Coherent pitch and intensity modification of speech signals

1. A method comprising:

operating one or more hardware processors for receiving an utterance embodied as digitized speech signal, an original pitch contour of the utterance, and a target pitch contour for the utterance, wherein the utterance comprises a plurality of consecutive frames, and wherein at least one of said frames is a voiced frame;

operating at least one of said one or more hardware processors for calculating an original intensity contour of said utterance;

operating at least one of said one or more hardware processors for generating a pitch-modified utterance based on the target pitch contour;

operating at least one of said one or more hardware processors for calculating an intensity modification factor for each of said frames, based on said original pitch contour and on said target pitch contour, to produce a sequence of intensity modification factors corresponding to said plurality of consecutive frames,

wherein:

the calculating of the intensity modification factor for each of said frames is further based on a pitch-to-intensity transformation modeling the relationship between the instantaneous pitch frequency and the instantaneous intensity of the utterance,

the pitch-to-intensity transformation is represented as a function of a pitch frequency and a set of control parameters, and

the pitch-to-intensity transformation is based on log-linear regression, and wherein the set of control parameters comprises the slope coefficient of the regression line of the log-linear regression;

operating at least one of said one or more hardware processors for calculating a final intensity contour for said utterance by applying said intensity modification factors to said original intensity contour; and

operating at least one of said one or more hardware processors for generating a coherently-modified speech signal by time-dependent scaling of the intensity of said pitch-modified utterance according to said final intensity contour.

2. The method of claim 1 , wherein the received utterance is natural speech, and wherein the method further comprises mapping each of said frames to a corresponding speech class selected from a predefined set of speech classes.

3. The method of claim 1 , wherein each of said frames is mapped to a corresponding speech class selected from a predefined set of speech classes, and wherein the method further comprising setting the values of said control parameters for each of said frames according to its corresponding speech class.

4. The method of claim 3 , further comprising operating at least one of said one or more hardware processors for offline modeling of the pitch to intensity relationship to receive said values for said control parameters according to said speech classes.

5. The method of claim 1 , further comprising operating at least one of said one or more hardware processors for setting said control parameters to constant predefined values.

6. The method of claim 1 , wherein the intensity modification factor is ten in the power of the twentieth of the ratio of average empirical decibels per octave multiplied by the extent of pitch modification expressed in octaves.

7. The method of claim 6 , wherein the value of the ratio of empirical decibels per octave is set to six decibels per octave.

8. The method of claim 1 , wherein the calculating of the intensity modification factor for each of said frames comprises:

calculating a reference value of the intensity corresponding to an original pitch frequency of the original pitch contour for said each frame, by applying the pitch-to-intensity transformation to the original pitch frequency;

calculating a reference value of the intensity corresponding to the target pitch frequency of the target pitch contour for each of said frames by applying the pitch-to-intensity transformation to the target pitch frequency; and

dividing the reference value of the intensity corresponding to the target pitch frequency by the reference value of the intensity corresponding to the original pitch frequency.

9. A computer program product comprising a non-transitory computer-readable storage medium having program code embodied therewith, the program code executable by at least one hardware processor to:

receive an utterance embodied as digitized speech signal, an original pitch contour of the utterance, and a target pitch contour for the utterance, wherein the utterance comprises a plurality of consecutive frames, and wherein at least one of said frames is a voiced frame;

calculate an original intensity contour of said utterance;

generate a pitch-modified utterance based on the target pitch contour;

calculate an intensity modification factor for each of said frames, based on said original pitch contour and on said target pitch contour, to produce a sequence of intensity modification factors corresponding to said plurality of said consecutive frames,

wherein:

the calculating of the intensity modification factor for each of said frames is further based on a pitch-to-intensity transformation modeling the relationship between the instantaneous pitch frequency and the instantaneous intensity of the utterance,

the pitch-to-intensity transformation is represented as a function of a pitch frequency and a set of control parameters, and

the pitch-to-intensity transformation is based on log-linear regression, and wherein the set of control parameters comprises the slope coefficient of the regression line of the log-linear regression;

calculate a final intensity contour for said utterance by applying said intensity modification factors to said original intensity contour; and

generate a coherently-modified speech signal by time-dependent scaling of the intensity of said pitch-modified utterance according to said final intensity contour.

10. The computer program product of claim 9 , wherein the received utterance is natural speech, and wherein the program code is further executable by said at least one hardware processor to map each of said frames to a corresponding speech class selected from a predefined set of speech classes.

11. The computer program product of claim 9 , wherein each of said frames is mapped to a corresponding speech class selected from a predefined set of speech classes, and wherein the program code is further executable by said at least one hardware processor to set the values of said control parameters for each of said frames according to its corresponding speech class.

12. The computer program product of claim 11 , wherein the program code is further executable by said at least one hardware processor to offline model the pitch to intensity relationship to receive said values for said control parameters according to said speech classes.

13. The computer program product of claim 9 , wherein the intensity modification factor is ten in the power of the twentieth of the ratio of average empirical decibels per octave multiplied by the extent of pitch modification expressed in octaves.

14. The computer program product of claim 9 , wherein the calculating of the intensity modification factor for each of said frames comprises:

calculating a reference value of the intensity corresponding to an original pitch frequency of the original pitch contour for said each frame, by applying the pitch-to-intensity transformation to the original pitch frequency;

calculating a reference value of the intensity corresponding to the target pitch frequency of the target pitch contour for each of said frames by applying the pitch-to-intensity transformation to the target pitch frequency; and

dividing the reference value of the intensity corresponding to the target pitch frequency by the reference value of the intensity corresponding to the original pitch frequency.

15. A system comprising:

(i) a non-transitory storage device having stored thereon instructions for:

receiving an utterance embodied as digitized speech signal, an original pitch contour of the utterance, and a target pitch contour for the utterance, wherein the utterance comprises a plurality of consecutive frames, and wherein at least one of said frames is a voiced frame,

calculating the original intensity contour of said utterance,

generating a pitch-modified utterance based on the target pitch contour,

calculating an intensity modification factor for each of said frames, based on said original pitch contour and on said target pitch contour, to produce a sequence of intensity modification factors corresponding to said plurality of said consecutive frames,

wherein:

the calculating of the intensity modification factor for each of said frames is further based on a pitch-to-intensity transformation modeling the relationship between the instantaneous pitch frequency and the instantaneous intensity of the utterance,

the pitch-to-intensity transformation is represented as a function of a pitch frequency and a set of control parameters, and

the pitch-to-intensity transformation is based on log-linear regression, and wherein the set of control parameters comprises the slope coefficient of the regression line of the log-linear regression,

calculating a final intensity contour for said utterance by applying said intensity modification factors to said original intensity contour, and

generating a coherently-modified speech signal by time-dependent scaling of the intensity of said pitch-modified utterance according to said final intensity contour; and

(ii) at least one hardware processor configured to execute said instructions.

16. The system of claim 15 , further comprising a database, wherein:

each of said frames is mapped to a corresponding speech class selected from a predefined set of speech classes,

the database comprises values of said control parameters per a speech class of said set of speech classes,

and wherein said storage device has further stored thereon instructions for setting values for said control parameters for each of said frames according to its corresponding speech class, wherein the values are fetched from said database.