LAUSR

Mobile intelligent terminal speaker identification for real-time monitoring system of sports training is presented in this manuscript. The purpose of speaker recognition is to use speaker’s voice characteristics to identify the speaker’s identity. No two people in the world have exactly the same voice characteristics, because from the perspective of the human physiological structure, different people’s vocal shapes, throat sizes, and other factors related to pronunciation are different, which makes it possible to recognize speech by speech. From the technology level, the proposed framework contains the four core aspects, namely: speaker identification, motion gesture detection, heart rate detection and breath detection. In the information collection part, we adopt the ZigBee + WIFI wireless transmission, GPRS data transmission and Bluetooth communication joint framework to capture the accurate data to rrealize the dual-network dual-pass function, the TD-SCDMA receiver must also work at the same time when the GSM radio module of the terminal is working. In the speaker identification part, we propose the novel deep neural network framework and revise the architecture. The experiment compared with the other state-of-the-art models have proven the robustness of the proposed framework.