A Gammatone-based Psychoacoustical Modeling Approach for Speech and Audio Coding

We propose a new approach for modeling auditory masking based on gammatone filters for application areas including speech/audio coding and audio watermarking. Besides the use of gammatone filters, this model differs from existing audio coding psychoacoustical models (e.g., the ones used in MPEG), in taking into account the contribution of a range of filters in computing the distortion, rather than considering the filter receiving most of the distortion. This is more in line with known psychoacoustical data and it more adequately describes the different masking behavior of tonal versus noisy components without the need for a separate tonality detector. keywords— audio/speech coding, auditory masking, distortion detectability, psychoacoustical model, gammatone filter bank.