英语发音通过嵌入式实时系统的识别设计及功能实现  

The Recognition Design and Function Realization of English Pronunciation Through Embedded Real-time System

作　　者：赵丽娜[1] Zhao Lina(Xi'an Peihua College,Xi'an 710125)

机构地区：[1]西安培华学院,陕西西安710125

出　　处：《中阿科技论坛（中英文）》2021年第1期85-88,共4页China-Arab States Science and Technology Forum

基　　金：西安培华学院校级课题“基于语音识别系统设计及功能在英语发音中的实现研究”(PHKT2051)。

摘　　要：本文基于连续隐含马尔可夫模型完成识别框架的构建,对通过使用嵌入式DSP语音芯片实现英语发音识别功能,并且综合运用时频分析及提取关联信息熵特征完成系统发音误差自动检测进行探讨。研究发现通过使用该嵌入式实时英语发音识别系统可有效提高英语发音的识别效率、准确率,以及系统发音误差自动检测能力,具有较高应用价值,可为优化英语发音识别系统的功能及应用效益研究提供参考。This paper adopts the optimal design of embedded technology sound recognition system,completes the construction of recognition framework based on continuous hidden Markov model,and realizes the function of English pronunciation recognition by using embedded DSP speech chip.By using time-frequency analysis and extracting entropy feature of related information,the automatic detection method of pronunciation error is designed.The experimental results show that the English pronunciation recognition system is feasible and improves the recognition efficiency and accuracy of English pronunciation,It improves the ability of automatic detection of pronunciation error and provides a reference for optimizing the function and application benefit of English pronunciation recognition system.

关 键 词：实现路径 发音误差自动检测 实时英语发音识别系统 

分 类 号：TN912.34[电子电信—通信与信息系统]