KARHUNEN-LOEVE BASED ADDITIVE SYNTHESIS OF MUSICAL TONES (AUDIO, SOUND, TRANSFORM, SIGNAL)
Abstract
An additive synthesis technique based on the Karhunen-Loeve (KL) transform is introduced. Current additive methods for musical tones are based on a truncated Fourier series representation. While Fourier-based additive synthesis yields tones of high quality, its computational load has limited its real time applications. A reduction in computation is achieved by replacing the sinusoids with a smaller set of basis functions derived from KL statistical techniques. Algorithms are developed for the analysis of tones, yielding the basis functions and synthesis parameters. Energy compression in the KL basis functions is enhanced with a classification procedure for the tones to be synthesized. The high level of natural quality is preserved in tones synthesized with this method.
Degree
Ph.D.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.