模拟航天环境下一种应激情绪的语音识别研究  

Speech Recognition of a Stressed Emotion in Simulated Environment of Spaceflight

作　　者：高慧[1] 陈善广[1] 安平[1] 王峻[1] 秦海波[1] 王博[1] 

机构地区：[1]中国航天员科研训练中心声学实验室,北京100094

出　　处：《航天医学与医学工程》2010年第4期248-252,共5页Space Medicine & Medical Engineering

基　　金：中国航天医学工程预先研究项目(SJ200603)

摘　　要：目的实现模拟航天环境下特定人、小词汇量的平静-烦躁语音情绪识别。方法在72h心理隔绝及睡眠剥夺实验、头低位60d模拟失重实验,及环控生保密闭舱60d实验中,录制志愿者的语音。采用基于Teager能量算子的非线性特征,运用HMM技术,实现烦躁情绪识别。结果 3项实验的平静-烦躁情绪平均识别率为98.6%。结论运用非线性特征可有效识别3种实验环境下的烦躁应激情绪,研究结果可为航天心理监测提供参考。Objective To research speech recognition of calm and annoying emotions with speech recognition system of small vocabulary speaker-dependent in simulated environment of spaceflight. Methods The mandarin speech database were established from volunteers speech in 72 h psychological isolation sleep deprivation experiment, 60 d head-down-tilt bed-rest experiment and 60 d sealed module of Environment control and life support experiment. Using nonlinear features based on Teager energy operator and adopting HMM recognition method, calm and annoying emotional status were recognized. Results The recognition rates of calm and annoyance were respectively 98.6% in above experiments. Conclusion The results of recognition demonstrate that the nonlinear features based on Teager energy operator can effectively recognize calm and annoyance in above 3 experiments. It may provide important references for space psychological monitoring in the future.

关 键 词：模拟航天环境 语音识别 应激 情绪 

分 类 号：R857.3[医药卫生—航空、航天与航海医学]