SPEECH RECEPTION VIA A PACKET TRANSMISSION FACILITY

摘要

Degradations in packetized voice communications received by a non-synchronized entity, via a packet network, are reduced by adjusting a depth of storage in a jitter buffer of the receiving entity. Units of voice samples data are stored in the jitter buffer as they are received. Stored units are normally extracted and delivered to a processor one at a time at a regular rate for the generation of audible speech. From time to time the rate of extraction can be accelerated by extracting two units while delivering only one. Also the rate of extraction can be retarded by not extracting a unit while delivering a substitute unit in place of the unit that would normally have been extracted. The depth of storage is thereby controllable in response to packet reception events such that delay is minimized while yet providing sufficient delay to smooth variances between reception events.