Combining Finite State Machines and LDA for Voice Activity Detection
Ilias Rentzeperis, Christos Boukis, Aristodemos Pnevmatikakis, Lazaros Polymenakos · 2007
A robust speech activity detection system is presented in this paper. The proposed approach combines the well-known linear discriminant analysis with a finite state machine in order to successfully identify speech patterns within a recorded audio signal. The derived method is compared with existing ones to demonstrate its superiority, especially when performing on noisy audio signals, obtained with far field microphones.