MATLAB-Tool für die Parameterberechnung zur Additiven Klangsynthese

Cornelia Falch

Diplomarbeit ( 821 KB pdf file )

One of the most powerful tools for resynthesing sound signals represents the additive sound synthesis (ASS). Since the speed and computational power of computers increased dramatically in recent years, the use of this synthesis has become even more important. Thus, also analysis algorithms for estimating the parameters needed in ASS, that are the frequency, amplitude and phase, respectively, of a sound signal, were improved and extended.

The aim of this diploma thesis is to investigate existing analysis software (i.e. Spectral Modeling Synthesis, SMS; MQ Algorithm; Reassignment), and combine them in order to obtain best detection results for the three desired parameters. An additional decomposition of the input into a deterministic and a stochastic signal seems to be reasonable as the ASS only takes account of the harmonic and inharmonic part. Thus, a separate processing of both signal types is employed, which leads to audible improvements when resynthesing the original.

Within this thesis a program called "SOUND ANALYSIS" is especially delveloped and designed to act as an analysis tool for an already existing synthesis algorithm, both written in MATLAB programming language. Several tests performed with different audio signals showed the limits and characteristica of the entire system.