Dynamic bit allocation subband excited transform coding method and apparatus

[54] DYNAMIC BIT ALLOCATION SUBBAND EXCITED TRANSFORM CODING METHOD AND APPARATUS

[75] Inventor: Baruch Mazor, Newton Center, Mass.

[73] Assignee: GTE Laboratories Incorporated, Waltham, Mass.

[21] Appl. No.: 597,438

[22] Filed: Oct. 15,1990

[51] Int. CI.' G10L9/02

[52] U.S. CI 395/2

[58] Field of Search 381/41-50,

381/29-40; 395/2

[56] References Cited

U.S. PATENT DOCUMENTS

4,184,049 1/1980 Crochiete et al 179/1 SA

4,330,689 5/1982 Kang et al 179/15.55 R

4,455,649 6/1984 Esteben et al 381/29

4,535,472 8/1985 Tomcik 381/31

4,538,234 8/1985 Honda et al 381/31

4,610,022 9/1986 Kitayama et al 381/36

4,790,016 12/1988 Mazor et al 381/36

4,914,701 4/1990 Zibman 381/36

4,956,871 9/1990 Swaminathan 381/31

5,042,069 8/1991 Chhatwal et al 381/36

FOREIGN PATENT DOCUMENTS 0176243A3 2/1986 European Pat. Off. .

OTHER PUBLICATIONS Mazor, B., "An Optimal Transform Trellis Code with

Applications to Speech" Oct., 1985, vol. COM-33, 10:1111-1112.

Flanagan, J. L. et al., "Speech Coding," IEEE Transactions on Communications, vol. Com-27, 4:710-736 (1979).

Kang, G. S. et al., "Mediumband Speech Processor with Baseband Residual Spectrum Encoding," Naval Research Lab, Washington, D.C. 20375, 1981, pp. 820-823.

Tribolet, J. M. et al., "Frequency Domain Coding of Speech," IEEE Transactions on Acoustics, Speech and Signal Processing, vol. ASSP-27, No. 5, Oct., 1979, pp. 512-530.

Westerink, P. H. et al, "An Optimal Bit Allocation Algorithm For Sub-Band Coding," Proceedings of 1988 IEEE International Conference on Acoustics, Speech and Signal Processing, vol. II, pp. 757-760, New York, 11-14 Apr. 1988.

Primary Examiner—Allen R. MacDonald

Assistant Examiner—Michelle Doerrler

Attorney, Agent, or Firm—Hamilton, Brook, Smith &

Reynolds

[57] ABSTRACT

In an adaptive subband excited transform speech encoding system, a range of quantizers are available and are dynamically selected for each window of speech. The quantizers designated for individual subbands are determined to minimize mean squared error distortion in the recreated signal while using no more than a predetermined number of quantization bits per window of speech.

18 Claims, 5 Drawing Sheets