Coding of Parametric Models with Randomized Quantization in a Distributed Speech and Audio Codec

Konferenz: Speech Communication - 12. ITG-Fachtagung Sprachkommunikation
05.10.2016 - 07.10.2016 in Paderborn, Deutschland

Tagungsband: ITG-Fb. 267: Speech Communication

Seiten: 5Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Baeckstroem, Tom; Fischer, Johannes (International Audio Laboratories Erlangen, a joint institution of the Friedrich-Alexander University Erlangen-Nürnberg (FAU) and Fraunhofer Institute for Integrated Circuits (IIS), Germany)

For efficient distributed coding of speech and audio signals, we need to encode the signal such that each device transmits unique information, to avoid redundancy. In a recent contribution, we have demonstrated that this is possible by application of a randomization operation before quantization of the signal spectrum. The proposed coding however relies on an assumption that the input data follows the normal distribution with zero-mean, whereby it does not directly apply on the coding of the parameters of parametric models of the signal. Importantly, the proposed coding does not preserve signal gain, which thus has to be parametrized and separately transmitted. To avoid overcoding also in transmission of the signal gain and the coefficients of other parametric models, in this paper, we propose a method for mapping parameters to the same probability distribution as spectral coefficients, as well as methods for perceptual weighting of these parameters.