Celp voice encoder

• US 6,804,639 B1
• Filed: 06/21/2000
• Issued: 10/12/2004
• Est. Priority Date: 10/27/1998
• Status: Expired due to Term

• Current Assignee: III Holdings 12, LLC (Intellectual Ventures LLC)
• Original Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Inventors: Ehara, Hiroyuki
• Primary Examiner(s): Knepper, David D.
• Application Number:
  US09/582,039
• Time in Patent Office: 1,574 Days
• Field of Search: 704/219-230
• US Class Current: 704/223
• CPC Class Codes: G10L 19/08 Determination or coding of the excitation function; Determination or coding of the long-term prediction parameters

Abstract

A CELP type speech coder performs quantization of pitch differential value on pitch information between subframes. The coder limits the number of preliminary selected candidates using threshold processing. The coder includes specialized pitches for a subframe on which quantization of pitch differential value is not applied. When pitch preliminary selection is performed on such a subframe, the coder limits the number of preliminarily selected candidates using threshold processing to avoid outputting, as a preliminarily selected candidate, the above-mentioned specialized pitches. The coder improves the accuracy of the pitch search (adaptive codebook search) while avoiding adverse effects on the quantization of pitch differential value.

Claims

1. A CELP type speech coding apparatus comprising:

parameter coding means for coding a parameter representative of a spectral characteristic of a speech;

periodicity coding means for coding a periodicity of an excitation vector using an adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using a random codebook storing predetermined excitation vectors,

wherein said periodicity coding means having;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

2. The CELP type speech coding apparatus according to claim 1, wherein said pitch candidate selecting means comprising:

auto-correlation function calculation means for obtaining a normalized auto-correlation function using a previous weighted input speech signal and a new weighted input speech signal;

sorting means for sorting auto-correlation functions into a plurality of ranges corresponding to the pitch in the adaptive codebook;

a plurality of search means for searching a maximum value of an auto-correlation function in a respective range, and a pitch corresponding to the auto-correlation function;

threshold calculating means for obtaining a predetermined threshold using the auto-correlation functions searched in the plurality of search means; and

selecting means for selecting a pitch corresponding to the auto-correlation function exceeding the threshold among the auto-correlation functions searched in the plurality of search means.

3. A speech signal transmission apparatus having a speech input apparatus that converts a speech signal into an electric signal, a CELP type speech coding apparatus that performs coding processing on a signal output from the speech input apparatus, and a transmission apparatus that transmits a coded signal output from the CELP type speech coding apparatus, said CELP type speech coding apparatus comprising:

parameter coding means for coding a parameter representative of a spectral characteristic of a speech;

periodicity coding means for coding a periodicity of an excitation vector using an adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using a random codebook storing predetermined excitation vectors,

wherein said periodicity coding means having;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

4. A speech coding/decoding apparatus comprising:

a CELP type speech decoding apparatus having

means for decoding coded information of a parameter representative of a spectral characteristic of a speech;

means for decoding an adaptive code vector using an adaptive codebook storing previously generated excitation vectors;

means for decoding a random code vector using a random codebook storing predetermined excitation vectors; and

means for decoding respective amplitudes of an adaptive codebook component and random codebook component; and

a CELP type coding apparatus having;

parameter coding means for coding the parameter representative of the spectral characteristic of the speech;

periodicity coding means for coding a periodicity of the excitation vector using the adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using the random codebook storing predetermined excitation vectors,

wherein said periodicity coding means including;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

5. A speech signal transmission/reception apparatus having a speech coding/decoding apparatus, a speech input apparatus that converts a speech signal into an electric signal, a transmission apparatus that transmits a coded signal output from a CELP type coding apparatus, and a speech output apparatus that converts a decoded signal into the speech signal, said speech coding/decoding apparatus comprising:

a CELP type speech decoding apparatus having;

means for decoding coded information of a parameter representative of a spectral characteristic of a speech;

means for decoding an adaptive code vector using an adaptive codebook storing previously generated excitation vectors;

means for decoding a random code vector using a random codebook storing predetermined excitation vectors; and

means for decoding respective amplitudes of an adaptive codebook component and random codebook component; and

a CELP type coding apparatus having;

parameter coding means for coding the parameter representative of the spectral characteristic of the speech;

periodicity coding means for coding the periodicity of the excitation using the adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using the random codebook storing predetermined excitation vectors,

wherein said periodicity coding means including;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

6. A base station apparatus provided with a speech signal transmission/reception apparatus, said speech signal transmission/reception apparatus having a speech coding/decoding apparatus, a speech input apparatus that converts a speech signal into an electric signal, a transmission apparatus that transmits a coded signal output from a CELP type coding apparatus, and a speech output apparatus that converts a decoded signal into the speech signal, said speech coding/decoding apparatus comprising:

a CELP type speech decoding apparatus having

means for decoding coded information of a parameter representative of a spectral characteristic of a speech;

means for decoding an adaptive code vector using an adaptive codebook storing previously generated excitation vectors;

means for decoding a random code vector using a random codebook storing predetermined excitation vectors; and

means for decoding respective amplitudes of an adaptive codebook component and random codebook component; and

a CELP type coding apparatus having;

parameter coding means for coding the parameter representative of the spectral characteristic of the speech;

periodicity coding means for coding a periodicity of an excitation vector using the adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using the random codebook storing predetermined excitation vectors,

wherein said periodicity coding means including;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

7. A communication terminal apparatus provided with a speech signal transmission/reception apparatus, said speech signal transmission/reception apparatus having a speech coding/decoding apparatus, a speech input apparatus that converts a speech signal into an electric signal, a transmission apparatus that transmits a coded signal output from a CELP type coding apparatus, and a speech output apparatus that converts a decoded signal into the speech signal, said speech coding/decoding apparatus comprising:

a CELP type speech decoding apparatus having

means for decoding coded information of a parameter representative of a spectral characteristic of a speech;

means for decoding an adaptive code vector using an adaptive codebook storing previously generated excitation vectors;

means for decoding a random code vector using a random codebook storing predetermined excitation vectors; and

means for decoding respective amplitudes of an adaptive codebook component and random codebook component; and

a CELP type coding apparatus having;

parameter coding means for coding the parameter representative of the spectral characteristic of the speech;

periodicity coding means for coding a periodicity of an excitation vector using the adaptive codebook storing previously generated excitation vectors; and

excitation component coding means for coding an excitation vector component that cannot be represented with the adaptive codebook, using the random codebook storing predetermined excitation vectors,

wherein said periodicity coding means including;

pitch candidate selecting means for performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

8. A CELP type speech coding method comprising:

the parameter coding step of coding a parameter representative of a spectral characteristic of a speech;

the periodicity coding step of coding a periodicity of an excitation vector using an adaptive codebook storing previously generated excitation vectors; and

the excitation component coding step of coding an excitation vector component that cannot be represented with the adaptive codebook, using a random codebook storing predetermined excitation vectors,

wherein said periodicity coding step having;

the pitch candidate selecting step of performing preliminary selection of a pitch for the adaptive codebook on a subframe on which the pitch for the adaptive codebook is not subjected to quantization of pitch differential value, among subframes obtained by dividing unit frame, and selecting at least one pitch candidate adaptively.

9. The CELP type speech coding method according to claim 8, wherein said pitch candidate selecting step having the steps of:

obtaining a normalized auto-correlation function using a previous weighted input speech signal and a new weighted input speech signal;

sorting auto-correlation functions into a plurality of ranges corresponding to the pitch in the adaptive codebook;

searching a maximum value of an auto-correlation function in a respective range and a pitch corresponding to the auto-correlation function;

obtaining a predetermined threshold using searched auto-correlation functions; and

selecting a pitch corresponding to the auto-correlation function exceeding the threshold among the searched auto-correlation functions.

10. A recording medium, readable by a computer, having a program recorded therein executable by the computer, said program comprising the procedures of:

obtaining a normalized auto-correlation function using a previous weighted input speech signal and a new weighted input speech signal;

sorting auto-correlation functions into a plurality of ranges corresponding to the pitch in the adaptive codebook;

searching a maximum value of an auto-correlation function in a respective range and a pitch corresponding to the auto-correlation function,

obtaining a predetermined threshold using searched auto-correlation functions; and

selecting a pitch corresponding to the auto-correlation function exceeding the threshold among the searched auto-correlation functions.

11. A CELP type speech coding apparatus comprising:

an analyzer that divides a speech signal into frames of predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into a plurality of subframes, to encode

a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector,

wherein said periodicity coder, when performing differential coding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

12. The CELP type speech coding apparatus according to claim 11, further comprising:

an autocorrelation function calculator that obtains a normalized autocorrelation function from a previous weighted input speech signal and a new weighted input speech signal;

a sorter that sorts autocorrelation functions into a plurality of ranges according to a pitch in the adaptive codebook;

a plurality of search devices that, for each range, search for a maximum value of the autocorrelation function and the pitch corresponding to said autocorrelation function;

a threshold calculator that calculates a predetermined threshold from the maximum value of the autocorrelation function; and

a selector that, from the pitches searched by the plurality of search devices, selects the pitch corresponding to the autocorrelation function that exceeds said threshold.

13. A speech signal transmitting apparatus comprising:

a speech input apparatus that converts a speech signal to an electric signal;

a CELP type speech coder that encodes the electric signal output from said speech input apparatus; and

a transmitting apparatus that transmits the coded signal output from said CELP type speech coding apparatus,

wherein said CELP type speech signal coder comprises;

an analyzer that divides the speech signal into frames of a predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into a plurality of subframes, to encode a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector, and

wherein said periodicity coder, when performing differential encoding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of the autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

14. A speech coding/decoding apparatus comprising:

a CELP type speech coder comprising an analyzer that divides a speech signal into frames of a predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into a plurality of subframes, to encode a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector, and

a CELP type speech decoder that decodes coded information of a parameter representing a spectral characteristic of speech, decodes an adaptive code vector using an adaptive codebook that stores a previously generated excitation vector, decodes a random code vector using a random codebook that stores a predetermined excitation vector, and decodes an amplitude of an adaptive codebook component and a random codebook component,

wherein the periodicity coder in said CELP type speech coder, when performing differential encoding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of the autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

15. A speech signal transmitting/receiving apparatus comprising:

a speech input apparatus that converts a speech signal to an electric signal;

a CELP type speech coder that encodes the signal output from said speech input apparatus;

a transmitting apparatus that transmits the coded signal output from said CELP type speech coder to a communication partner;

a receiving apparatus that receives a signal transmitted from the communication partner;

a CELP type speech decoder that decodes the received signal output from the receiving apparatus; and

a speech output apparatus that converts the decoded signal output from said CELP type speech decoder to a speech signal and outputs said speech signal,

wherein said CELP type speech coder comprises;

an analyzer that divides the speech signal into frames of a predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into a plurality of subframes, to encode a periodicity of excitation vector using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector, and

wherein said periodicity coder, when performing differential encoding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of the autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

16. A base station apparatus comprising:

a CELP type speech coder that encodes an electric signal to which a speech signal has been converted;

a transmitting apparatus that transmits the coded signal output from said CELP type speech coder to a communication partner;

a receiving apparatus that resolves a signal transmitted from the communication partner; and

a CELP type speech decoder that decodes the received signal output from said receiving apparatus, wherein said CELP type speech coder comprises;

an analyzer that divides a speech signal into frames of a predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into a plurality of subframes, to encode a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector,

wherein said periodicity coder, when performing differential encoding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of the autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

17. A communication terminal apparatus comprising:

a speech input apparatus that converts a speech signal to an electric signal;

a CELP type speech coder that encodes the electric signal output from said speech input apparatus;

a transmitting apparatus that transmits the coded signal output from said CELP type speech coder to a communication partner;

a receiving apparatus that receives a signal transmitted from the communication partner;

a CELP type speech decoder that decodes the received signal output from the receiving apparatus; and

a speech output apparatus that converts the decoded signal output from said CELP type speech decoder to a speech signal and outputs said speech signal,

wherein said CELP type speech coder comprises;

an analyzer that divides the speech signal into frames of a predetermined length and performs linear prediction analysis on a per frame basis;

a linear prediction parameter coder that encodes a linear prediction parameter obtained in said analyzer;

a periodicity coder that operates on a per subframe basis, where a frame is divided into at plurality of subframes, to encode a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector; and

an excitation component coder that encodes an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector, and

wherein said periodicity coder, when performing differential encoding such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, (i) selects a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculates a threshold from a maximum value of the autocorrelation functions of the selected pitch period candidates, and (iii) preliminarily selects, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

18. A CELP type speech coding method comprising the steps of:

dividing a speech signal into frames of a predetermined length and performing linear prediction analysis on a per frame basis;

encoding a linear prediction parameter obtained in the linear prediction analysis;

encoding a periodicity of excitation using an adaptive codebook that stores a previously generated excitation vector on a per sub frame basis, where a frame is divided into a plurality of sub frames; and

encoding an excitation component that cannot be represented by means of the adaptive codebook, using a random codebook storing a predetermined excitation vector,

wherein the periodicity encoding step, when differential coding is performed such that pitch periods are encoded differentially between the subframes and the pitch period in at least one of said subframes is represented by its differential relative to the pitch period encoded in an earlier subframe, further comprises the steps of;

(i) selecting a plurality of pitch period candidates in a subframe where the pitch period is not encoded differentially, based on a scale of an autocorrelation function of an input speech signal or an excitation signal, (ii) calculating a threshold from a maximum value of the autocorrelation function of the selected pitch period candidates, and (iii) preliminarily selecting, from the selected pitch period candidates, at least one pitch period with the autocorrelation function above the threshold.

19. The CELP type speech coding method according to claim 18, further comprising the steps of:

obtaining a normalized autocorrelation function using a previous weighted input speech signal and a new weighted input speech signal;

sorting auto correlation functions into a plurality of ranges according to a pitch in the adaptive codebook;

searching, for each range, a maximum value of the autocorrelation function and the pitch corresponding to said autocorrelation function;

calculating a predetermined threshold from the maximum value of the autocorrelation function; and

selecting, from the pitches searched in the searching step, the pitch corresponding to the autocorrelation function that exceeds said threshold.