Delayless soft-decision decoding of high-quality audio with adaptively shaped priors

For high-quality digital audio transmitted over error-prone short-range wireless channels robust source decoding with low decoding delay is desired. However, many previous approaches from the field of audio error concealment are solely intuitively motivated or introduce algorithmic delay. In contrast, this paper deals with a Bayesian framework for delayless full-band soft-decision error concealment for quantized but uncompressed audio utilizing only residual redundancy in the audio signal and channel reliability information. In principle, it can be applied to any channel providing reliability information. As a novelty, we employ multiple adaptively shaped prior probability distributions in the decoding process. These are utilized in conjunction with the autocorrelation method or the normalized least-mean-square (NLMS) algorithm in order to compute prediction probabilities within the Bayesian framework. Experiments carried out on representative audio data transmitted over additive white Gaussian noise (AWGN) channels show significant enhancements in audio quality.