Near-transparent audio coding at low bit-rate based on minimum noise loudness criterion

A low bit-rate audio coding method based on minimum noise loudness criterion of a psychoacoustic model is proposed. The lower bound of the bit rate versus the suggested perceptual distortion is also discussed. The problem encountered in processing wideband audio signal is its high bit-rate in limited bandwidth channel and high capacity requirement in storage media. Most of high quality audio compression methods heavily rely on human perception, or more precisely on simultaneous masking, to reduce the bit rate. In order to make an explicit and separate control to encode different frequency regions in the auditory spectrum, many compression methods for wideband audio are based on spectral decomposition via linear transform or subband coding. A transform coder based on the human auditory model is designed.