摘要
针对传统英语发音学习过程中存在的发音不准确、缺乏发音评价和纠错指导的现状。提出了基于Android平台开发一款发音跟读、发音比对、发音评分和纠错多功能应用的英语发音训练系统。基于短时过零率端点检测进行语音预处理,获得较为稳定的语音段信号,利用美尔倒普系数(MFCC)提取语音信号特征值,获得每帧语音频谱特性。通过矢量间距离计算表征信号的匹配度,在自适应(AP)评价法来实现平均匹配距离与发音评分间的逻辑关系,得到发音共振峰包络图,利用生成的发音共振峰对比图构建发音共振峰和读音口型模型进行发音跟读质量反馈。实际应用结果表明:开发的系统应用能够准确进行定性化的发音口型纠正,有效满足现代英语学习过程中的智能化、实时性和便携化需求。
In view of the present situation of inaccurate pronunciation,lack of pronunciation evaluation and error correction guidance in the process of traditional English pronunciation learning.An English pronunciation training system based on Android platform is proposed for pronunciation following reading,pronunciation comparison,pronunciation scoring and error correction.speech preprocessing based on short-time zero-crossing rate endpoint detection to obtain a more stable speech segment signal,using the mer inverted coefficient(mfcc)to extract the speech signal eigenvalues and obtain the speech spectrum characteristics of each frame.Using vector-to-vector distance to calculate the matching degree of the representation signal,the average matching distance and pronunciation evaluation are realized in the adaptive(AP)evaluation method In this paper,the phonetic resonance peak envelope map is obtained,and the model of pronunciation resonance peak and pronunciation mouth pattern is constructed by using the generated phonetic resonance peak contrast map.The practical application results show that the developed system can accurately carry out the correction of pronunciation and mouth shape,and effectively meet the needs of intelligence,real-time and portability in the process of modern English learning.
作者
梁玮
LIANG Wei(Public Basic Teaching Department,Zhaoqing Medical College,Zhaoqing,Guangdong 526020,China)
出处
《计算技术与自动化》
2020年第2期155-159,共5页
Computing Technology and Automation
基金
广东省高职教育教学改革课题项目(20171035)。