Perceptual Audio Coding Using Wavelet Packet Analysis

Cem Dogan, Leibniz Universität Hannover, Bachelorarbeit
11/2004

A perceptual audio coder analyzing each audio segment adaptively using a best wavelet basis according to the time-varying characteristics of the audio signal has been constructed. The adaptive wavelet packet tree is obtained minimizing a perceptual cost function implementing a masking model mimicking the human auditory system. The resulting scheme is a variable bit-rate audio coder providing compression ratios comparable to MPEG-I layers 1 and 2 with almost transparent quality. A short overview of the structure of the audio coder will be given.

Kontakt: Niklas Peinecke