Error concealment for speech decoder

摘要

Provided is a system, method, and computer program product for improving the quality of speech reproduction in wireless applications where the received speech frames are subject to transmission and packet losses. The speech decoding process is dynamically delayed by at least one frame period in order to perform additional error correction and concealment techniques during times when the wireless link quality if below a predetermined threshold. The wireless link is monitored and if the link quality falls below a predetermined threshold, the decoding process is delayed by at least one frame period so that one or more error correcting techniques can be performed to increase the quality of the reconstructed speech.

主权项

1. A method for improving the quality of a voice call transmitted over a cellular network through a cellular wireless link to a communication device having a voice decoder for generating reconstructed speech, the method comprising the steps of:
monitoring the cellular wireless link to determine a link quality; receiving a voice packet from the cellular wireless link at the decoder; determining if the voice decoder is a normal speech decoder or a delayed speech decoder; if the voice decoder is said delayed speech decoder and said link quality is above a predetermined threshold, dynamically decreasing the latency of the voice call to generate a low-latency voice call by adjusting the voice decoder to said normal speech decoder during silence or background noise periods, and thereafter decoding the voice packet immediately; and if the voice decoder is said normal speech decoder and said link quality is below said predetermined threshold, dynamically increasing the latency of the voice call to generate a high-latency voice call by adjusting the voice decoder to said delayed speech decoder during silence or background noise periods, and thereafter delaying the decoding process to perform at least one error correcting technique using newer information contained in speech packets received after a missing frame for increasing the quality of the reconstructed speech.