An Application of the Piecewise Autoregressive Model in Lossless Audio Coding

The paper is focused on the prediction stage for lossless audio coding. We rely on the piecewise autoregressive model to design a compressor dubbed PAR for which the complexity of the decoder is low such that to be easily implemented on economic devices. The performances of the new coder are evaluated for two data sets of monaural audio signals sampled at 44.1 KHz and recorded with 16 bits/sample. The results obtained with PAR compares favorably with the popular MAC Audio Coder.