Low complexity no delay reconstruction of missing packets for LPC decoder

1. A method for reconstruction of lost frames, comprising:

a) analyzing a previous good or reconstructed frame to determine a category for the lost frame;

b) associating a percentage P i with the determined category for the lost frame;

c) zeroing out a top P i percent magnitude samples in an excitation of the previous good or reconstructed frame to produce a reconstruction excitation; and

d) applying the reconstruction excitation to one or more linear prediction coefficients for the previous good or reconstructed frame to generate a reconstructed frame.

2. The method of claim 1 wherein the lost frame and previous good or reconstructed frame are audio frames.

3. The method of claim 2 wherein a) includes determining whether the lost frame was a voice frame, an unvoiced frame, a high-to-low energy transition frame or a low-to-high energy transition frame.

4. The method of claim 3 wherein:

P i =P 1 , if the lost frame is a voice frame;

P i =P 2 , if the lost frame is an unvoiced frame,

P i =P 3 , if the lost frame is a high-to-low energy transition frame,

P i =P 4 , if the lost frame is a high-to-low energy transition frame, wherein

P 1 <P 2 <P 3 <P 4 or P 1 <P 3 <P 2 <P 4 .

5. The method of claim 1 , further comprising updating an adaptive codebook with the reconstruction excitation.

6. The method of claim 1 wherein a) includes determining a behavior of a signal energy magnitude as a function of time during the previous good or reconstructed frame.

7. The method of claim 6 wherein a) includes categorizing the previous good or reconstructed frame as a voice frame if the energy magnitude is determined to be relatively large and constant.

8. The method of claim 6 wherein a) includes categorizing the previous good or reconstructed frame as an unvoiced frame if the energy magnitude is determined to be relatively small and constant.

9. The method of claim 6 wherein a) includes categorizing the previous good or reconstructed frame as a high-to-low transition frame if the energy magnitude is determined to decrease with time.

10. The method of claim 6 wherein a) includes categorizing the previous good or reconstructed frame as a low-to-high transition frame if the energy magnitude is determined to increase with time.

11. The method of claim 1 , wherein a) includes assigning a category to the lost frame that is the same as a category of the previous good or reconstructed frame.

12. The method of claim 1 , further comprising adjusting a formula used to generate the percentage P i based on a listener's experience with sound quality of speech synthesized with the reconstructed frame.

13. The method of claim 1 , wherein, if two or more consecutive frames are lost frames, the lost frames are reconstructed by performing a) through d) for an earliest of the two or more consecutive frames to generate a first reconstructed frame and repeating a) through d) for a subsequent on of the two or more consecutive frames using the first reconstructed frame as the previous good or reconstructed frame.

14. The method of claim 1 , further comprising generating a final signal using the reconstructed frame, wherein the final signal is configured for presentation on a transducer.

15. The method of claim 14 , further comprising presenting the final signal with the transducer.

16. A method for reconstruction of lost frames in conjunction with decoding a plurality of frames, comprising:

receiving a plurality of frames including a lost frame;

analyzing a previous good or reconstructed frame to determine a category for the lost frame;

associating a percentage P i with the determined category for the lost frame;

zeroing out a top P i percent magnitude samples in an excitation of the previous good or reconstructed frame to produce a reconstruction excitation; and

applying the reconstruction to one or more linear prediction coefficients for the previous good or reconstructed frame to generate a reconstructed frame.

17. A method for reconstruction of lost frames in conjunction with encoding a plurality of frames, comprising:

generating a plurality of frames including a lost frame;

analyzing a previous good or reconstructed frame to determine a category for the lost frame;

associating a percentage P i with the determined category for the lost frame;

zeroing out a top P i percent magnitude samples in an excitation of the previous good or reconstructed frame to produce a reconstruction excitation; and

applying the reconstruction to one or more linear prediction coefficients for the previous good or reconstructed frame to generate a reconstructed frame.

18. An apparatus for reconstruction of lost frames, comprising:

a processor module having a processor with one or more registers;

a memory operably coupled to the processor; and

a set of processor executable instructions adapted for execution by the processor, the processor executable instructions including:

one or more instructions that when executed on the processor analyze a previous good or reconstructed frame to determine a category for the lost frame;

one or more instructions that when executed on the processor associate a percentage P i with the category determined for the lost frame;

one or more instructions that when executed on the processor zero out a top P i percent magnitude samples in an excitation of the previous good or reconstructed frame to produce a reconstruction excitation; and

one or more instructions that when executed on the processor apply the reconstruction excitation to linear prediction coefficients for the previous good or reconstructed frame to generate a reconstructed frame.

19. A non-transitory computer readable medium encoded with a program for implementing a method for reconstruction of lost frames, the method comprising:

analyzing a previous good or reconstructed frame to determine a category for the lost frame;

associating a percentage P i with the determined category for the lost frame;

zeroing out a top P i percent magnitude samples in an excitation of the previous good or reconstructed frame to produce a reconstruction excitation; and

applying the reconstruction excitation to one or more linear prediction coefficients for the previous good or reconstructed frame to generate a reconstructed frame.