Potential possibilities of voice image recognition by the analysis of distributed fiber-optic sensor signals

Abstract

This study aimed to investigate the potential application of distributed fiber-optic acoustic sensors for speech recognition amidst complex audio backgrounds. The experimental measurements were carried out on the mock-up room with a fiberoptic link. The possibility of application the fiber-optic distributed acoustic sensor for localization of sound source and recognition of musical and human speech signals amidst complex audio backgrounds was demonstrated. The signal detected by DAS provides good quality of music composition recognition and clear speech recognition. The difference in spectrum of signals detected by microphone and DAS was analyzed. The signal noise ratios were calculated for test speech signal with different average levels and its influence on the speech recognition was analyzed.