Claims Analysis
Total Claims: 10
Claim 1
Independent Claim
186 words
A multi-pulse type vocoder comprising: first means for extracting spectrum information of an input speech signal X(n) in an analysis frame; second means for developing an impulse response h(n) of a filter specified by said spectrum information; third means for developing a cross-correlation series φ hx (mi) between said input speech signal X(n) and said impulse response h(n) at a time lag mi within a predetermined time range, n representing a sampling time point; fourth means for developing an auto-correlation series R hh (n) of said impulse response h(n) and a normalized auto-correlation series R hh ,(n) normalized by a power of the auto-correlation series R hh (n) ; fifth means for determining the most similar portion of said cross-correlation series φ hx to the auto-correlation series R hh (n) ; sixth means for developing a similarity between the cross-correlation series φ hx (n) and the normalized auto-correlation series R hh '.sub.(n) ; and seventh means for providing a pulse having the maximum similarity value and a time position thereat of the most similar portion of said cross-correlation series φ hx as one of said multi-pulses.
Claim 2
Dependent Claim
186 words
A multi-pulse type vocoder comprising: first means for extracting spectrum information of an input speech signal X(n) in an analysis frame; second means for developing an impulse response h(n) of a filter specified by said spectrum information; third means for developing a cross-correlation series φ hx (mi) between said input speech signal X(n) and said impulse response h(n) at a time lag mi within a predetermined time range, n representing a sampling time point; fourth means for developing an auto-correlation series R hh (n) of said impulse response h(n) and a normalized auto-correlation series R hh ,(n) normalized by a power of the auto-correlation series R hh (n) ; fifth means for determining the most similar portion of said cross-correlation series φ hx to the auto-correlation series R hh (n) ; sixth means for developing a similarity between the cross-correlation series φ hx (n) and the normalized auto-correlation series R hh '.sub.(n) ; and seventh means for providing a pulse having the maximum similarity value and a time position thereat of the most similar portion of said cross-correlation series φ hx as one of said multi-pulses.
Claim 3
Dependent Claim
49 words
The multi-pulse type vocoder as defined in claim 1, further comprising eighth means for correcting said cross-correlation series φ hx by subtracting a weighted auto-correlation series by the maximum similarity from the most similar portion of said cross-correlation series and providing the corrected cross-correlation series to said fifth means.
Claim 4
Dependent Claim
21 words
The multi-pulse type vocoder as defined in claim 1, wherein said first means includes means for extracting a linear prediction parameter.
Claim 5
Dependent Claim
32 words
The multi-pulse type vocodor as defined in claim 1, wherein said first means includes means for weighting said input speech signal and extracting the spectrum information from the weighted input speech signal.
Claim 6
Dependent Claim
58 words
The multi-pulse type vocoder as defined in claim 1, wherein said sixth means includes a similarity calculator calculating the similarity b mi according to the following expression: ##EQU15## where S represents time point; m i , time point shifted from the S; and N R , predetermined effective duration time of the normalized autocorrelation series R hh '.sub.(S).
Claim 7
Dependent Claim
59 words
The multi-pulse type vocoder as defined in claim 1, wherein said sixth means includes a similarity calculator calculating the similarity C mi according to the following expression: ##EQU16## where S represents time point; m i , time point shifted from the S; and N R , a predetermined effective duration time of the normalized auto-correlation series R hh '.sub.(S).
Claim 8
Dependent Claim
42 words
The multi-pulse type vocoder as defined in claim 1, wherein said first means includes means for extracting a pitch of said input speech signal and supplying the pitch to said sixth means to determine the total number of multi-pulses to be provided.
Claim 9
Dependent Claim
36 words
The multi-pulse type vocoder as defined in claim 1, wherein said seventh means includes means for determining a quotient obtained by dividing said analysis frame period by said pitch period as the total number of multi-pulses.
Claim 10
Dependent Claim
30 words
The multi-pulse type vocoder as defined in claim 1, further comprising a synthesis filter operable by the spectrum information from said first means and the multi-pulses from said sixth means.