Multi Channel Voice Active Detection Using Instance Filed Auto-Interrelation Function
Murugavelu Mathivanan, S. Chenthur Pandian · International Review on Computers and Software (IRECOS) · 2013
This research work presents a method of Voice Activity Detection (VAD) in a multi-channel communication for vigorous noise reduction using the Noise Strong Voice Speech Discovery system. This method is based on the amalgamation of two phases. The initial phase utilizes the utmost apex of the normalized Instance Filed Auto-Interrelation function in multi channel communication. The second phase uses a strange mixing of cross- interrelation and zero-transit rate of the standardize auto-interrelation to the suitable measure of signal pitch and periodicity for strong noise reduction. Existing VAD coding techniques in multi channel communication are not appropriate to specific location and disgrace the speech quality while decoding. The score outcomes of the two phases are combined using the TotalWeigh Fusion to create the proposed Auto-Interrelation Zero Transit Rate (AIZT) Voice Active Detection. Accuracy of the AIZT is compared with the standardized LTSV-based VAD scheme and VAD methods to outperform the best accomplishing method. Performance of the Auto-Interrelation Zero Transit Rate (AIZT) Voice Active Detection is measured in terms of multi channel communication efficiency, noise level. Experiments are conducted in the M2VTS audio-visual database, TIMIT dataset and the Dutch subset of a test database to evaluate the communication in multiple channels.