Finite-state transducers for speech-input translation

F. Casacuberta1
1Dpt. de Sistemes Infomàtics i Computació, Institut Tecnològic dE28099Infomàtica, Universitat Politècnica de València, Valencia, Spain

Tóm tắt

Nowadays, hidden Markov models (HMMs) and n-grams are the basic components of the most successful speech recognition systems. In such systems, HMMs (the acoustic models) are integrated into a n-gram or a stochastic finite-state grammar (the language model). Similar models can be used for speech translation, and HMMs (the acoustic models) can be integrated into a finite-state transducer (the translation model). Moreover, the translation process can be performed by searching for an optimal path of states in the integrated network. The output of this search process is a target word sequence associated to the optimal path. In speech translation, HMMs can be trained from a source speech corpus, and the translation model can be learned automatically from a parallel training corpus. This approach has been assessed in the framework of the EUTRANS project, founded by the European Union. Extensive speech-input experiments have been carried out with translations from Spanish to English and from Italian to English translation, in an application involving the interaction (by telephone) of a customer with a receptionist at the front-desk of a hotel. A summary of the most relevant results are presented in this paper.

Từ khóa

#Hidden Markov models #Acoustic transducers #Stochastic processes #Speech recognition #Stochastic systems #Telephony #Prototypes #Search engines #Natural languages #Decoding

Tài liệu tham khảo

brown, 1993, The mathematics of statistical machine translation: Parameter estimation, Computational Linguistics, 19, 263 10.1109/89.817451 10.1023/A:1010880113956 llorens, 2000, Suavizado de autmatas y traductores fini-los estocsticos vidal, 1993, Learning subsequential transducers for pattern recognition interpretation tasks, IEEE Transactions on Pattern Analysis and Machine Intelligence, 15, 448, 10.1109/34.211465 10.1109/ICASSP.1999.758176 10.1007/978-94-017-1183-8_2 llorens, 1999, Acustical and syntactical modeling in ATROS system, Proceedings of the IEEE International Conference on Acoustic Speech and Signal Processing, 641 young, 1997, The HTK Book (Version 2 1) Cambridge University Department and Entropic Research Laboratories Inc 10.1109/ICASSP.1998.675399 amengual, 2001, The EUTRANS-I speech translation system, J Mach Translat, 15, 75, 10.1023/A:1011116115948 10.1109/ICASSP.1997.599563 bangalore, 2000, Stochastic finite-state models for spoken language machine translation, in Workshop on Embeded Machine Translation Systems 10.1109/ICASSP.2001.940906 10.1007/BFb0033362 bangalore, 2001, A finite-state approach to machine translation, The Second Meeting of the North American Chapter of the Association for Computational Linguistics, 10.3115/1073336.1073354 vilar, 2000, Improve the learning of subsequential transducers by using alignments and dictionaries, Grammatical Inference Algorithms and Applications, 1891, 298, 10.1007/978-3-540-45257-7_24 casacuberta, 2000, Inference of finite-state transducers by using regular grammars and morphisms, Gram-matical Inference Algorithms and Applications, 1891, 1, 10.1007/978-3-540-45257-7_1 brown, 1990, A statistical approach to machine translation, Computational Linguistics, 16, 79 casacuberta, 0, Statistical and finite-state approaches to speech-to-speech translation