Audio dejittering using delay standard deviation

1. A method for audio dejittering at a device, comprising:

determining a mean delay applied to packets in a packet voice communication system;

calculating, from the mean delay, a standard deviation for each packet of at least a subset of packets in a talk spurt;

determining a target delay for the talk spurt based at least in part on a moving average corresponding to the standard deviation for each packet of at least the subset of packets in the talk spurt;

applying the target delay to a first packet from the talk spurt; and

processing at least the subset of packets in the talk spurt after determining the target delay applied to the first packet is lapsed.

2. The method of claim 1 , further comprising:

calculating, from the mean delay, a standard deviation for each packet in the talk spurt, wherein determining the target delay for the talk spurt by applying the moving average to the standard deviation is based at least in part on calculating the standard deviation for each packet in the talk spurt.

3. The method of claim 1 , further comprising:

determining that the determined target delay is less than a predetermined minimum delay; and

setting the target delay to the predetermined minimum delay.

4. The method of claim 1 , further comprising:

determining that the determined target delay is greater than a predetermined maximum delay; and

setting the target delay to the predetermined maximum delay.

5. The method of claim 1 , wherein the target delay is based at least in part on an underflow rate of a dejitter buffer of the digitized voice system.

6. The method of claim 1 , wherein the target delay is based at least in part on a number of packets in the talk spurt.

7. The method of claim 1 , wherein determining the mean delay comprises:

measuring a mean delay for each packet in a plurality of packets from the packet system and determining whether the measured mean delay of each packet converges.

8. The method of claim 1 , further comprising:

selecting a factor f and a scalar w; and

determining the target delay based at least in part on the factor f and the scalar w, wherein f, or w, or both, are tuned to achieve a balance between mean opinion score (MOS) of the packet voice communication system and the target delay.

9. The method of claim 8 , wherein selecting the factor f and the scalar w comprises:

using supervised learning, or non-linear regression, or both to select the factor f and the scalar w.

10. An apparatus for audio dejittering, comprising:

a processor, memory in electronic communication with the processor; and

instructions stored in the memory and executable by the processor to cause the apparatus to:

determine a mean delay applied to packets in a packet voice communication system;

calculate, from the mean delay, a standard deviation for each packet of at least a subset of packets in a talk spurt;

determine a target delay for the talk spurt based at least in part on a moving average corresponding to the standard deviation for each packet of at least the subset of packets in the talk spurt;

apply the target delay to a first packet from the talk spurt; and

process at least the subset of packets in the talk spurt after determining the target delay applied to the first packet is lapsed.

11. The apparatus of claim 10 , wherein the instructions are further executable by the processor to cause the apparatus to:

calculate, from the mean delay, a standard deviation for each packet in the talk spurt, wherein determining the target delay for the talk spurt by applying the moving average to the standard deviation is based at least in part on calculating the standard deviation for each packet in the talk spurt.

12. The apparatus of claim 10 , wherein the instructions are further executable by the processor to cause the apparatus to:

determine that the determined target delay is less than a predetermined minimum delay; and

set the target delay to the predetermined minimum delay.

13. The apparatus of claim 10 , wherein the instructions are further executable by the processor to cause the apparatus to:

determine that the determined target delay is greater than a predetermined maximum delay; and

set the target delay to the predetermined maximum delay.

14. The apparatus of claim 10 , wherein the target delay is based at least in part on an underflow rate of a dejitter buffer of the digitized voice system.

15. The apparatus of claim 10 , wherein the target delay is based at least in part on a number of packets in the talk spurt.

16. The apparatus of claim 10 , wherein the instructions to determine the mean delay are executable by the processor to cause the apparatus to:

measure a mean delay for each packet in a plurality of packets from the packet system and determine whether the measured mean delay of each packet converges.

17. An apparatus for audio dejittering, comprising:

means for determining a mean delay applied to packets in a packet voice communication system;

means for calculating, from the mean delay, a standard deviation for each packet of at least a subset of packets in a talk spurt;

means for determining a target delay for the talk spurt based at least in part on a moving average corresponding to the standard deviation for each packet of at least the subset of packets in the talk spurt;

means for applying the target delay to a first packet from the talk spurt; and

means for processing at least the subset of packets in the talk spurt after determining the target delay applied to the first packet is lapsed.