CLAIMS

What is claimed is :

1. Apparatus for removing the periodicity from a speech signal in a transform coder prior to the quantization of said speech signal, which speech signal is a sampled time domain speech signal composed of information samples, said transform coder sequentially segregating said speech signal into blocks of information samples, comprising:

filter means for filtering each of said blocks of samples to remove spurious peaks;

clipping means for enhancing certain samples contained in said blocks necessary to determine pitch;

function means for generating an autocorrelation function of each of said sample blocks after it has been operated upon by said clipping means;

pitch means for determining the maximum value in said autocorrelation function;

LTP means for determining long term predictor parameters in relation to said maximum value and other values contained in said autocorrelation function; and

a difference means for calculating a periodicity value for each sample in said block wherein the calculation of said periodicity value is based upon said maximum value and said long term predictor parameter and for generating a revised block of difference samples by subtracting said periodicity value from the corresponding sample.

2. The apparatus of claim 1, wherein said filter means comprises a low pass filter having a frequency range from approximately 0 Hz to approximately 1650 Hz.

3. The apparatus of claim 1, wherein said filter means comprises an eight-tap finite impulse response filter having 3 dB cutoff frequencies at 1800 Hz and 2400 Hz.

4. The apparatus of claim 1, further comprising calculating means for calculating pitch gain in relation to said autocorrelation function, and threshold means for determining when said pitch gain exceeds a reference value.

5. The apparatus of claim 4, wherein said reference value is 0.25.

6. The apparatus of claim 1, wherein said clipping means comprises:

dividing means for dividing said blocks into a plurality of smaller blocks;

search means for searching said smaller blocks for the maximum value in each of said smaller blocks;

enhancing means for identifying those samples in each of said smaller blocks which exceed a threshold value; and

combination means for combining all samples identified by said enhancing means into a single block.

7. The apparatus of claim 6, wherein said dividing means divides said blocks into two smaller blocks.

8. The apparatus of claim 7, wherein said enhancing means identifies samples according to the following formula:

c(n) = +1 s(n) ≥ T_{c} (1)

= -1 s(n) ≤ T_{c}

= 0 otherwise

where T_{c} = amplitude threshold.

9. The apparatus of claim 1, wherein said difference signal is generated according to the following formula:

e(n) = s(n) - β_{1}. s (n-K) - β_{2}.s(n-M-1)

where M = the pitch; and

β_{1} and β_{2} = the long term predictor parameters;

and wherein the long term predictor parameters are determined according to the formula:

β1 = R(O)R(M) - R(1)R(M+1)_{:}

[R(O)^{2} - R(1)^{2}]

β2 = R(O)R(M+1) - R(1)R(M)

[R(0)^{2} - R(1)^{2}] where R(O) - the ACF value at the origin;

R(l) = the ACF value at 1;

R(M-1) = the ACF value at the pitch -1;

R(M) = the ACF value at the pitch; and

R(M+1) = the ACF value at the pitch +1.

10. The apparatus of claim 9, further comprising a comparator for comparing the sum of β_{1} and β_{2} to a reference value.

11. The apparatus of claim 10, wherein said reference value is 8/9.

12. The apparatus of claim 10, further comprising scaling means for scaling β_{1} and β_{2} so that β_{1} + β_{z} = said reference value.

13. The apparatus of claim 9, further comprising a comparator for determining whether R(M+1) is greater than R(M-1).

14. The apparatus of claim 13, further comprising means for substituting the value of R(M-1) for R(M+1) prior to the calculation of β_{1} and β_{2} interchange means for interchanging the values calculated for β_{1} and β_{2} and decrement means for decrementing Pitch (M) by one prior to transmission.

15. Apparatus for removing the periodicity from a speech signal in a transform coder prior to the quantization of said speech signal, which speech signal is a sampled time domain speech signal composed of information samples, said transform coder sequentially segregating said speech signal into blocks of information samples, comprising:

pitch means for determining the pitch in each of said sample blocks;

LTP means for determining a long term prediction parameter for each of said blocks based on the pitch determined for each block;

a difference means for calculating a periodicity value for each sample in said block wherein the calculation of said periodicity value is based upon said pitch and said long term predictor parameter and for generating a revised block of difference samples by subtracting said periodicity value from the corresponding sample; and

adaptive transform coding means for performing adaptive transform coding on each of said difference blocks.

16. A method for removing the periodicity from a speech signal in a transform coder prior to the quantization of said speech signal, which speech signal is a sampled time domain speech signal composed of information samples, said transform coder sequentially segregating said speech signal into blocks of information samples, comprising the steps of:

filtering each of said blocks of samples to remove spurious peaks;

enhancing certain samples contained in said blocks necessary to determine pitch;

generating an autocorrelation function of each of said sample blocks after it has been operated upon by said clipping means;

determining the pitch by determining the maximum value in said autocorrelation function;

determining long term predictor parameter in relation to said maximum value and other values contained in said autocorrelation function;

calculating a periodicity value for each sample in said block wherein the calculation of said periodicity value is based upon said maximum value and said long term predictor parameter; and

generating a revised block of difference samples by subtracting said periodicity value from the corresponding sample.

17. The method of claim 16, wherein said step of filtering comprises providing a low pass filter having a frequency range from approximately 0 Hz to approximately 1650 Hz.

18. The method of claim 16, wherein said step of filtering comprises providing an eight-tap finite impulse response filter having 3 dB cutoff frequencies at 1800 Hz and 2400 Hz.

19. The method of claim 16, further comprising the steps of calculating pitch gain in relation to said autocorrelation function, and determining when said pitch gain exceeds a reference value.

20. The method of claim 19, wherein said reference value is 0.25.

21. The method of claim 16, wherein the step of enhancing comprises the steps of:

dividing said blocks into a plurality of smaller blocks;

searching said smaller blocks for the maximum value in each of said smaller blocks;

identifying those samples in each of said smaller blocks which exceed a threshold value; and

combining all samples identified by said enhancing means into a single block.

22. The method of claim 21, wherein the step of dividing comprises dividing said blocks into two smaller blocks.

23. The method of claim 22, wherein said step of enhancing identifies samples according to the following formula: c(n) = +1 s(n) > T_{c} (1)

= -1 s(n) ≤ T_{c}

= 0 otherwise

where T_{c} = amplitude threshold.

24. The apparatus of claim 16, wherein said step of generating a difference signal is accomplished according to the following formula: e(n) = s(n) - β_{1}.s(n-M) - β_{2}.s(n-M-1)

where M = the pitch; and

β_{1} and β_{2} = the long term predictor parameters;

and wherein the step of determining said long term predictor parameters are determined according to the formula:

β1 = R(0)R(M) - R(1)R(M+1),

[R(0)^{2} - R(1)^{2}]

β2 = R(O)R(M+1) - R(1)R(M)

[R(0)^{2} - R(1)^{2}]

where R(0) = the ACF value at the origin;

R(1) = the ACF value at 1;

R(M-1) = the ACF value at the pitch -1;

R(M) = the ACF value at the pitch; and

R(M+1) = the ACF value at the pitch +1.

25. The method of claim 24, further comprising the step of comparing the sum of β_{1} and β_{2} to a reference value.

26. The method of claim 25, wherein said reference value is 8/9.

27. The method of claim 25, further comprising the step of scaling β_{1} and β_{2} so that β_{1} + β_{2} = said reference value.

28. The method of claim 24, further comprising the step of determining whether R(M+1) is greater than R(M-1).

29. The method of claim 28, further comprising the steps of substituting the value of R(M-l) for R(M+1) prior to the calculation of β_{1} and β_{2}, interchanging the values calculated for β_{1} and β_{2} and decrementing Pitch (M) by one prior to transmission.

30. A method for removing the periodicity from a speech signal in a transform coder prior to the quantization of said speech signal, which speech signal is a sampled time domain speech signal composed of information samples, said transform coder sequentially segregating said speech signal into blocks of information samples, comprising the steps of:

determining the pitch in each of said sample blocks; determining a long term prediction parameter for each of said blocks based on the pitch determined for each block;

calculating a periodicity value for each sample in said block wherein the calculation of said periodicity value is based upon said pitch and said long term predictor parameter;

generating a revised block of difference samples by subtracting said periodicity value from the corresponding sample; and

performing adaptive transform coding on each of said difference blocks.

31. Apparatus for decoding a coded speech signal wherein such coded speech signal includes sequential blocks of transform coefficients which have been quantized in relation to a bit allocation signal generated in relation to scaled spectral envelope information and side information including pitch, long term predictor parameter and linear prediction coefficients, representative of the variance of said quantized transform coefficients, comprising:

envelope generation means for generating the spectral envelope of each of said blocks of information samples based upon said linear prediction coefficients;

bit allocation means for generating a bit allocation signal in relation to said spectral envelope;

de-quantization means for de-quantizing said transform coefficients in response to said bit allocation signal and for generating blocks of de-quantized transform coefficients;

inverse transformation means for transforming said de-quantized transform coefficients from said transform domain into said time domain; and

summation means for calculating a periodicity value for each sample in said block wherein the calculation of said periodicity value is based upon said pitch and said long term predictor parameter and for generating a revised block of difference samples by adding said periodicity value to the corresponding sample.