An asymptotically optimal predictor for stereo lossless audio compression

Summary form only given. A new generalized stereo decorrelation algorithm and optimal predictor for lossless audio compression is proposed. The proposed algorithm yields around 1.5% better compression than the previous state-of-the-art algorithms. With this approach, the stereo decorrelation and the prediction can be merged into a single step, which can lead to a global minimum for the combined problem. The integer valued audio signal to be encoded is considered. The two discrete time equations, which model the combined problem using an order S intra-channel prediction and an order D inter-channel prediction, are given.