摘要
语音数据库评价算法和识别函数是影响语音识别率和识别时间的关键因素。本文利用NOSE算法和梅尔频率倒谱系数加窗函数设计了在线英语语音交互识别系统。该系统能够准确、快速的识别英语语音信息,并利用参考数据库进行评价以及专家知识数据库进行语音纠错,极大的提高了识别效率。与同类主流英语语音识别系统相比,该系统在识别率和识别时间上都存在一定的优势。本文设计为英语语音识别技术的开发提供了参考。
Speech database evaluation algorithm and recognition function are the key factors affecting speech recognition rate and recognition time.In this paper,an online English speech recognition system is designed by using NOSE algorithm and Mel frequency cepstrum coefficient plus window function.The system can recognize English speech information accurately and quickly,and use reference database for evaluation and expert knowledge database for speech error correction,which greatly improves the recognition efficiency.Compared with the same mainstream English speech recognition system,the system has certain advantages in recognition rate and recognition time.This paper provides a reference for the development of English speech recognition technology.
作者
田烨
Tian Ye(School of foreign languages,Xianyang Normal University,Xianyang Shaanxi 712000)
出处
《现代科学仪器》
2020年第3期34-38,共5页
Modern Scientific Instruments
基金
2018年度校级“青年骨干教师”科学研究基金项目,编号:XSYGG201705。
