Der temporale Charakter von Klangfarbe

Diploma Thesis (pdf 2.680 KB)

The thesis is about timbre analysis and is aimed at the identification of timbral structures in music progression, which are characteristic for a particular class of sounds, namely, short polyphonic signal bursts. It points out the importance of a particular sequence of timbral features as being a crucial information carrier for the purpose of sound classification and identification.

First, an overview of existing signal analysis methods is presented. Those are usually conditioned by the degree of the signal’s spectral complexity, spanning a range from simple, monophonic harmonic sounds, over polyphonic timbre mixtures, to noise-like textures. Each of these categories demands a particular analysis method in order to sufficiently describe the timbre’s perceptually relevant features.

This work further investigates the possibilities of applying formal techniques like partial envelope modelling from the field of monophonic timbre analysis for re-organizing the content gained by methods of polyphonic timbre analysis like Mel Frequency Cepstral Coefficient (MFCC) representations.