Forgent Networks, Inc.

Paul H Whitford

James H Parry

See-Mong Tan

Disclosed is a method and system for detecting processor congestion during decompression of a stream of video and audio data. The system and method includes a processor receiving and decoding a first frame of audio data in accordance with an audio decode software algorithm. The processor generates a first audio time stamp ATS indicating the time at which the processor finishes decoding the first frame of audio data. Subsequently, the processor receives and decodes a second frame of audio data in accordance with the same audio decode software algorithm and generates a second audio time stamp ATS indicating the time at which the processor finishes decoding the second audio data frame. The first audio time stamp ATS is added to a predetermined amount of time T, the result of which is compared with ATS . T, in one embodiment, is the time it takes a speaker to generate audio corresponding to a decoded frame of audio data. If ATS is later in time than (ATS &plus;T) by a predetermined amount TMIN, a signal is generated indicating that the processor is not decoding received audio frames fast enough due to processor congestion. In response to this signal, the processor workload can be redistributed in favor of increased audio decoding. If ATS is not later in time than (ATS &plus;T) by a predetermined amount TMIN, then no modifications to the processor workload need be made.

Method and apparatus for detecting processor congestion during audio and video decode

Disclosed is a method and system for reducing audio artifacts and/or avoiding invalid reference memory in a compressed video decoder due to processor congestion. The system and method includes decoding compressed frames of video data, and decoding compressed frames of audio data. The system and method determines whether audio data can be generated without audio artifacts. If it is determined that audio can be generated without audio artifacts, then images corresponding to the decoded frames of video data, respectively, are subsequently displayed. If however it is determined that the audio can not be generated without artifacts due to processor congestion, images are displayed and redisplayed while processor power is shifted to decoding frames of audio data.

Growing community of inventors

Santa Clara, CA, United States of America

Paul H Whitford