A decoder method for generating comfort noise across multiple audio channels using weighted combination of intra-frame and inter-frame predictions of spatial coherence values.

Patentradars sammanfattning

This invention addresses comfort noise generation in multi-channel audio decoding by combining intra-frame and inter-frame predictions of spatial coherence vectors weighted by a prediction weighting factor. The decoder reconstructs each vector element representing spatial coherence per frequency band by blending two prediction types and applying decoded residuals, enabling improved comfort noise synthesis during silence periods in audio streaming.

AI-genererad och redaktionellt processad.