Robust peak detection algorithms for fundamental frequency estimation of sustained vowel phonation in pathological voices
Abstract
This paper presents design of two new methods of speech fundamental frequency (f0) detection for vowel sustained phonations and the detection method, which use cross-corelation to detect f0, is tested. The algorithm consists of certain preprocessing and processing methods.
The first method is based on the detection of maxima and the second method is based on band pass filtration. In comparison with the other commonly used f0 detection methods, our algorithms are designed with respect to speech pathology detection.
These methods lead to detection of the other voice parameters such as jitter, shimmer and harmonic-to-noise ratio (HNR). The results of this study are compared with database, which is labeled by the help of Praat algorithm.
The results for maximum method succeed at 88.4% and for pass band method at 83.9%. The detection leads to create self-automated method, which detect robustly f0.