Language model adaptation in speech recognition using document maps
Tóm tắt
We present speech experiments that were carried out to evaluate a topically focusing language model in large vocabulary speech recognition. An ordered topical clustering is first computed as a self-organized mapping of a large document collection. Language models are then trained for each text cluster or for several neighboring clusters. The obtained organized collection of language models is efficiently utilized in continuous speech recognition to concentrate on the model that corresponds closest to the current topic of discussion. The speech recognition experiments are carried out on a novel Finnish speech database. A property of Finnish that is particularly challenging for speech recognition is the extremely fast vocabulary growth that makes many of the standard word-based language modeling methods impractical for large vocabulary tasks.
Từ khóa
#Natural languages #Adaptation model #Speech recognition #Vocabulary #Probability #Intelligent networks #Neural networks #Speech analysis #Databases #Ultraviolet sourcesTài liệu tham khảo
kohonen, 2001, Self-Organizing Maps, 10.1007/978-3-642-56927-2
10.1109/72.846729
kurimo, 1997, Using Self-Organizing Maps and Learning Vector Quantization for Mixture Density Hidden Markov Models
kurimo, 2002, An Efficiently Focusing Large Vocabulary Language Model, ICANN'06 International Conference on Artificial Neural Networks
10.1109/ICASSP.1993.319225
10.1109/89.784107
siivola, 2001, Large Vocabulary Statistical Language Modeling for Continuous Speech Recognition, Proc of 7th European Conference on Speech Communication and Technology
2001, CSC-Tieteellinen laskenta Oy Finnish Language Text Bank Corpora Books Newspapers Magazines and Other
clarkson, 1997, Statistical language Modeling Using CMU-Cambridge Toolkit, Proc of 5th European Conference on Speech Communication and Technology, 2707
10.1006/csla.2001.0174
gildea, 1999, Topic-based Language Modeling Using EM, Proc 6th European Conf on Speech Communication and Technology, 2167
10.1109/89.736328
honkela, 1996, Newsgroup exploration with WEBSOM method and browsing interface, TR A32
10.1109/ICASSP.1997.596049
10.1109/5.880084
10.1109/ICASSP.1999.758059