摘要
针对用户演讲与口才表达能力提高的需求以及语音识别应用中用户口音、习惯用词等个性化因素所造成的识别困难和耗时长的问题,提出一种基于语音识别的演讲教学安卓App。该系统在提供教学视频的同时,利用语音识别技术为用户提供了演讲稿的语音文字转换、储存以及数据分析,并为用户演讲能力的改进提供实时指导。系统利用词法分析筛选出用户常用词,采用热词优先识别算法,使识别结果贴近用户的想法和个人风格,进而提高语音识别的准确率和用户体验。测试结果表明系统的语音文字转化速率达到每分钟180字,语音识别准确率高于95%。所提系统能够有效提高演讲语音识别速度和准确率,提高用户的体验质量。
Focusing on the issues that users have the requirement to improve their ability of speech and eloquence and the user requirement of speech and eloquence expression and personalized factors,such as accents,idioms,etc.,making recognition difficult and time-consuming in the application of speech recognition,a speech teaching App based on speech recognition was designed. It not only provides teaching video,but also provides speech-to-text conversion,storage and data analysis for users by using automatic speech recognition. A real-time guidance for users was also provided to improve their speech ability. Meanwhile,the hot words were identified in priority to make the recognition results closer to users more humanized to improve identification accuracy and user experience. The testing results show that the speech is converted to text by 180 words per minute,and the recognition accuracy is over 95%. The proposed system can effectively improve the speed and accuracy of speech recognition and improve the quality of user experience.
作者
戴瑾
巫丰
DAI Jin;WU Feng(College of Information Science and Engineering,Jinling College of Nanjing University,Nanjing Jiangsu 210089,China)
出处
《计算机应用》
CSCD
北大核心
2020年第S01期289-292,共4页
journal of Computer Applications
基金
国家自然科学基金资助项目(61371035)。
关键词
自动语音识别
演讲表达
词法分析
语音文字转换
App
Automatic Speech Recognition(ASR)
speech expression
lexical analysis
speech-to-text conversion
App
