A novel endpoint detection method and its application on human-computer interaction of intelligent wheelchair
一种新的语音端点检测方法及在智能轮椅人机交互中的应用

Voice activity detection lower the speech recognition rate of intelligence wheelchair in the general noise environment. Short-term energy and cepstral distance combined endpoint detection method is proposed in this paper. Experiments show that this method can enhance the overall robustness of speech recognition system and realize the voice control of the intelligent wheelchair by using five basic movements commands including forward, backward, turn left, turn right, stop.