Method and apparatus for speech reconstruction within a distributed speech recognition system
View Patent ↗A method and apparatus for speech reconstruction within a distributed speech recognition system is provided herein. Missing MFCCs are reconstructed and utilized to generate speech. Particularly, partial recovery of the missing MFCCs is achieved by exploiting the dependence of the missing MFCCs on the transmitted pitch period P as well as on the transmitted MFCCs. Harmonic magnitudes are then obtained from the transmitted and reconstructed MFCCs, and the speech is reconstructed utilizing these harmonic magnitudes.
1. A method for speech reconstruction, the method comprising the steps of:
receiving a first plurality of Mel-frequency cepstral coefficients (MFCCs);
calculating a second plurality of MFCCs; and
utilizing the received and the calculated MFCCs for reconstructing speech.
2. The method of claim 1 wherein the step of utilizing the received and the calculated MFCCs for reconstructing speech comprises the steps of:
transforming the received and the calculated MFCCs into harmonic magnitudes; and
utilizing the harmonic magnitudes for reconstructing the speech.
3. The method of claim 1 wherein the step of receiving the first plurality of MFCCs comprises the step of receiving coefficients C 0 –C 12 .
4. The method of claim 3 wherein the step of calculating the second plurality of MFCCs comprises the step of calculating coefficients C 13 through C 22 .
5. The method of claim 4 wherein the step of utilizing the received and the calculated MFCCs for reconstructing speech comprises the steps of:
transforming coefficients C 0 through C 22 into harmonic magnitudes; and
utilizing the harmonic magnitudes for reconstructing the speech.
6. The method of claim 1 wherein the step of receiving the first plurality of MFCCs comprises the step of receiving the first plurality of MFCCs via an over-the-air communication link.
7. The method of claim 1 further comprising the step of:
receiving a pitch period along with the first plurality of MFCCs.
8. The method of claim 1 wherein the step of calculating the second plurality of MFCCs comprises the step of calculating the second plurality of MFCCs, wherein the second plurality of MFCCs are based on the pitch period.
9. The method of claim 1 wherein the step of calculating the second plurality of MFCCs comprises the step of utilizing a single, pre-stored vector for the second plurality of MFCCs.
10. The method of claim 1 wherein the step of calculating the second plurality of MFCCs comprises the step of deriving the second plurality of MFCCs from the first plurality of MFCCs.
11. A method for speech reconstruction, the method comprising the steps of:
receiving Mel-frequency cepstral coefficients C 0 –C 12 ;
calculating Mel-frequency cepstral coefficients C 13 –C 22 ; and
utilizing coefficients C 0 –C 22 for reconstructing speech.
12. The method of claim 11 wherein the step of utilizing coefficients C 0 –C 22 for reconstructing speech comprises the steps of:
transforming the coefficients C 0 –C 22 into harmonic magnitudes; and
utilizing the harmonic magnitudes for reconstructing the speech.
13. The method of claim 11 wherein the step of receiving coefficients C 0 –C 12 comprises the step of receiving coefficients C 0 –C 12 via an over-the-air communication link.
14. The method of claim 11 further comprising the step of:
receiving a pitch period along with coefficients C 0 –C 12 .
15. The method of claim 14 wherein the step of calculating coefficients C 13 –C 22 comprises the step of calculating coefficients C 13 –C 22 based on the pitch period.
16. The method of claim 11 wherein the step of calculating coefficients C 13 –C 22 comprises the step of utilizing a single, pre-stored vector for coefficients C 13 –C 22 .
17. The method of claim 11 wherein the step of calculating coefficients C 13 –C 22 comprises the step of deriving coefficients C 13 –C 22 from the coefficients C 0 –C 12 .
18. An apparatus comprising:
a receiver receiving a first plurality of Mel-frequency cepstral coefficients (MFCCs);
a MFCC reconstructor calculating a second plurality of MFCCs; and
a speech processor utilizing the received and the calculated MFCCs for reconstructing speech.
19. The apparatus of claim 18 wherein the receiver is a radio-frequency (RF) receiver.
20. The apparatus of claim 18 wherein:
the first plurality of MFCCs comprise coefficients C 0 –C 12 ; and
the second plurality of MFCCs comprise coefficients C 13 –C 22 .