基于隐马尔可夫模型的连续语音同步识别系统  

Continuous speech synchronization recognition system based on hidden Markov model

作　　者：李玉华[1] LI Yuhua(Nanchang University College of Science and Technology,Nanchang 330029,China)

机构地区：[1]南昌大学科学技术学院

出　　处：《现代电子技术》2019年第11期64-67,71,共5页Modern Electronics Technique

基　　金：江西省级教改课题:网络环境下独立学院非英语专业大学生英语听力学习策略的研究与实践(JXJG-08-78-17);江西省社会科学规划项目:基于CBI的大学英语ESP教学改革探究——以独立学院“旅游管理”专业为例(15WX316)~~

摘　　要：语音同步识别系统的发展方向是连续性的人机交互,采用传统系统易受到突发性噪声影响,致使识别效果较差,提出基于隐马尔可夫模型的连续语音同步识别系统。结合语音识别原理,设计系统硬件总体结构。利用JFET输入高保真运放的OPA604低通滤波器,保证信号处理结果的有效性。通过OMAP5912ZZG型号芯片对处理后的信号进行存储,使用矢量图缓冲音频,经由以太网接口移植相关语音识别序列,由此实现连续语音同步识别。由实验对比结果可知,该系统比传统系统识别效果最高值高出48%,推进了语音识别技术研究的快速发展。The current development direction of the speech synchronization recognition system is successive human-computer interaction.The traditional system is easily affected by the sudden noise,which may cause the poor recognition effect.Therefore,a continuous speech recognition system based on hidden Markov model is proposed.In combination with the principle of speech recognition,the overall hardware structure of the system is designed.The low-pass filter of JFET-input high-fidelity operational amplifier OPA604 is utilized to ensure the validity of signal processing results.The chip OMAP5912ZZG is used to store the processed signals after acquisition.The vector map is used to buffer the audio frequency signal,and transplant the related speech recognition sequence via the Ethernet interface,thus the continuous speech synchronization recognition is realized.The experimental results show that the recognition effect of the proposed system is 48%higher than that of the traditional system,and the system can promote the rapid development of speech recognition technology research.

关 键 词：隐马尔可夫模型 连续语音识别 同步识别 信号处理 人机交互 系统结构设计 

分 类 号：TN912.34-34[电子电信—通信与信息系统] TP391.42[电子电信—信息与通信工程]