SSM - A novel method to recognize the fundamental frequency in voice signals

Abstract

Nowadays the detection of the fundamental frequency (F0) in voice signals can be evaluated by several algorithms. There are two main attributes of these algorithms: exactness and calculation time. A considerable part of the algorithms are based on the well-known Fast Fourier Transformation (FFT). The Smoothed Spectrum Method is an FFT based process, which was developed for the F0 detection of recorded voice signals especially the infant cry. As it will be shown the SSM provides a better accuracy than regular FFT based algorithms or the Autocorrelation Function. In case of sound recordings in noisy environment the modified SSM is able to recognize significant noise components in the recorded signal. A further advantage of SSM is that additional information of the analyzed signal can be given to improve the performance of the method. © Springer-Verlag Berlin Heidelberg 2007.