Automatic Inverse Filtering Method for the Analysis of Glottal Waveforms

摘要

The thesis describes the development of an inverse filtering technique for theanalysis of glottal excitation. The method is based on the separated speech production model. The thesis consists of nine publications. Four of the papers deal with the application of the glottal wave analysis to low bit rate speech coding. One discusses from a different point of view the role of pre-emphasis in linear predictive coding. The papers describe the development of an analysis method. The starting point was a fairly simple algorithm, called Adaptive Inverse Filtering (AIF). The knowledge and experience obtained by studying the AIF-algorithm was of great importance in the development ofa more sophisticated technique, called Iterative Adaptive Inverse Filtering (IAIF). The author believes that the IAIF-method has several features that make it a promising technique in the analysis of the glottal pulseform. However, the performance of the algorithm needs to be improved in order to obtain more reliable results especially in the analysis of vowels of low first formant. A first step in the direction was the introduction of Pitch Synchronous Iterative Adaptive Inverse Filtering (PSIAIF).