Adaptive pre- and post-filtering for a subband ADPCM-based low delay audio codec

This paper addresses the combination of a low delay subband ADPCM-based audio codec with adaptive pre- and post-filtering for psychoacoustic noise shaping. We present how our basic scheme for error robust subband coding can be combined with two cascades of band shelving filters. The gain parameters of these filters are adapted by an algorithm that is based on power estimates which are obtained from the bandpass filtered subband signals of the codec. Since the subband coding as well as the adaptive pre- and post-filtering include several partially interacting parameters that are hard to adjust manually, a framework for their psychoacoustic controlled global optimization is presented. Experiments and their results show that the combination of our low delay subband ADPCM-based audio codec with adaptive pre-and post-filtering allows for a significant improvement of audio quality without any additional signaling overhead. Furthermore, the global optimization enables finding a psychoacoustical meaningful operating point.