VAD INNES: A voice activity detector for noisy industrial environments
Julie Séris, Christian S. Gargour, Frédéric Laville · Conference proceedings · 2007
A new algorithm for voice activity detection (VAD) in noisy industrial environments is presented in this paper. It is based on a Mel scale wavelet packet partition and on a decision-making process performed according to the Johnstone and Silverman (JSTP) threshold parameter value used as a decision criterion and to energy considerations. The experimental results obtained are very satisfactory.