Authors : T. Ravichandran and K. Durai Samy
Abstract: One of the most difficult problems in speech analysis is reliable discrimination among Silence, Unvoiced and Voiced speech. Although several methods have been proposed for making this three level decision, this study provides a method to detect weak fricatives by Voice Activity Detection (VAD). VAD can be viewed as a decision problem in which detector decides between speech and silence. The signal is sliced into contiguous frames. Energy of a frame indicates the possible presence of voice data as it has high energy. Based on the assumption that zero crossing rates for speech and noise are generally different, low energy speech can be refined. High correlation found in speech samples can be exploited to detect weak fricatives. Various test samples are used by varying accent, loudness of male voice and female voice. A comparison of relative merits and demerits of three time domain VAD algorithms along with the subjective quality of speech after pruning of silence periods is presented .VAD is used in a variety of speech communication systems such as speech coding, speech recognition, hands-free telephony, audio conferencing and echo cancellation.
T. Ravichandran and K. Durai Samy , 2007. Performance Evaluation and Comparison of Voice Activity Detection Algorithms. International Journal of Soft Computing, 2: 257-261.