Oral Objective Evaluation and Study Based on Speech Feature Parameters
Starting with the perspective of human sense of hearing,this paper uses the characteristics of speech signal to study extraction algorithm for five kinds of speech feature parameters related to foreign oral language,and takes it as a basis of objective evaluation.Establishing neural network model of objective evaluation,combining technology of speech processing and neural network,it will achieve the objective evaluations of performance on oral practice by computer and uses it as a supplement of subjective evaluation.Test results showed that the five kinds of speech feature parameters extracted from speech signal,when trained by Elman neural network,the objective evaluation of the automatic feedback can be more closely to subjective evaluation value,and takes it as the supplement of subjective evaluation.
speech signal feature parameters objective evaluation neural networks Oral language
Jianchao Yu Ruilin Zhang
Software Engineering Institute,Zhejiang Sci-Tech University Hangzhou,Zhejiang Province,China
国际会议
太原
英文
217-220
2011-02-26（万方平台首次上网日期，不代表论文的发表时间）