Fixed rate speech compression system and method

• US 6,959,274 B1
• Filed: 09/15/2000
• Issued: 10/25/2005
• Est. Priority Date: 09/22/1999
• Status: Expired due to Term

• Current Assignee: DigiMedia Tech, LLC (IP Investments Group LLC)
• Original Assignee: Mindspeed Technologies Inc. (MACOM Technology Solutions Holdings, Inc.)
• Inventors: Thyssen, Jes, Shlomot, Eyal, Su, Huan-Yu, Gao, Yang, Benyassine, Adil
• Primary Examiner(s): Ometz, David L.
• Assistant Examiner(s): Opsasnick, Michael N.
• Application Number:
  US09/663,662
• Time in Patent Office: 1,866 Days
• Field of Search: 704/222, 704/207, 704/219, 704/230
• US Class Current: 704/219
• 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 19/20 using sound class specific coding, hybrid encoders or object based coding
• CPC Class Codes: G10L 19/22 Mode decision, i.e. based on audio signal content versus external parameters
• 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: G10L 2019/0001 Codebooks
• CPC Class Codes: H03G 3/00 Gain control in amplifiers or frequency changers

Abstract

The invention improves the encoding and decoding of speech by focusing the encoding on the perceptually important characteristics of speech. The system analyzes selected features of an input speech signal, and first performing a common frame based speech coding of an input speech signal. The system then performs a speech coding based on either a first speech coding mode or a second speech coding mode. The selection of a mode is based on characteristics of the input speech signal. The first speech coding mode uses a first framing structure and the second speech coding mode uses a second framing structure.

Claims

1. A fixed rate speech compression system for processing a frame of a speech signal, the fixed rate speech compression system comprising:

an encoder operable to encode u first part of the frame using common frame based encoding;

the common frame based encoding comprising pitch pre-processing to modify the waveform of the speech signal as a function of classification of the frame;

the encoder operable to select one of a first speech coding mode and a second speech coding mode to encode a second part of the frame, wherein the first speech coding mode uses less bits than the second speech coding mode to code a fixed codebook contribution, wherein the first speech coding mode uses a first vector quantizer to jointly code both an adaptive codebook gain and a fixed codebook gain, and wherein the second speech coding;

mode uses a second vector quantizer to code the adaptive codebook gain and a third vector quantizer to code the fixed codebook gain.

2. The fixed rate speech compression system of claim 1, where the encoder is operable to continuously time warp the speech signal during pitch pre-processing when the frame is classified as indicative of increased voicing strength.

3. The fixed rate speech compression system of claim 1, where the encoder is operable to selectively perform continuous time warping of the speech signal during pitch preprocessing to introduce a variable delay of up to about twenty samples.

4. The fixed rate speech compression system of claim 1, where the encoder is operable to selectively estimate continuous time warping of the speech signal during pitch pre-processing by interpolation with Hamming weighted Sinc interpolation filters.

5. The fixed rate speech compression system of claim 1, where the encoder is operable to select the first speech coding mode as a function of classification of the frame as at least one of silence/background noise, noise-like unvoiced speech, unvoiced speech, onset speech, plosive speech and non-stationary voiced speech.

6. The fixed rate speech compression system of claim 1, where the encoder is operable to select the second speech coding mode as a function of classification of the frame as stationary voiced speech.

7. The fixed rate speech compression system of claim 1, where a frame classified as at least one of background noise and unvoiced speech remains unchanged by pitch pre-processing.

8. The fixed rate speech compression system of claim 1, where the encoder is operable to time shift the speech signal with pitch pre-processing in a frame classified as predominantly pulse-like unvoiced speech.

9. The fixed rate speech compression system of claim 1, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein fourteen bits are allocated to the two-dimensional vector quantization gain codebook.

10. The fixed rate speech compression system of claim 1, wherein the second speech coding mode uses two three-dimensional vector quantization gain codebooks.

11. The fixed rate speech compression system of claim 1, wherein the system is configured to operate at approximately 4 kbits/s.

12. The fixed rate speech compression system of claim 1, wherein 21 bits are allocated to code linear prediction coefficients.

13. The fixed rate speech compression system of claim 1, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein the two-dimensional code-vector is selected from the two-dimensional vector quantization gain codebook.

14. The fixed rate speech compression system of claim 1, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein the two-dimensional vector quantization gain codebook has an adaptive codebook gain and a fixed codebook gain.

15. A method of processing a frame of a speech signal with a fixed rate speech compression system, the method comprising:

encoding a first part of the frame with common frame based encoding, the common frame based encoding comprising;

classifying the frame;

pitch pre-processing to modify the waveform of the speech signal as a function of classification of the frame; and

selecting one of a first speech coding mode and a second speech coding mode to encode a second part of the frame, wherein the first speech coding mode uses less bits than the second speech coding mode to code a fixed codebook contribution, wherein the first speech coding mode uses a first vector quantizer to jointly code both an adaptive codebook gain and a fixed codebook gain, and wherein the second speech coding mode uses a second vector quantizer to code the adaptive codebook gain and a third vector quantizer to code the fixed codebook gain.

16. The method of claim 15, where classifying the frame comprises classifying the frame as a function of pitch correlation information.

17. The method of claim 15, where pitch preprocessing comprises:

classifying the speech signal as indicative of increased voicing strength; and

continuously

time warping the frame of the speech signal to introduce a variable delay.

18. The method of claim 15, where pitch pre-processing comprises:

classifying the speech signal as predominantly pulse-like unvoiced speech; and

time shifting the waveform as a function of an accumulated delay.

19. The method of claim 15, where pitch pre-processing comprises:

classifying the speech signal as at least one of predominantly background noise and predominantly unvoiced speech; and

resetting an accumulated delay without modification of the waveform.

20. The method of claim 15, where pitch pre-processing comprises modifying at least one pitch cycle of the speech signal to provide continuous time warping of the speech signal.

21. The method of claim 15, where selecting the first speech coding mode comprises classifying the frame as at least one of silence/background noise, noise-like unvoiced speech, unvoiced speech, onset speech, plosive speech and non-stationary voice speech.

22. The method of claim 15, where selecting the second speech coding mode comprises classifying the frame as stationary voiced speech.

23. The method of claim 15, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein the method further comprising allocating fourteen bits to the two-dimensional vector quantization gain codebook.

24. The method of claim 15, wherein the second speech coding mode uses two three-dimensional vector quantization gain codebooks.

25. The method of claim 15, wherein the system operates at approximately 4 kbits/s.

26. The method of claim 15, further comprising allocating 21 bits to code linear prediction coefficients.

27. The method of claim 15, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein the method further comprising selecting the two-dimensional code-vector from the two-dimensional vector quantization gain codebook.

28. The method of claim 15, wherein the first speech coding mode uses a two-dimensional vector quantization gain codebook and a two-dimensional code-vector, and wherein the two-dimensional vector quantization gain codebook has an adaptive codebook gain and a fixed codebook gain.