欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
首先,给出结合韵律信息的系统框架。然后,针对汉语的特点,解决了韵律相关的语音识别系统中建模单元选择、模型训练等问题,并在多空间概率分布隐马尔可夫模型（multiple-space distribution hidden Markov mod-el,MSD-HMM）框架下构建了韵律相关的语音识别系统。最后,通过语音识别的实验验证了方法的有效性。在＂863＂测试集上,该方法能够达到76.18%的带调音节识别正确率。
英文摘要：
First,this paper proposed the prosody dependent system framework.Then,according to the characters about Chinese mandarin,resolved these problems about modeling unit selection and model training when constructing prosody dependent speech recognition system,and built prosody dependent speech recognition system under MSD-HMM framework.Finally,verified the proposed method through the experiments of speech recognition.In ＂863＂ testing set,the proposed method can achieve 76.18% correct rate of the tonal syllable.
同期刊论文项目
同项目期刊论文
位置：