Multilingual acoustic models for the recognition of non-native speech
Tóm tắt
We report on the use of multilingual hidden Markov models for the recognition of non-native speech. Based on the design of a common phoneme set that provides a phone compression rate of almost 80 percent compared to a conglomerate of language dependent phone sets, we create acoustic models that share training data from up to 5 languages. Results obtained on two different data bases of non-native English demonstrate the feasibility of the approach, showing improved recognition accuracy in case of sparse training material, and also for speakers whose native language is not in the training data.
Từ khóa
#Speech recognition #Natural languages #Hidden Markov models #Training data #Loudspeakers #Vocabulary #Noise robustness #Speech enhancement #Adaptation model #Information systemsTài liệu tham khảo
petrovska, 1998, Poly-cost: A Telephone-Speech Database for Speaker Recognition, Proc of the Workshop on Speaker Recognition and its Commercial and Forensic Applications
schultz, 2000, Multilinguale Spracherkennung: Kombination akustischer Modelle zur Portiening auf neue Sprachen
schultz, 2000, Language Portability in Acoustic Modeling, Proc of the IEEE Workshop on Multilingual Speech Communications
uebler, 1999, Recognition of Non-Native German Speech with Multilingual Recognizers, Proc 6th European Conf on Speech Communication and Technology, 2, 911
ward, 1998, Towards Speech Understanding Across Multiple Languages, Proc Int Conf Spoken Language Processing
bonaventura, 1997, Multilingual Speech Recognition for Flexible Vocabularies, Proc of the 5th European Conference on Speech Communication andTechnology, 355
beyerlein, 1999, Towards language independent acoustic modeling, Workshop Automatic Speech Recognition Understanding ASRU) 1999
fischer, 1998, Speaker independent upfront dialect adaptation in a large vocabulary continuous speech recognizer, Proc Int Conf Spoken Language Processing
chen, 1998, Clustering via the Bayesian Information Criterion with Applications to Speech Recognition, Proc of the IEEE Int Conference on Acoustics Speech and SignalProcessing, 645
10.1109/89.279278
fischer, 2000, Towards Multilingual Acoustic Modeling for Large Vocabulary Continuous Speech Recognition, Proc of the IEEE Workshop on Multilingual Speech Communications
10.1109/ICASSP.1994.389238
1993, International Phonetic Association. IPA chart, revised version of: The Principles of the International Phonetic Association, 1949, Journal of the International Phonetic Association, 1
cambra, 2000, Towards a common alphabet for multilingual speech recognition, Proc Int Conf Spoken Language Processing