Voice data transmission with adaptive redundancy

摘要

Voice data transmission with adaptive redundancy is described. Voice data is received 502 and encoded as a voice data payload. A voice data packet is created by packetizing the voice data payload and a number of redundant payloads selected from a set of previous voice data payloads. The voice data from the voice data payload is analysed to determine whether it is a critical or non-critical payload by classifying the received voice data as voiced or unvoiced 504. Voice data created by vocal chords vibrating is classified as voiced. If at least a portion of the voice data is classified as unvoiced, the voice data payload is determined to be a critical payload 506. If it is a critical payload, then the voice data payload is added to the set of previous voice data payloads for inclusion as a redundant payload in subsequent voice data packets. The voice data packet is then forwarded for transmission over the network. A critical payload is defined as a payload whose loss, in transmission across the network, cannot be concealed by the receiver. Voiced data payloads may also be classified 512 as critical from a determined variance from a waveform similarity factor 508. The redundancy level may be dynamic depending on a network statistics report such as for out-of-order or lost packet statistics.