摘要
基于标准美音声学模型对英语朗读发音进行评测,准确性相对较低,因此本文设计了英语朗读发音质量自适应评测模型,并以实验分析测试了自适应评测模型。结果表明,系统可准确测评朗读发音质量,从而切实反应朗读者英语口语水平;模型在评测朗读者英语发音时适应性与标准性较高,基于发音准确性、流利性、完整性,可全面衡量朗读发音质量;智能校对系统基于灭错计算进行发音识别评测与校对,可提升语音识别能力,提高评测准确性,以根本上避免评测混乱;利用支持向量回实现特征融合,接近评分特征与人工评分的非线性关系,可保障发音质量评测准确性与可靠性。
Based on the standard American acoustic model,the accuracy of English pronunciation evaluation is relatively low,so this paper designs an adaptive evaluation model of English pronunciation quality,and tests the adaptive evaluation model with experimental analysis.The results show that the system can accurately evaluate the pronunciation quality of the readers,so as to effectively reflect the oral English level of the readers;the model has a high adaptability and standard when evalu-ating the English pronunciation of the readers,and can comprehensively measure the pronunciation quality of the readers based on the accuracy,fluency and integrity of the pronunciation;the intelligent proofreading system can evaluate and proofread the pronunciation recognition based on the error elimination calculation,so as to improve the voice recognition In order to avoid the confusion of evaluation,we can improve the accuracy of evaluation,and use support vector regres-sion to achieve feature fusion,which is close to the non-linear relationship between scoring features and artificial scoring,which can guarantee the accuracy and reliability of pronunciation quality evaluation.
作者
臧瑞婷
ZANG Rui-ting(Xianyang Normal University,Xianyang 712000 China)
出处
《自动化技术与应用》
2021年第10期117-120,共4页
Techniques of Automation and Applications
关键词
英语朗读
发音质量
自适应
评测模型
English reading
pronunciation quality
self-adaption
evaluation model
