| 摘要 |
Temporal drift correction may be provided in a real-time audio communication system by measuring a size of a receiving data buffer and comparing that size to a predetermined nominal data buffer size. An amount of temporal drift is characterized as a number of samples per audio playback data block based on the measured data buffer size and the nominal data buffer size. A number of samples to be inserted or removed for each audio playback data block to correct the temporal drift may be determined, and the number of samples for each audio playback data block may be modified. For example, an instantaneous size of the receiving data buffer may be measured, and if measured multiple times, may be averaged over a time period. Heuristic resampling of the audio playback data block also may be performed. When heuristic resampling is performed, multiple consecutive samples of audio data in an audio buffer may be analyzed, consecutive samples with minimal variation in a parameter of their data may be identified, and the number of samples in the identified consecutive samples may be adjusted. A sample may be removed from or added to the identified consecutive samples.
|
