Audio Transcoding Using Covered Compression Scheme on Heterogenous Compressed Domain

Abstract

With the recent advancement in audio transcoding, the multimedia telecommunication services are predictable to use pre-encoded audio for storage and transmission. The major concern related to transcoding B-picture to P-picture from MPEG-2 to H.263 are that the incoming discrete consine transform (DCT) coefficients of B-picture are predicted from the forward and backward predictions whereas the prediction error in the DCT domain occurring from the previous frame are not available. This requires the new prediction errors to be re-estimated resulting in high complexity not only during the heterogeneous compressed domain process but also while re-estimatng the new errors. To provide an elegant solution to the predicted errors and to perform heterogeneous compessed domain, an innovative approach called as the Covered Coding Technique based on Ascendable Sub-band Coding (CC-ASC) for encoding and decoding audio frames for real time transmission is presented. The objective of the research is to support heterogeneous domain with a single bit-stream, from which a number of sub-streams of varying bit-rates are extracted. A muticast transmission is introduced, where the dissimilar receivers are capable of receiving different bit rate streams from the same source in an efficient and scalable way. The multi rate property of CC-ASC also allows us to provide elegant solution to the predicted errors in the frequency domain when used over networks which support multiple priorities.