A one-pass decoder based on polymorphic linguistic context assignment
Tóm tắt
In this study, we examine how fast decoding of conversational speech with large vocabularies profits from efficient use of linguistic information, i.e. language models and grammars. Based on a re-entrant single pronunciation prefix tree, we use the concept of linguistic context polymorphism to allow an early incorporation of language model information. This approach allows us to use all available language model information in a one-pass decoder, using the same engine to decode with statistical n-gram language models as well as context free grammars or re-scoring of lattices in an efficient way. We compare this approach to our previous decoder, which needed three passes to incorporate all available information. The results on a very large vocabulary task show that the search can be speeded up by almost a factor of three, without introducing additional search errors.
Từ khóa
#Decoding #Context modeling #Automatic speech recognition #Vocabulary #Acoustic beams #Speech recognition #History #Interactive systems #Laboratories #Natural languagesTài liệu tham khảo
alex, 2001, Advances in meeting recogniton, Proceedings of the First International Conference on Human Language Technology Conference
aubert, 1999, One pass cross word decoding for large vocabularies based. on a lexical tree search organization, Proceedings of the Eurospeech
ravishankar, 1996, Efficient Algorithms for Speech Recognition
10.1109/ICASSP.1996.543251
soltau, 2001, The ISL evaluation system for Verbmobil-IL, Proceedings of the ICASSP
finke, 1999, Modeling and efficient decoding of large vocabulary conversational speech, Proc EU-ROSPEECH
10.1109/ICASSP.1996.540309
odell, 1995, The use of context in large vocabulary speech recognition
ney, 1992, Improvements in beam search for 100000-word continuous speech recognition, Proceedings of the ICASSP
10.1109/ICASSP.1995.479666
10.1109/ICASSP.1998.675390
10.1109/ICASSP.1996.540308