a first rate and a second rate and a type classification comprising one of a first type and a second type, the frame processing module comprising a pitch pre-processing module operable to modify the speech signal to improve encoding; and

a subframe processing module operable to encode parameters representative of a subframe of the frame as a function of the rate selection and the type classification;

where the pitch pre-processing module is operable to modify the speech signal as a function of a grade characterization of the speech signal.

22. The variable rate speech compression system of claim 21, where the pitch pre-processing module is operable to perform continuous time warping of the speech signal.

23. The variable rate speech compression system of claim 21, where the pitch pre-processing module is operable to perform waveform interpolation of the speech signal.

24. The variable rate speech compression system of claim 21, where the frame processing module comprises an interpolation module operable to select an interpolation path when the type classification is the first type and the rate selection is the first rate, the frame processing module operable to encode short-term predictor parameters as a function of the interpolation path.

25. The variable rate speech compression system of claim 21, where the frame processing module comprises a predictor switch module operable to select predictor coefficients when the rate selection is the second rate, the frame processing module operable to encode short-term predictor parameters as a function of the predictor coefficients.

26. The variable rate speech compression system of claim 21, where the frame processing module is operable to encode short-term predictor parameters, an open loop pitch lag, an adaptive codebook gain and a fixed codebook gain when the type classification is the second type.

27. The variable rate speech compression system of claim 21, where the frame processing module comprises a 3D/4D open loop VQ module operable to encode an adaptive codebook gain with a pre-gain quantization table when the type classification is the second type.

28. The variable rate speech compression system as in claim 21, where the frame processing module comprises a 3D/4D VQ gain codebook, operable to encode a fixed codebook gain with a delayed gain quantization table when the type classification is the second type.

29. The variable rate speech compression system of claim 21, where the subframe processing module is operable to encode a closed loop pitch lag, a fixed codebook contribution, an adaptive codebook gain and a fixed codebook gain when the type classification is the first type.

34. The variable rate speech compression system as in claim 21, where subframe processing module is operable to perform enhancements to a fixed codebook, the enhancements comprising at least one of injection of high frequency noise, incorporation of additional pulses and filtration.

35. The variable rate speech compression system of claim 21, where the subframe processing module is operable to determine a quantized gain vector for the subframe when the rate selection is one of the first rate and the second rate, and the type classification is the first type, the quantized gain vector being representative of joint encoding of an adaptive codebook gain and a fixed codebook gain.

36. The variable rate speech compression system of claim 21, where the first rate is 8.5 kilobits per second and the second rate is 4.0 kilobits per second.

30. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

an encoding system operable to extract and encode parameters of a frame of the speech signal, the encoding system comprising;

a frame processing module operable to encode parameters representative of the frame as a function of a rate selection comprising one of a first rate and a second rate and a type classification comprising one of a first type and a second type, the frame processing module comprising a pitch pre-processing module operable to modify the speech signal to improve encoding; and

a subframe processing module operable to encode parameters representative of a subframe of the frame as a function of the rate selection and the type classification;

where the subframe processing module is operable to search a fixed codebook to determine a fixed codebook contribution for the subframe, when the rate selection is the first rate, and the type classification is the first type;

where the fixed codebook comprises three 5-pulse codebooks.

31. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

an encoding system operable to extract and encode parameters of a frame of the speech signal, the encoding system comprising;

a frame processing module operable to encode parameters representative of the frame as a function of a rate selection comprising one of a first rate and a second rate and a type classification comprising one of a first type and a second type, the frame processing module comprising a pitch pre-processing module operable to modify the speech signal to improve encoding; and

a subframe processing module operable to encode parameters representative of a subframe of the frame as a function of the rate selection and the type classification;

where the subframe processing module is operable to search a fixed codebook to determine a fixed codebook contribution for the subframe, when the rate selection is the first rate, and the type classification is the second type;

where the fixed codebook comprises an 8-pulse codebook.

32. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

an encoding system operable to extract and encode parameters of a frame of the speech signal, the encoding system comprising;

a frame processing module operable to encode parameters representative of the frame as a function of a rate selection comprising one of a first rate and a second rate and a type classification comprising one of a first type and a second type, the frame processing module comprising a pitch pre-processing module operable to modify the speech signal to improve encoding; and

a subframe processing module operable to encode parameters representative of a subframe of the frame as a function of the rate selection and the type classification;

where the subframe processing module is operable to search a fixed codebook to determine a fixed codebook contribution for the subframe, when the rate selection is the second rate, and the type classification is the first type;

the fixed codebook comprising a 2-pulse codebook, a 3-pulse codebook and a Gaussian codebook.

33. A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

an encoding system operable to extract and encode parameters of a frame of the speech signal, the encoding system comprising;

a frame processing module operable to encode parameters representative of the frame as a function of a rate selection comprising one of a first rate and a second rate and a type classification comprising one of a first type and a second type, the frame processing module comprising a pitch pre-processing module operable to modify the speech signal to improve encoding; and

a subframe processing module operable to encode parameters representative of a subframe of the frame as a function of the rate selection and the type classification;

where the subframe processing module is operable to search a fixed codebook to determine a fixed codebook contribution for the subframe, when the rate selection is the second rate, and the type classification is the second type;

where the fixed codebook comprises a 2-pulse codebook and a 3-pulse codebook.

37. A system for processing a frame of a speech signal, the system comprising:

a first codec selectively activated as a function of a rate selected for the frame, the first codec comprising;

a plurality of frame processing modules operable to encode parameters representative of the frame as a function of a type classification of the frame to form a first portion of an encoded speech signal; and

a plurality of subframe processing modules operable to encode parameters representative of each of a plurality of subframes of the frame as a function of the type classification to form a second portion of the encoded speech signal; and

a second codec selectively activated as a function of the rate selected for the frame, the second codec comprising;

an initial frame-processing module operable to encode parameters representative of the frame regardless of the type classification; and

a rate module operable to encode parameters of the frame regardless of the type classification;

where the first codec is operable to encode an adaptive codebook gain and a fixed codebook gain as a function of the rate and the type classification, and where the adaptive codebook gain is encoded exclusive of the fixed codebook gain.

38. The system of claim 37, where the first codec comprises one of a full-rate codec and a half-rate codec.

39. The system of claim 37, where the second codec comprises one of a quarter-rate codec and an eighth-rate codec.

40. The system of claim 37, where the rate selected for the first codec comprises one of 8.5 kilobits per second and 4.0 kilobits per second.

41. The system of claim 37, where the rate selected for the second codec comprises one of 2.0 kilobits per second and 0.8 kilobits per second.

42. The system of claim 37, where the type classification comprises one of a first type and a second type.

43. The system of claim 42, where the second type represents frames exhibiting stationary voiced behavior and the first type represents all other types of behavior.

44. The system of claim 37, where the rate module of the second codec is operable to encode a predicted energy gain representative of the frame when the rate selected is an eighth rate.

47. The method of claim 41, further comprising:

h) post-processing the frame as a function of the first rate selection when the first rate selection was previously selected; and

i) post-processing the frame as a function of the second rate selection when the second rate selection was previously selected.

48. The method of claim 47, where h) comprises performing a fine tuning search for a pitch period as a function of pitch correlation and gain controlled harmonic filtering.

45. A method of processing a speech signal with a speech compression system, the method comprising:

a) extracting parameters from a frame of the speech signal;

b) selecting from one of a first rate selection and a second rate selection as a function of the parameters extracted from the frame;

c) selecting a type classification as a function of the parameters extracted from the frame when the first rate selection is selected;

d) encoding the frame as a function of the first rate selection and the type classification when the first rate selection is selected; and

e) encoding the frame as a function of the second rate selection, regardless of the type classification, when the second rate selection is selected;

where b) comprises characterizing the frame as a function of the parameters., and where the characterizing the frame comprises;

performing an initial characterization, and performing a final characterization.

46. The method of claim 45, further comprising:

f) decoding the frame as a function of the first rate selection and the type classification when the first rate selection was previously selected; and

g) decoding the frame as a function of the second rate selection regardless of the type classification when the second rate selection was previously selected.

49. The method of claim 45, where d) comprises encoding with a full-rate codec and a half-rate codec.

50. The method of claim 45, where e) comprises encoding with one of a quarter-rate codec and an eighth-rate codec.

51. The method of claim 45, where d) comprises:

generating a first portion of a bitstream by encoding parameters representative of the frame; and

generating a second portion of the bitstream by encoding parameters representative of each of a plurality of subframes of the frame.

52. The method of claim 51, where d) further comprises determining as a function of the type classification the parameters representative of the frame and the parameters representative of one of the subframes.

53. The method of claims 45, where c) comprises:

characterizing the frame as at least one of silence/background noise, stationary noise-like unvoiced speech, non-stationary unvoiced speech, onset speech and non-stationary voiced speech; and

selecting type zero.

54. The method of claim 45, where c) comprises:

characterizing the frame as stationary voiced speech; and

selecting type one.

55. A method of processing a speech signal, the method comprising:

a) extracting parameters from a frame of the speech signal;

b) selecting a bit rate as a function of the extracted parameters;

c) selecting a type classification as a function of the extracted parameters;

d) encoding extracted parameters representative of the frame as a function of the bit rate and the type classification; and

e) encoding extracted parameters representative of each of a plurality of subframes of the frame as a function of the bit rate and the type classification;

where e) comprises encoding a fixed codebook contribution when the type classification is type one, and where encoding the fixed codebook contribution comprises searching an 8-pulse codebook when the bit rate selected is a full rate.

56. The method of claim 55, where b) comprises modifying the speech signal to fit a pitch track.

57. The method of claim 55, where b) comprises selecting the bit rate as a function of a desired average bit rate.

58. The method of claim 55, where b) comprises selecting the bit rate as one of 8.5 kilobits per second and 4.0 kilobits per second.

59. The method of claim 55, where d) comprises choosing an interpolation path to encode short-term prediction parameters when the type classification is type one and the bit rate selected is a full rate.

60. The method of claim 55, where d) comprises identifying predictor coefficients to encode short-term prediction parameters when the bit rate selected is a half rate.

61. The method of claim 55, where d) comprises encoding short-term prediction parameters, an open loop pitch lag, an adaptive codebook gain and a fixed codebook gain when the type classification is type one.

63. The method of claim 55, where d) comprises encoding short-term prediction parameters when the type classification is type zero.

64. The method of claim 55, where e) comprises encoding a closed loop pitch lag, a fixed codebook contribution, a fixed codebook gain and an adaptive codebook gain when the type classification is type zero.

65. The method of claim 64, where encoding the fixed codebook contribution when the bit rate selected is a half rate comprises:

injecting high frequency noise into a plurality of codebooks;

incorporating additional pulses into the codebooks;

filtering the codebooks; and

searching the codebooks to determine the fixed codebook contribution.

66. The method of claim 64, where encoding the fixed codebook gain and the adaptive codebook gain comprises:

jointly quantizing the fixed codebook gain and the adaptive codebook gain for each of the subframes; and

determining a quantized gain vector representing the joint quantization.

62. A method of processing a speech signal, the method comprising:

a) extracting parameters from a frame of the speech signal;

b) selecting a bit rate as a function of the extracted parameters;

c) selecting a type classification as a function of the extracted parameters;

d) encoding extracted parameters representative of the frame as a function of the bit rate and the type classification; and

e) encoding extracted parameters representative of each of a plurality of subframes of the frame as a function of the bit rate and the type classification;

where e) comprises encoding a fixed codebook contribution when the type classification is type one, and where encoding the fixed codebook contribution comprises searching a 2-pulse codebook and a 3-pulse codebook when the bit rate selected is a half rate.