Glottal Doppler radar system and its applications to communication and speaker recognition

摘要

A glottal radar system has been successfully developed to extract human speech information with great immunity to acoustic interference, which enables numerous speech related applications. This work proposed and demonstrated two potential uses. First, for the communication quality enhancement of mobile handsets in a noisy environment, the background acoustic noise can be effectively detected and eliminated based on the glottal radar detection of the unvoiced segments. Second, for the use in speaker recognition, the identification rate of the microphone detection system drops to 82% even though the system is in a 15-dB SNR condition. In contrast, the glottal signal radar still remains 95% identification accuracy.