摘要
恶劣的军事环境为语音识别系统应用于军事指挥带来了巨大的困难 ,本文基于语音识别的基本原理 ,并根据军事指挥这一特殊的应用环境 ,提出了一种高鲁棒性的语音识别方案。经过战场仿真环境测试表明 ,该方案的语音识别率在高噪声环境中能达到 90 %以上 。
Adverse military environment often leads to many difficulties for reliable application in speech recognition. In this paper, a highly robust speech recognition system is presented based on the basic principle of speech recognition and analysis of military environment. The recognition experiment shows that the proposed ASR system can achieve more than 90% recognition accuracy in adverse environment.
出处
《兵工学报》
EI
CAS
CSCD
北大核心
2004年第4期509-512,共4页
Acta Armamentarii
基金
北京理工大学基础研究基金 (BIT UBF 2 0 0 3 0 1F0 3 )
关键词
军事指挥
鲁棒性
语音识别系统
信息处理技术
感知线性预测
噪声抑制
information processing technique, speech recognition, robustness, perceptually based linear prediction, noise suppression, speaker adaptation
