Concatenative Music Composition Based on Recontexturalization Utilizing Rhythm Synchronous Music Releated Feature Extration

Diploma Thesis (pdf. 2.553 KB)

The aim of this diploma thesis is to design a software tool that makes a new kind of transformation of audio signals, be it short sounds or whole songs, possible. This can subsequently be used to synthesize music from audio signal fragments (“Concatenative Sound Synthesis“).

Many applications, e.g. genre classification of music or generation of playlists etc., analyse and compare audio signals using coefficients (“features”) describing, for example, timbre and harmony properties. These features are extracted from the audio signal itself.

This diploma thesis attempts a new approach to resynthesis of audio signals. As in a mosaic, an existing song is newly constructed from small parts (“frames”) of other songs already stored in a database. The length of the frames corresponds to musically meaningful units.

For the implementation suitable onset- and beattracking methods are evaluated and the selection of suitable parameters describing the subjective semantic similarities is determined by listening tests.