Estimation of pulsed speech model parameters

• US 8,036,886 B2
• Filed: 12/22/2006
• Issued: 10/11/2011
• Est. Priority Date: 12/22/2006
• Status: Active Grant

• Current Assignee: Digital Voice Systems, Inc.
• Original Assignee: Digital Voice Systems, Inc.
• Inventors: Griffin, Daniel W.
• Primary Examiner(s): Opsasnick; Michael N
• Application Number:
  US11/615,414
• Publication Number:
  US 20080154614A1
• Time in Patent Office: 1,754 Days
• Field of Search: 704/203, 704/214, 704/216
• US Class Current: 704/216
• CPC Class Codes: G10L 19/0204 using subband decomposition
• CPC Class Codes: G10L 19/10 the excitation function being a multipulse excitation

Abstract

Methods for estimating speech model parameters are disclosed. For pulsed parameter estimation, a speech signal is divided into multiple frequency bands or channels using bandpass filters. Channel processing reduces sensitivity to pole magnitudes and frequencies and reduces impulse response time duration to improve pulse location and strength estimation performance. These methods are useful for high quality speech coding and reproduction at various bit rates for applications such as satellite and cellular voice communication.

Claims

1. A method of analyzing a digitized signal to determine model parameters for the digitized signal, the method comprising:

receiving a digitized signal;

dividing the digitized signal into at least two frequency band signals;

performing an operation to emphasize pulse positions on at least two frequency band signals to produce modified frequency band signals;

determining pulsed parameters from the at least two modified frequency band signals.

2. The method of claim 1 wherein pulsed parameters are determined at regular intervals of time.

3. The method of claim 1 wherein the pulsed parameters are used to encode the digitized signal.

4. The method of claim 1 wherein the pulsed parameters include a pulsed strength.

5. The method of claim 4 wherein a voiced strength is used in determining the pulsed strength.

6. The method of claim 1 wherein the pulsed parameters include pulse positions.

7. The method of claim 4 wherein the pulsed strength is determined using one or more pulse positions estimated from the digitized signal.

8. The method of claim 4 wherein the pulsed strength is used to estimate one or more model parameters.

9. The method of claim 1 wherein the operation to emphasize pulse positions includes a nonlinearity.

10. The method of claim 1 wherein the operation to emphasize pulse positions includes an operation to reduce sensitivity to pole magnitudes.

11. The method of claim 1 wherein the operation to emphasize pulse positions includes an operation to reduce sensitivity to pole frequencies.

12. The method of claim 1 wherein the operation to emphasize pulse positions includes an operation to reduce pulse time duration.

13. The method of claim 9 wherein the operation to emphasize pulse positions further includes an operation which quickly follows a rise in the output of the nonlinearity and slowly follows a fall in the output of the nonlinearity to produce fast rise slow decay frequency band signals.

14. The method of claim 13 wherein the fast rise slow decay frequency band signals are further processed to emphasize pulse onsets.

15. The method of claim 14 wherein pulse onsets are emphasized by subtracting a weighted sum of previous samples of the fast rise slow decay frequency band signals from the current value to produce emphasized frequency band signals.

16. The method of claim 15 wherein the emphasized frequency band signals are further processed by a rectifier operation that preserves positive values and clamps negative values to zero.

17. The method of claim 6 wherein the pulse positions are estimated from a combination of the modified frequency band signals.

18. The method of claim 17 wherein the pulse positions are estimated from the combination by correlation with a pulse location signal.

19. The method of claim 18 wherein the pulse location signal is low pass.

20. The method of claim 18 wherein a pulse position is estimated by choosing the location at which the correlation is maximum.

21. The method of claim 1 wherein the modified frequency band signals are remapped into a set of remapped modified frequency band signals.

22. The method of claim 21 wherein the pulsed strength of a remapped modified frequency band signal is determined using one or more pulse positions estimated from the digitized signal.

23. The method of claim 22 wherein the pulsed strength is determined by comparing a weighted sum of the remapped modified frequency band signal around the estimated pulse positions to the total weighted sum over the frame window.