This shows you the differences between two versions of the page.

Link to this comparison view

fundamental_frequency_estimation_employing_wavelet_parametrization [2014/03/26 17:05] (current)
bziolko created
Line 1: Line 1:
 +Jakub Gałka, Rafał Samborski (Poland)
 +**Fundamental frequency estimation employing wavelet parametrization**
 +AGH University of Science and Technology
 +This paper describes a novel method for estimating the fundamental frequency of speech. The method employs wavelet decomposition of signal, Gaussian filters and integral transform with impulse train kernel. The method does not rely on the F0 frequency component. It obtains all harmonics within an intentionally limited frequency range and detects the differences between strongest of them. Such approach makes the results very robust especially for low quality signals. The method outperformed two state-of-the-art reference methods in simulations for low signal-to-noise ratio. Computational load of the method remains reasonable and allows the real-time implementation in speech processing systems. Data of speakers from a database containing both sound recording and electroglottograph data were used in verification experiments. Preliminary results are given and discussed.
Copyright © XXII PVC Organizing Committee 2013. All Rights Reserved.