Digital speech coder having improved long term lag parameter determination

Claim 2

A digital speech encoding method that produces parameters representative of samples of digitized speech comprising the steps of: storing a plurality of excitation signals; filtering said excitation signals using a long term filter to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; generating error signals based upon the difference between said filtered signals and a sample of digitized speech; selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; generating said time lag parameter by: calculating correlation values based on trial time lags of different lengths; evaluating said correlation values and selecting a predetermined number of the trial time lags having the larger of said correlation values, the maximum value of said number having a corresponding lag D p ; determining if at least one of said number of trial time lags is harmonically related to lag D p , if at least one of said predetermined number of lags is harmonically related to lag D p selecting the smallest to said harmonically related lags for use as said lag parameter, if none of said number of trial time lags are harmonically related to lag D p selecting D p as said lag parameter; filtering a sample of digitized speech using the long term filter with the time lag filter characteristic controlled by said selected lag parameter.

Claim 3

The method according to claim 1 wherein said determining step further comprises the steps of defining a range of lags consisting of continuous lags and including lag D p and said calculating step includes calculating correlation values based on said continuous lags, and making said harmonically related determination based on an integer multiple of said number of lags being within said range.

Claim 4

A digital speech encoder that produces parameters representative of samples of digitized speech comprising: codebook means for storing a plurality of excitation signals; long term filter which filters said excitation signals to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; means for generating error signals based upon the difference between said filtered signals and a sample of digitized speech; means for selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; means for generating said time lag parameter comprising: means for calculating correlation values based on trial time lags of different lengths; means for evaluating said correlation values and means for selecting a predetermined number of the trial time lags having the larger of said correlation values, the maximum value of said number having a corresponding lag D p ; means for determining if at least one of said predetermined number of trial time lags is harmonically related to lag D p ; means for selecting the smallest of said harmonically related lags as said lag parameter if a harmonically related lag exists, if none of said number of trial time lags are harmonically related to lag D p selecting D p as said lag parameter; said long term filter filtering samples of digitized speech using the time lag filter characteristic controlled by said selected lag parameter.

Claim 5

A digital speech encoder that produces parameters representative of samples of digitized speech comprising: codebook means for storing a plurality of excitation signals; long term filter which filters said excitation signals to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; means for generating error signals based upon the difference between said filtered signals and a sample of digitized speech; means for selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; means for generating said time lag parameter comprising: means for calculating correlation values based on trial time lags of different lengths; means for evaluating said correlation values and means for selecting a predetermined number of the trial time lags having the larger of said correlation values, the maximum value of said number having a corresponding lag D p ; means for determining if at least one of said predetermined number of trial time lags is harmonically related to lag D p ; means for selecting the smallest of said harmonically related lags as said lag parameter if a harmonically related lag exists, if none of said number of trial time lags are harmonically related to lag D p selecting D p as said lag parameter; said long term filter filtering samples of digitized speech using the time lag filter characteristic controlled by said selected lag parameter.

Claim 6

The encoder according to claim 3 further comprising means for defining a range of lags consisting of continuous lags and including lag D p , said calculating means calculating correlation values based on said continuous lags, and said means for determining making said determination dependent on whether an integer multiple of one of said number of lags is within said range.

Claim 7

A digital speech encoding method that produces parameters representative of samples of digitized speech comprising the steps of: storing a plurality of excitation signals; filtering said excitation signals using a long term filter to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; generating error signals based upon the difference between said filtered signals and a sample of digitized speech; selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; generating said time lag parameter by: calculating an open loop lag parameter L open ; conducting a predetermined series of tests of closed loop lag parameters dependent on the value of open loop lag parameter L open to determine the error associated with each closed loop lag parameter tested; selecting the closed loop lag parameter with the smallest error as said time lag parameter; filtering samples of digitized speech using the long term filter with the time lag filter characteristic controlled by lag parameter L.

Claim 8

A digital speech encoding method that produces parameters representative of samples of digitized speech comprising the steps of: storing a plurality of excitation signals; filtering said excitation signals using a long term filter to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; generating error signals based upon the difference between said filtered signals and a sample of digitized speech; selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; generating said time lag parameter by: calculating an open loop lag parameter L open ; conducting a predetermined series of tests of closed loop lag parameters dependent on the value of open loop lag parameter L open to determine the error associated with each closed loop lag parameter tested; selecting the closed loop lag parameter with the smallest error as said time lag parameter; filtering samples of digitized speech using the long term filter with the time lag filter characteristic controlled by lag parameter L.

Claim 9

The method according to claim 5 wherein said step of conducting tests comprises the steps of testing closed loop lag parameters harmonically related to open loop lag parameter L open .

Claim 10

The method according to claim 6 wherein the number of harmonics of L open tested depends on the value of parameter L open relative to a predetermined maximum value.

Claim 11

The method according to claim 6 wherein said conducting step conducts said predetermined series of closed loop lag parameter tests using undecimated samples of digitized speech.

Claim 12

The method according to claim 5 wherein said conducting step conducts said predetermined series of closed loop lag parameter tests using undecimated samples of digitized speech.

Claim 13

The method according to claim 9 wherein said calculating of the open loop lag parameter L open is based on a decimated sample of the digitized speech.

Claim 14

A digital speech encoder that produces parameters representative of samples of digitized speech comprising: codebook means for storing a plurality of excitation signals; long term filter which filters said excitation signals to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; means for generating error signals based upon the difference between said filtered signals and a sample of digitized speech; means for selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; means for generating said time lag parameter comprising: means for calculating an open loop lag parameter L open ; means for conducting a predetermined series of tests of closed loop lag parameters dependent on the value of open loop lag parameter L open to determine the error associated with each test; means for selecting said closed loop lag parameter with the smallest error as said time lag parameter; said filter filtering samples of digitized speech with the time lag filter characteristic controlled by said time lag parameter.

Claim 15

A digital speech encoder that produces parameters representative of samples of digitized speech comprising: codebook means for storing a plurality of excitation signals; long term filter which filters said excitation signals to produce corresponding filtered signals, said long term filter having a time lag filter characteristic controlled by a time lag parameter; means for generating error signals based upon the difference between said filtered signals and a sample of digitized speech; means for selecting the excitation signal corresponding to the smallest error signal for use with said sample of digitized speech; means for generating said time lag parameter comprising: means for calculating an open loop lag parameter L open ; means for conducting a predetermined series of tests of closed loop lag parameters dependent on the value of open loop lag parameter L open to determine the error associated with each test; means for selecting said closed loop lag parameter with the smallest error as said time lag parameter; said filter filtering samples of digitized speech with the time lag filter characteristic controlled by said time lag parameter.

Claim 16

The encoder according to claim 11 wherein said means of conducting tests comprises means for testing closed loop lag parameters harmonically related to open loop lag parameter L open .

Claim 17

The encoder according to claim 12 wherein the number of harmonics of L open tested depends on the value of parameter L open relative to a predetermined maximum value.

Claim 18

The encoder according to claim 12 wherein said conducting means conducts said predetermined series of closed loop lag parameter tests using undecimated samples of digitized speech.

Claim 19

The encoder according to claim 11 wherein said conducting means conducts said predetermined series of closed loop lag parameter tests using undecimated samples of digitized speech.

Claim 20

The encoder according to claim 15 wherein said means for calculating calculates the open loop lag parameter L open based on a decimated sample of the digitized speech.