Speech parameter coding method and appparatus

• US 5,487,128 A
• Filed: 02/26/1992
• Issued: 01/23/1996
• Est. Priority Date: 02/26/1991
• Status: Expired due to Term

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Ozawa, Kazunori
• Primary Examiner(s): Knepper, David D.
• Application Number:
  US07/841,726
• Time in Patent Office: 1,427 Days
• Field of Search: 381/36, 381/29-40, 395/2, 395/2.28-2.32, 395/2.38, 395/2.39
• US Class Current: 704/222
• CPC Class Codes: G10L 19/038 Vector quantisation, e.g. TwinVQ audio
• CPC Class Codes: G10L 2019/0005 Multi-stage vector quantisation
• CPC Class Codes: H03M 7/3082 Vector coding for television signals, see H04N19/94

Abstract

A speech parameter coding method and apparatus which can quantize a spectrum parameter of a speech signal with a smaller number of bits smaller than ever. A dividing section divides a predetermined order number of spectrum parameters, obtained by division of a speech signal into frames, for each order number of spectrum parameters smaller than the divisional order number. A vector quantizing section searches codebooks for the divided spectrum parameters for each order number and outputs a plurality of candidates of codevector in order of magnitude from the minimum one. An accumulated distortion calculating section calculates accumulated distortions for the entire order number for combinations of codevectors. A minimum judging section selectively outputs a combination of codevectors which minimizes the accumulated distortion thereby to quantize the spectrum parameter.

Claims

1. A speech parameter coding method, comprising the steps of:

receiving an input speech signal;

dividing the received speech signal into a plurality of frames each of a predetermined time length;

finding spectrum parameters, each of which corresponds to first to P-th order, of the speech signal for each of the frames;

dividing the spectrum parameters into a plural number of said spectrum parameters in accordance with their orders;

searching a spectrum codebook for the divided spectrum parameters and calculating a distortion due to quantizing by use of codevectors in said codebook to obtain a plurality of candidate codevectors.

2. A speech parameter coding method, comprising the steps of:

receiving an input speech signal;

dividing the received speech signal into a plurality of frames;

finding spectrum parameters, each of which corresponds to first to P-th order, of the speech signal for each of the frames;

representing the spectrum parameters in a multi-stage cascade connection of a plurality of codebooks;

obtaining a plurality of candidate codevectors at at least one of stages of the multi-stage cascade connection;

dividing, at the other stage or at at least one of the other stages of the multi-stage cascade connection, the spectrum parameters into a plural member of said spectrum parameters in accordance with their orders;

searching other codebooks for the divided spectrum parameters to obtain a plurality of candidates of codevectors; and

selecting a combination of codevectors which minimizes an accumulated distortion thereby to quantize the spectrum parameters.

3. A speech parameter coding method, comprising the steps of:

receiving an input speech signal;

dividing the input speech signal into a plurality of frames;

dividing each of the frames into a plurality of shorter subframes;

finding spectrum parameters, each of which corresponds to first to P-th order, for the speech signal of at least one of the subframes;

quantizing the spectrum parameters of at least one of the subframes using a first codebook constructed in advance; and

quantizing parameters of the other subframe or at least of the other subframes using the quantized values and a second codebook constructed in advance or non-linear processing.

4. A speech parameter coding method, comprising the steps of:

receiving an input speech signal;

dividing the input speech signal into a plurality of frames;

dividing each of the frames into a plurality of shorter subframes;

finding spectrum parameters, each of which corresponds to first to P-th order, for the speech signal of at least one of the subframes;

quantizing the spectrum parameters of at least one of the subframes using a first codebook constructed in advance;

finding difference signals between the spectrum parameters of the other subframe or at least one of the other subframes and the quantized values; and

quantizing the difference signals using a second codebook constructed in advance.

5. A speech parameter coding apparatus, comprising:

means for dividing an input speech signal into a plurality of frames;

means for finding spectrum parameters, each of which corresponds to first to P-th order, from the speech signal;

means for vector quantizing the spectrum parameters to obtain vector quantized values;

means for scalar quantizing difference signals between the spectrum parameters and the vector quantized values; and

means for determining a predetermined order of quantizing ranges for the scalar quantization for each of a predetermined number of codevectors of said vector quantizing means so as to cause said scalar quantizing means to perform scalar quantization within the quantizing ranges.

6. A speech parameter coding apparatus, comprising:

means for dividing an input speech signal into a plurality of frames;

means for finding a predetermined order number of spectrum parameters from the speech signal;

means for vector quantizing the spectrum parameters;

means for scalar quantizing difference signals between the spectrum parameters and the vector quantized values;

means for producing a plurality of candidate quantizing values for each order number of the spectrum parameters at said scalar quantizing means;

means for modifying, using each of the candidates, a quantizing range of an adjacent spectrum parameter; and

means for accumulating a predetermined order number of quantizing distortions by the quantized values and producing a quantized value sequence which minimizes the accumulated distortion.

7. A speech parameter coding apparatus, comprising:

means for dividing an input speech signal into a plurality of frames;

means for finding a predetermined order number of spectrum parameters from the speech signal;

means for vector quantizing the spectrum parameters;

means for scalar quantizing difference signals between the spectrum parameters and the vector quantized values;

means for determining a predetermined order number of quantizing ranges for the scalar quantization for each of a predetermined number of codevectors from said vector quantizing means;

means for producing a plurality of candidate quantizing values for each order number of the spectrum parameters for the scalar quantization;

means for modifying, using each of the candidates, a quantizing range of an adjacent spectrum parameter; and

means for accumulating a predetermined order number of quantizing distortions by the quantizing values and producing a quantized value sequence which minimizes the accumulated distortion.