Publication Details
Bayesian Models for Unit Discovery on a Very Low Resource Language
Godard Pierre (LIMSI)
Besacier Laurent (UGA)
Larsen Elin (INRIA)
Hasegawa-Johnson Mark (UILLINOIS)
Scharenborg Odette (RUN)
Dupoux Emmanuel (ENS)
Burget Lukáš, doc. Ing., Ph.D. (DCGM FIT BUT)
Yvon Francois (GET/ENST)
Khudanpur Sanjeev (JHU)
Acoustic Unit Discovery, Low-Resource ASR, Bayesian Model, Informative Prior.
Developing speech technologies for low-resource languages has become a very active research field over the last decade. Among others, Bayesian models have shown some promising results on artificial examples but still lack of in situ experiments. Our work applies state-of-the-art Bayesian models to unsupervised Acoustic Unit Discovery (AUD) in a real low-resource language scenario. We also show that Bayesian models can naturally integrate information from other resourceful languages by means of informative prior leading to more consistent discovered units. Finally, discovered acoustic units are used, either as the 1-best sequence or as a lattice, to perform word segmentation. Word segmentation results show that this Bayesian approach clearly outperforms a Segmental-DTW baseline on the same corpus.
@INPROCEEDINGS{FITPUB11719, author = "Francois Antoine Lucas Yang Ondel and Pierre Godard and Laurent Besacier and Elin Larsen and Mark Hasegawa-Johnson and Odette Scharenborg and Emmanuel Dupoux and Luk\'{a}\v{s} Burget and Francois Yvon and Sanjeev Khudanpur", title = "Bayesian Models for Unit Discovery on a Very Low Resource Language", pages = "5939--5943", booktitle = "Proceedings of ICASSP 2018", year = 2018, location = "Calgary, CA", publisher = "IEEE Signal Processing Society", ISBN = "978-1-5386-4658-8", doi = "10.1109/ICASSP.2018.8461545", language = "english", url = "https://www.fit.vut.cz/research/publication/11719" }