Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Optimized covariance domain framework for time-frequency processing of spatial audio

: Vilkamo, Juha; Bäckström, Tom; Kuntz, Achim

Journal of the Audio Engineering Society 61 (2013), Nr.6, S.403-411
ISSN: 0004-7554
ISSN: 1549-4950
Fraunhofer IIS ()

The covariance matrix of a multichannel audio signal is a measure that contains the channel energies and the inter-channel dependencies. This measure is perceptually relevant in frequency bands, since with the effect of the acoustic transfer path, it forms the inter-aural cues based on which the human spatial hearing system decodes the spatial sound. This is the foundation to several state of the art perceptual spatial processing techniques. In this paper we propose a generalized framework for spatial sound processing that operates directly in the covariance matrix domain. The technique optimizes the sound quality by a least mean square solution with respect to a defined prototype signal. Furthermore, the technique uses all available independent signal components in the input channels in order to minimize the undesired effects due to the decorrelators that are often employed in the concerned use cases. A wide variety of applications of the proposed method has been identified.