Title :
Automatic Training Set Segmentation for Multi-pass Speech Recognition
Author :
Mao, Mark Z. ; Vanhoucke, Vincent ; Strope, Brian
Author_Institution :
Dept. of Electr. Eng., Stanford Univ., CA, USA
fDate :
March 18-23, 2005
Keywords :
acoustic signal processing; learning (artificial intelligence); natural languages; speech recognition; American-English digit strings; acoustic models; automatic speech recognition; automatic training set segmentation; channel labeling; gender labeling; multi-pass speech recognition; relative error rate reduction; second pass model training; training data clustering; Automatic speech recognition; Character recognition; Clustering algorithms; Decoding; Hidden Markov models; Labeling; Loudspeakers; Speech recognition; Target recognition; Training data;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2005. Proceedings. (ICASSP '05). IEEE International Conference on
Print_ISBN :
0-7803-8874-7
DOI :
10.1109/ICASSP.2005.1415206
