Apparatus and quality enhancement algorithm for mixed excitation linear predictive (MELP) and other speech coders

• US 6,453,287 B1
• Filed: 09/29/1999
• Issued: 09/17/2002
• Est. Priority Date: 02/04/1999
• Status: Expired due to Fees

• Current Assignee: Georgia Tech Research Corporation (University System of Georgia)
• Original Assignee: Georgia Tech Research Corporation (University System of Georgia)
• Inventors: Truong, Kwan K., Barnwell, Thomas P. III, Unno, Takahiro
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): NOLAN, DANIEL A
• Application Number:
  US09/408,195
• Time in Patent Office: 1,084 Days
• Field of Search: 704/200, 704/208, 704/209, 704/230, 704/261, 704/223, 704/226, 704/219, 704/264, 704/265
• US Class Current: 704/219
• CPC Class Codes: G10L 19/10 the excitation function being a multipulse excitation
• CPC Class Codes: G10L 25/90 Pitch determination of speech signals

Abstract

A system and method for enhancing the speech quality of the mixed excitation linear predictive (MELP) coder and other low bit-rate speech coders. The system and method employ a plosive analysis/synthesis method, which detects the frame containing a plosive signal, applies a simple model to synthesize the plosive signal, and adds the synthesized plosive to the coded speech. The system and method remains compatible with the existing MELP coder bit stream.

Claims

1. A method of enhancing the speech quality of a speech coder encoded data transmission, comprising:

digitally sampling speech to create a speech waveform over a plurality of frames;

identifying frames that contain a plosive signal distinguished from other transitory signals;

analyzing the plosive signal to create plosive signal parameters;

applying the plosive signal parameters to a linear prediction residual plosive signal to synthesize the plosive signal for frames that contain a plosive signal; and

adding the synthesized plosive signal to the synthesized speech for the frame that contains the plosive.

2. The method of claim 1, wherein the step of identifying frames that contain a plosive signal comprises detecting peakiness in a linear prediction residual signal.

3. The method of claim 1, wherein the step of applying further comprises:

applying the plosive signal parameters to a previously-stored linear prediction residual plosive signal and applying a linear prediction synthesis filter.

4. The method of claim 1, wherein the step of analyzing comprises identifying a subdivision of the frame that contains the plosive and calculating the amplitude of the plosive.

5. The method of claim 3, wherein applying the plosive signal parameters comprises scaling a previously-stored plosive signal by the plosive amplitude.

6. The method of claim 5, wherein the length of a previously-stored linear prediction residual plosive signal is equal to the length of a subdivision of the frame containing the plosive.

7. The method of claim 2, wherein detecting peakiness in the linear prediction residual signal comprises computing the ratio of the L1 and L2 norm of the linear prediction residual signal with a sliding sample window.

8. The method of claim 4, wherein the step of adding the synthesized plosive signal comprises adding the synthesized plosive signal to the identified subdivision of the frame.

9. A speech coder, comprising:

means for digitally sampling speech to create a speech waveform over a plurality of frames;

means for identifying frames that contain a plosive signal distinguished from other transitory signals;

means for analyzing the plosive signal to create plosive signal parameters;

means for applying the plosive signal parameters to a linear prediction residual signal to synthesize the plosive signal for frames that contain the plosive; and

means for adding the plosive signal to the synthesized speech for frames that contain the plosive.

10. The coder of claim 9, wherein the means for identifying frames that contain a plosive signal detects peakiness in a linear prediction residual signal.

11. The coder of claim 9, wherein the means for synthesizing the plosive signal applies the plosive signal parameters to a previously-stored linear prediction residual plosive signal and applies a linear prediction synthesis filter.

12. The coder of claim 9, wherein the means for analyzing the plosive signal to create plosive parameters, identifies a subdivision of the frame that contains the plosive and calculates the amplitude of the plosive.

13. The coder of claim 12, wherein the length of a previously-stored linear prediction residual plosive signal is substantially equivalent to the length of the subdivision.

14. The coder of claim 11, wherein the means for applying the plosive signal parameters further comprises:

scaling a previously-stored signal by the plosive amplitude.

15. The coder of claim 10, wherein the means for identifying further comprises:

detecting peakiness in the linear prediction residual signal.

16. The coder of claim 15, wherein detecting peakiness comprises computing the ratio of the L1 and L2 norm of the linear prediction residual signal with a sliding sample window.

17. The coder of claim 12, wherein the means for adding the plosive signal to the synthesized speech comprises adding the synthesized plosive signal to the subdivision.