Method and apparatus for encoding speech

• US 4,914,701 A
• Filed: 08/29/1988
• Issued: 04/03/1990
• Est. Priority Date: 12/20/1984
• Status: Expired due to Term

• Current Assignee: Verizon Laboratories Incorporated (Verizon Communications Inc.)
• Original Assignee: GTE Laboratories Incorporated (Lumen Technologies, Inc.)
• Inventors: Zibman, Israel B.
• Primary Examiner(s): KEMENY, EMANUEL
• Application Number:
  US07/239,042
• Time in Patent Office: 582 Days
• Field of Search: 381/36-39, 381/50
• US Class Current: 704/203
• CPC Class Codes: G10L 19/02 using spectral analysis, e.g. transform vocoders or subband vocoders
• CPC Class Codes: G10L 21/038 using band spreading techniques
• CPC Class Codes: H04B 1/66 for reducing bandwidth of signals; for improving efficiency of transmission H04B1/68 takes precedence

Abstract

In a speech encoder a Fourier transform of the speech is provided. The Fourier transform is equalized by normalizing the spectrum coefficients to a curve which approximates the shape of the spectrum. Both the curve and the equalized spectrum are encoded. Preferably, only a baseband of the normalized spectrum is encoded and that baseband is repeated in the decoder. The spectrum is normalized by scaling different regions (subbands) of the spectrum differently to flatten the spectrum.

Claims

1. A speech encoder comprising:

Fourier transform means for performing a Fourier transform of a window of speech with formants to generate a Fourier transform spectrum;

normalizing means for defining from the Fourier transform spectrum at least one curve of different magnitudes approximating different magnitudes of the Fourier transform spectrum across the spectrum, for digitally encoding the at least one defined curve and for defining the Fourier transform spectrum relative to the at least one defined curve to provide a normalized spectrum; and

means for encoding at least a portion of the normalized spectrum.

2. A speech encoder as claimed in claim 1 wherein the normalizing means comprises:

means for determining the maximum magnitude of Fourier transform spectrum within each of a plurality of regions of the spectrum;

means for digitally encoding the maximum magnitude of each region; and

means for scaling each coefficient of the Fourier transform spectrum in each region to the maximum magnitude of each region to provide a first set of normalized outputs.

3. A speech encoder as claimed in claim 2 wherein the normalizing means further comprises:

means for determining the maximum magnitude of the first set of normalized outputs in each of a plurality of subregions of the spectrum;

means for digitally encoding the maximum magnitude of each subregion; and

means for scaling each output of the first set of normalized outputs to the maximum magnitude of each subregion to provide a second set of normalized outputs.

4. A speech encoder as claimed in claim 3 wherein each of the maximum magnitudes is logarithmically encoded.

5. A speech encoder as claimed in claim 3 wherein the maximum magnitude is determined for each of four regions corresponding to the first four formants.

6. A speech encoder as claimed in claim 2 wherein only a baseband of the normalized spectrum is encoded.

24. A speech encoder as claimed in claim 1 wherein the means for encoding at least a portion of the normalized spectrum encodes phase and amplitude information.

7. A speech encoder comprising:

means for sampling a speech signal;

an analog to digital converter for providing digital representations of the speech samples;

a preemphasis filter;

Fourier transform means for performing a Fourier transform of a window of digital speech samples to generate a Fourier transform spectrum;

means for determining the maximum magnitude of the Fourier transform spectrum within each of a plurality of regions of the spectrum;

means for digitally encoding the maximum magnitude of each region;

means for dividing each coefficient of the Fourier spectrum in each region by the maximum magnitude of each region to provide a first set of normalized outputs;

means for determining the maximum magnitude of the first set of normalized outputs in each of a plurality of subregions of the spectrum;

means for digitally encoding the maximum magnitude of each subregion;

means for dividing each output of the first set of normalized outputs by the maximum magnitude of each subregion to provide a second set of normalized outputs; and

means for encoding a baseband of the second set of normalized outputs.

8. A method of encoding speech comprising:

performing a Fourier transform of a window of speech with formants to generate a Fourier transform spectrum;

providing a normalized spectrum by defining from the Fourier transform spectrum at least one curve of different magnitudes approximately different magnitudes of the Fourier transform spectrum across the spectrum, digitally encoding the at least one defined curve and defining the Fourier transform spectrum relative to the at least one defined curve; and

encoding at least a portion of the normalized spectrum.

9. A method as claimed in claim 8 wherein the normalized spectrum is provided by;

determining a maximum magnitude of the Fourier transform within each of a plurality of regions of the spectrum;

digitally encoding the maximum magnitude of each region; and

scaling each coefficient of the Fourier spectrum in each region to the maximum magnitude of each region.

10. A method as claimed in claim 9 wherein the normalized spectrum is provided by further:

determining the maximum magnitude of the first set of normalized outputs in each of a plurality of subregions of the spectrum;

digitally encoding the maximum magnitude of each subregion; and

scaling each output of the first set of normalized outputs to the maximum magnitude of each subregion to provide a second set of normalized outputs.

11. A method as claimed in claim 10 wherein each of the maximum magnitudes is logarithmically encoded.

12. A method as claimed in claim 10 wherein the maximum magnitudes are determind for four regions corresponding to the first four formants.

13. A method as claimed in claim 8 wherein only a baseband of the normalized spectrum is encoded.

25. A method as claimed in claim 8 wherein the step of encoding at least a portion of the normalized spectrum includes encoding phase and amplitude information.

14. A speech encoder comprising:

transform means for performing a transform of an incoming speech signal with formants to generate a transform spectrum which varies significantly in magnitude across the spectrum;

equalizing means for modifying the transform spectrum to provide a substantially flat spectrum and for encoding a function derived from the transform spectrum by which the transform spectrum is modified; and

means for encoding at least a portion of the equalized spectrum.

15. A speech encoder as claimed in claim 14 wherein the transform means performs a Fourier transform.

16. A speech encoder as claimed in claim 15 wherein only a baseband of the equalized spectrum is encoded.

17. A speech encoder as claimed in claim 14 wherein only a baseband of the equalized spectrum is encoded.

26. A speech encoder as claimed in claim 14 wherein the means for encoding at least a portion of the equalized spectrum encodes both phase and amplitude information.

18. A speech encoder comprising:

transform means for performing a transform of a window of speech with formants to generate a transform spectrum;

normalizing means for defining a magnitude relative to each of a plurality of regions of the transform spectrum and for scaling each coefficient of the transform spectrum, in each region of at least a portion of the spectrum, to the defined magnitude of the region of provide a normalized spectrum; and

means for encoding the defined magnitudes and at least a portion of the normalized spectrum.

19. A speech encoder as claimed in claim 18 wherein the transform means performs a Fourier transform.

27. A speech encoder as claimed in claim 18 wherein the means for encoding the normalized spectrum encodes both phase and amplitude information.

20. A method of encoding speech comprising:

performing a transform of an incoming speech signal to generate a transform spectrum which varies significantly in magnitude across the spectrum;

modifying the transform spectrum by a function derived from the transform spectrum to provide a substantially flat spectrum; and

encoding the function derived from the transform spectrum by which the transform spectrum is modified and encoding at least a portion of the modified spectrum.

21. A method as claimed in claim 20 wherein the transform performed is a Fourier transform.

23. A method as claimed in claim 21 wherein the transform performed is a Fourier transform.

28. A method as claimed in claim 20 wherein the step of encoding at least a portion of the modified spectrum includes encoding both phase and amplitude information.

22. A method of encoding speech comprising:

performing a transform of a window of speech with formants to generate a transform spectrum;

defining a magnitude relative to each of a plurality of regions of the transform spectrum and scaling each coefficient of the transform spectrum, in each region of at least a portion of the spectrum, to the defined magnitude of the region; and

encoding the defined magnitudes and at least a portion of the scaled coefficients of the transform spectrum.

29. A method as claimed in claim 22 wherein the step of encoding at least a portion of the scaled coefficients of the transform spectrum includes encoding both phase and amplitude information.