Lossless Encoding of Signals from Microphone Array

Lossless audio coding is increasingly important for properly archiving and preserving audio in its original form. To achieve a good compression ratio, lossless encoding techniques such as linear prediction and Rice coding are often applied to the original audio in order to minimize its entropy and preserve the original signal bit-precisely with a reduced size. This thesis explores the possibilities of efficiently encoding multi-channel audio in a way that exploits the similarity between multiple channels in order to achieve better compression ratios. This thesis also explores the techniques employed by FLAC in more depth while also providing solutions to a few problems that FLAC fails to address.

lossless speech coding, lossless audio coding, linear prediction, adaptive Rice coding, decorrelation, audio processing, dynamic audio encoding