Degree Name

Doctor of Philosophy


School of Electrical, Computer and Telecommunications Engineering


This thesis deals with scalable perceptual audio compression. Two scalable perceptual solutions as well as a scalable to lossless solution are proposed and investigated. One of the scalable perceptual solutions is built around sinusoidal modelling of the audio signal whilst the other is built on a transform coding paradigm. The scalable coders are shown to scale both in a waveform matching manner as well as a psychoacoustic manner. In order to measure the psychoacoustic scalability of the systems investigated in this thesis, the similarity between the original signal's psychoacoustic parameters and that of the synthesized signal are compared. The psychoacoustic parameters used are loudness, sharpness, tonahty and roughness. This analysis technique is a novel method used in this thesis and it allows an insight into the perceptual distortion that has been introduced by any coder analyzed in this manner.