摘要
针对传统英语机器人识别准确率低的问题,提出基于语音特征提取的英语机器人识别方法。基于HMM模型,加入分段平均算法和聚类交叉分组算法,构建一个基于HMM分组模型的语音识别系统。在系统中输入初始语音后,利用MFCC进行语音特征提取和数据预处理,最后采用HMM分组模型输出识别结果。仿真结果表明,相较于改进前的HMM算法和ANN算法,本算法识别准确率高达99.65%,比另外两种算法分别高出了4.23%和1.82%,且在不同实验环境下,本系统的语音识别准确率均保持在90%及以上。综合分析可知,本方法能够实现英语机器人的准确识别,识别效率显著提升,具备一定的可行性和有效性。
Based on the low recognition accuracy of traditional English robot recognition,a recognition method based on speech feature extraction is proposed.Based on HMM model,segment averaging algorithm and cluster cross-grouping algorithm are added to construct a speech recognition system based on HMM grouping model.After inputting the initial speech in the system,the MFCC is used for speech feature extraction and data preprocessing,and finally the HMM grouping model is used to output the identification results.Simulation results show that compared with the improved HMM and ANN algorithms,the recognition accuracy of this algorithm is as high as 99.65%,13.6% and 9.8% higher than the other two algorithms,and the speech recognition accuracy of this system is maintained at 90% or above in different experimental environments.Comprehensive analysis shows that this method can realize the accurate identification of English robot,significantly improve the identification efficiency,and has certain feasibility and effectiveness.
作者
陈琳
CHEN Lin(Xianyang Normal University,Xianyang Shaanxi 712000,China)
出处
《自动化与仪器仪表》
2022年第8期234-239,共6页
Automation & Instrumentation
基金
咸阳师范学院教改项目《基于协同教学的地方院校卓越英语师范生培养模式研究》(2019Y048)
咸阳师范学院教改项目《基于“1-2-1-1”教学理念的“英语公众演讲”课程思政之路径与时间》。
