Digital filtering for audio coding

In this paper we review the role of digital filters for audio coding. Most specifically we discuss the general subband/transform coding framework used in state of the art perception-based audio coders. Such strategies rely on the use of digital filter banks to decompose signals into time-frequency components to achieve better source coding gains by exploiting the signal redundancy and also facilitating the direct use of psychoacoustic masking models on quantization noise shaping to remove signal irrelevancy. The paper focuses on the use of various digital filtering designs and implementation issues which have found their place in audio coding. Examples of their roles and computational requirements are illustrated by MPEG audio coding standards.