论文部分内容阅读
介绍了一个舰船指挥舱室强噪声环境下的汉语语音识别系统。根据汉语语音特点，用时频参数，实现了强背景噪声下的音节分割和端点检测，用RASTA-PLP和SDA技术提取鲁棒的语音特征。在85～90dB的背景噪声下，对200条口令，第一候选正识率大于90％，并在TMS320C30支持下，响应时间小于ls，做到了实时识别。
This paper introduces a Chinese speech recognition system under the strong noise environment of a ship’s command cabin. According to the characteristics of Chinese speech, syllable segmentation and endpoint detection with strong background noise are realized with time-frequency parameters. Robust speech features are extracted using RASTA-PLP and SDA techniques. Under the background noise of 85 ~ 90dB, the recognition rate of the first candidate is greater than 90% for 200 passwords and the response time is less than 1s under the support of TMS320C30, achieving real-time identification.