Efficient calculation of autocorrelation coefficients for CELP vocoder adaptive codebook

• US 5,179,594 A
• Filed: 06/12/1991
• Issued: 01/12/1993
• Est. Priority Date: 06/12/1991
• Status: Expired due to Term

• Current Assignee: General Dynamics C4 Systems Incorporated (General Dynamics Corporation)
• Original Assignee: Motorola, Inc. (Motorola Solutions, Inc.)
• Inventors: Barron, David L., Yip, William C.
• Primary Examiner(s): KEMENY, EMANUEL
• Application Number:
  US07/714,409
• Time in Patent Office: 580 Days
• Field of Search: 381/29-40
• US Class Current: 704/217
• CPC Class Codes: G10L 19/12 the excitation function being a code excitation, e.g. in code excited linear prediction [CELP] vocoders
• CPC Class Codes: G10L 2019/0002 Codebook adaptations
• CPC Class Codes: G10L 2019/0014 Selection criteria for distances
• CPC Class Codes: G10L 25/06 the extracted parameters being correlation coefficients
• CPC Class Codes: G10L 25/18 the extracted parameters being spectral information of each sub-band

Abstract

A new way of determining autocorrelation coefficients for adaptive codebook vectors for CELP coding of speech simplifies and improves the accuracy of the autocorrelation coefficient determination for the situation where the codebook vector length being analyzed is less than a speech frame length. This is important in synthesizing short pitch period speech. Copy-up of the shortened codebook vector to equal the frame length is not needed and autocorrelation coefficient errors associated with copy-up are avoided. The improved system relies on calculating autocorrelation coefficients of the first (shortest) vector and then obtaining subsequent autocorrelation coefficients for successive vectors of increasing length by a simple end correction technique until the vector length equals the frame length. The autocorrelation coefficients are scaled by multiplying them by the ratio of the frame length to the vector length.

Claims

1. A method for CELP coding speech employing autocorrelation coefficients of vectors of an adaptive codebook of vector length N wherein analysis initially utilizes a subset of samples M<

N with a speech analysis frame of length L, comprising;

calculating autocorrelation coefficients U_k (m) of a first vector C_k (n) of length M, where k=1 and m is an autocorrelation lag index and n is an index of the successive samples in the codebook vector, according to, ##EQU18## for m=0 to T<

M;

calculating the autocorrelation coefficients U_k (m) of the remaining codebook vectors incrementally where k≧

2 according to,

space="preserve" listing-type="equation">U'"'"'.sub.k (m)=[U'"'"'.sub.k-1 (m)+C.sub.k (M+k-1)C.sub.k (M+k-1+m)](3) ##EQU19## for m=0 to T<

M;

repeating the second calculating step until (M+k-1)=L; and

using the above-determined autocorrelation coefficients in determining which of the codebook vectors C_k (n) produces the least error when compared to input speech.

2. The method of claim 1 wherein T=M-1.

3. The method of claim 1 further comprising determining the coefficients according to Eq. (1) of claim 1, storing such coefficients in a memory and then scaling the stored coefficients according to Eq. (2) of claim 1 and transferring the scaled coefficients to an output.

4. The method of claim 1 further comprising determining the coefficients for k=1 according to Eq. (1) of claim 1, storing such coefficients in a memory and then using the stored coefficients for k=1 to determine the coefficients for k=2 according to Eq. (3) of claim 1 and then updating the stored coefficients to have the values so determined, and then scaling the stored coefficients according to Eq. (4) of claim 1 and transferring the scaled coefficients to an output.

5. The method of claim 4 further comprising repeating the determining, storing, updating, scaling and transferring steps for each subsequent value of k until (M+k-1)=L.

6. A method for CELP coding speech employing autocorrelation coefficients of vectors of an adaptive codebook identified by an index k, wherein analysis by synthesis initially utilizes M<

L codebook values where L is the speech analysis frame length and m is an index running from 0 to M-1 describing the autocorrelation lag, comprising;

calculating m=0 to m-1 autocorrelation coefficients of a first codebook vector k having n=1 to M values therein where n is an index of the code vector values;

placing the m=0 to M-1 calculated autocorrelation coefficients in a temporary store;

scaling the coefficients in the temporary store by a multiplying factor L/M and transferring the result to an output;

multiplying codebook values for n=M+j where j=1 by codebook values for n=M+j down to n=1 and adding the products to the m=0 to M-1 autocorrelation coefficients, respectively, from the temporary store to produce a result;

replacing the autocorrelation coefficients in the temporary store by the result;

scaling the coefficients in the temporary store by a multiplying factor L/(M+j) and transferring the result to the output;

repeating the multiplying, replacing, scaling and transferring steps for j=2 to j=k-1 and k=(L+1-M); and

using the autocorrelation coefficients transferred to the output to determining which of the codebook vectors provides better CELP coding of speech.

7. An apparatus for CELP coding of speech employing autocorrelation coefficients of vectors of an adaptive codebook wherein analysis initially utilizes a subset of samples M in connection with a speech analysis frame of length L>

M, comprising;

means for determining autocorrelation coefficients U_k'"'"' (m) of a first vector C_k (n) of length M, where k=1 and m is an autocorrelation lag index and n is an index of successive samples in the codebook vector, according to, ##EQU20## for m=0 to T<

M;

means for determining autocorrelation coefficients U_k (m) of remaining codebook vectors incrementally where k≧

2 according to,

space="preserve" listing-type="equation">U'"'"'.sub.k (m)=[U'"'"'.sub.k-1 (m)+C.sub.k (M+k-1)C.sub.k (M+k-1+m)](2)

for m=0 to T<

M until (M+k-1)=L;

means for scaling the result of Eq. (1) according to ##EQU21## and scaling the result for Eq. (2) according to ##EQU22## for m=0 to T<

M to produce a result for each m and each k; and

means for using the result to evaluate which codebook vector provides a least error compared to input speech.

8. The apparatus of claim 7 wherein the first means comprises memory means for receiving n=1 to n=M samples from a first codebook vector and means for multiplying vector values from the memory means by vector values from the memory means delayed by successive increments m=0 to M-1 to produce products for each value of n and m.

9. The apparatus of claim 8 wherein the first means further comprises, accumulator means for adding the multiplied values for n=1 to M to produce a sum of the products for each value of m=0 to M-1.

10. The apparatus of claim 9 further wherein the first means comprises switch means for permitting transfer of the sum of the products for each value of m to an updateable store.

11. The apparatus of claim 10 wherein the second means comprises memory means for receiving a vector value having an index one greater than a largest index of vector values used to produce the sum of the products in the updateable store.

12. The apparatus of claim 11 wherein the second means further comprises means for multiplying the vector value of one greater index by vector values of successively smaller indices to produce further vector value products thereof.

13. The apparatus of claim 12 wherein the second means further comprises means for adding the further vector value products for each index n to a corresponding sum of the products for each value of m in the updateable store, to produce new sums for each value of m which replace the sum of the products previously resident in the updateable store.

14. The apparatus of claim 12 further comprising means for coupling the updateable store to the scaling means so that the scaling means operates on each value of the sums in the updateable store.