Very large vocabulary proper name recognition for directory assistance
Tóm tắt
This paper deals with the difficult task of recognition of a large vocabulary of proper names in a directory assistance application. After a presentation of the related work, it introduces a methodology for rescoring the N-best hypotheses generated by a first step recognition. First experiments give encouraging results and several topics for future research are presented.
Từ khóa
#Vocabulary #Acoustic distortion #Speech recognition #Error analysis #Lattices #Research and development #Robustness #Automatic speech recognition #Hidden Markov modelsTài liệu tham khảo
10.1109/ICASSP.2001.940897
kim, 2000, An Effective Acoustic Modeling of Names Based on Model Induction, Proc IEEEI CASSP
kim, 2000, A Rule-Based Named Entity Recognition System for Speech Input, Proc ICSLP
korkmazskiy, 1999, Pronunciation Modeling with Genetic Algorithms, Proc ASRU Workshop, 528
10.1109/ICASSP.1998.674429
riley, 1999, Stochastic Pronunciation Modeling from, Hand Labeled Phonetic Corpora Speech Communication, 29, 2
suchato, 2000, Framework for Joint Recognition of Pronounced and Spelled Proper Names, MIT
10.1016/S0167-6393(99)00038-2
witt, 1999, Off-line Acoustic Modelling of Non-Native Accents, Proc EUROSPEECH, 1367
10.1016/S0167-6393(99)00034-5
chomsky, 1968, The Sound Pattern of English
deshmukh, 1996, Automatic Generation of N-Best Pronunciations of Proper Nouns, Proc ICASSP, 283
10.1109/ICASSP.2001.940894
fosler-lussier, 1999, Contextual Word and Syllable Pronunciation Models, Proc ASRU Workshop
10.1016/S0167-6393(99)00035-7
byrne, 1998, Is Automatic Speech Recognition Ready for Non-Native Speakers, Proc STiLL, 37
boves, 2000, ASR for Automatic Directory Assistance: the SMADA Project, Proc ASR
gao, 2001, Innovative Approaches for Large Vocabulary Name Recognition, Proc ICASSP