摘要
本系统基于反馈调节HMM改进模型方法,依据HMM数学模型原理,找到这种模型的内在缺陷,提出一种反馈学习的修正原理,通过实验找出最优数据,正确反映汉语语音的数学模型,大大提高语音识别率。
Based on the feedback adjust,the intrinsic defects of the HMM mathematical model are found and a modified principle of feedback study is suggested. Then the optimal data are obtained experimentally. The resultant mathematical model can correctly reflect the pronunciation of Chinese language and the automatic speech recognition efficiency is greatly improved.
出处
《吉林工学院学报(自然科学版)》
CAS
2002年第4期46-48,共3页
Journal of Jilin Institute of Technology