摘要
通过对四种智能语音识别系统进行语音识别实验,验证语音识别率受噪声因素影响的程度,并基于智能语音识别的系统结构,从端点检测、语音增强、特征提取和模型空间四个方面,进行了智能语音识别噪声鲁棒性优化方法研究。
In this paper, four kinds of intelligent speech recognition systems are tested to verify the influence of noise on speech recognition rate. Based on the system structure of intelligent speech recognition, the optimization methods of noise robustness of intelligent speech recognition are studied from four aspects: endpoint detection, speech enhancement, feature extraction and model space.
出处
《信息技术与标准化》
2019年第6期40-42,46,共4页
Information Technology & Standardization
关键词
语音识别
噪声
鲁棒性
句识别正确率
speech recognition
noise
robustness
recognition correctness rate
