Design and Realization of Dialect Interaction System Based on VAD

Abstract

In view of the miscommunication problem between Mandarin users and dialect users, we design a portable dialect interaction system based on the raspberry pi. We use IFLYTEK for recognition and synthesis, Turing Robot for intention understanding and response. Recognition results and interaction effect can be more accurate by using voice activity detection (VAD) to preprocess speech signal, which removes noise and separates speech signals by judging the starting and ending points of speech. The strong scalability and good hardware adaptability of raspberry pi make the system have good practicability.