Chinese speech recognition system and method
作者： Sin-Horng Chen , Yih-Ru Wang , Chen-Yu Chiang , Yuan-Fu Liao , Ming-Chieh Liu
DOI:
关键词: Word (computer architecture) 、 Artificial intelligence 、 State model 、 Structure (mathematical logic) 、 Speech recognition 、 Segmentation 、 Factored language model 、 Natural language processing 、 Computer science 、 Syllable 、 SIGNAL (programming language) 、 Speech synthesis
摘要: A Chinese speech recognition system and method is disclosed. Firstly, a signal received recognized to output word lattice. Next, the lattice received, arcs of are rescored reranked with prosodic break model, state syllable prosodic-acoustic syllable-juncture model factored language so as tag, tag phonetic segmentation which correspond signal. The present invention performs rescoring in two-stage way promote rate basic information labels provide structure for rear-stage voice conversion synthesis.
nctu.edu.tw参考文章(3)
Mari Ostendorf, Rebecca Bates, Izhak Shafran, PROSODY MODELS FOR CONVERSATIONAL SPEECH RECOGNITION ,(2003)
Sankaranarayanan Ananthakrishnan, Shrikanth Narayanan, Improved Speech Recognition using Acoustic and Lexical Correlates of Pitch Accent in a N-Best Rescoring Framework international conference on acoustics, speech, and signal processing. ,vol. 4, pp. 873- 876 ,(2007) , 10.1109/ICASSP.2007.367209
Chen-Yu Chiang, Sin-Horng Chen, Hsiu-Min Yu, Yih-Ru Wang, Unsupervised joint prosody labeling and modeling for Mandarin speech. Journal of the Acoustical Society of America. ,vol. 125, pp. 1164- 1183 ,(2009) , 10.1121/1.3056559