舰船指挥舱室强噪声环境下语音识别

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，根据充语语音特点，同时频参数，实现了强背景噪声下的音分分割和端点检测，用ＲＡＳＴＡ－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ＴＳ３２０Ｃ３０支持下，响应时间小于１ｓ，做到了实时识别。

Sound Recognition under Strong Environment in Command Compartment for Naval Ship

Based on the features of Chinese pronounciation and by using time-frequency parameters,the syllablification and end point detection are realized under strong noise environment.The robust sound parameters can be abstracted using RASTAPLP and SDA techniques.In the presence of 85-90 dB background noise,the first preferential recognition ratio of 200 Chinese words is more than 90%,while the recognition response time is less than one second with the help of TMS320C30.Therefore,the real time recognition can be achieved.