| Summary: | The article studies the process of creating a perceptually-motivated dictionary of the timefrequency functions based on the wavelet packet transform optimized for the input signal frame and its utilization in the universal scalable real-time audiocoder. The article points out the importance of the topic, great attention is paid to the psychoacoustic modelling. It describes the following algorithms: sparse approximation, perceptual adaptation of the wavelet packet decomposition tree, input signal encoding/decoding schemes. The results of the experimental research of the developed coding algorithm and comparison with the modern coding schemes such as Opus and Vorbis based on the objective quality assessment PEAQ – ODG were also given.
|
|---|