专利名称：Discriminative training of document
transcription system
发明人：Lambert Mathias,Girija
Yegnanarayanan,Juergen Fritsch
申请号：US14244053申请日：20140403公开号：US09190050B2公开日：20151117
专利附图：
摘要：A system is provided for training an acoustic model for use in speechrecognition. In particular, such a system may be used to perform training based on a
spoken audio stream and a non-literal transcript of the spoken audio stream. Such asystem may identify text in the non-literal transcript which represents concepts havingmultiple spoken forms. The system may attempt to identify the actual spoken form inthe audio stream which produced the corresponding text in the non-literal transcript, andthereby produce a revised transcript which more accurately represents the spoken audiostream. The revised, and more accurate, transcript may be used to train the acousticmodel using discriminative training techniques, thereby producing a better acousticmodel than that which would be produced using conventional techniques, which performtraining based directly on the original non-literal transcript.
申请人：MModal IP LLC
地址：Franklin TN US
国籍：US
代理机构：Robert Plotkin, P.C.
更多信息请下载全文后查看
因篇幅问题不能全部显示，请点此查看更多更全内容