TY - GEN
T1 - Bayesian Models for Unit Discovery on a Very Low Resource Language
AU - Ondel, Lucas
AU - Godard, Pierre
AU - Besacier, Laurent
AU - Larsen, Elin
AU - Hasegawa-Johnson, Mark
AU - Scharenborg, Odette
AU - Dupoux, Emmanuel
AU - Burget, Lukas
AU - Yvon, Francois
AU - Khudanpur, Sanjeev
N1 - Funding Information:
This work was started at JSALT 2017 in CMU, Pittsburgh, and was supported by JHU and CMU (via grants from Google, Microsoft, Amazon, Facebook, Apple), by the Czech Ministry of Education, Youth and Sports from the National Programme of Sustainability (NPU II) project ”IT4Innovations excellence in science - LQ1602” and by the French ANR and the German DFG under grant ANR-14-CE35-0002 (BULB project). This work used the Extreme Science and Engineering Discovery Environment (NSF grant number OCI-1053575 and NSF award number ACI-1445606).
Funding Information:
O. Scharenborg was supported by a Vidi-grant from NWO (grant number: 276-89-003)
Publisher Copyright:
© 2018 IEEE.
PY - 2018/9/10
Y1 - 2018/9/10
N2 - 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 I-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.
AB - 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 I-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.
KW - Acoustic Unit Discovery
KW - Bayesian Model
KW - Informative Prior
KW - Low-Resource ASR
UR - http://www.scopus.com/inward/record.url?scp=85054248119&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85054248119&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2018.8461545
DO - 10.1109/ICASSP.2018.8461545
M3 - Conference contribution
AN - SCOPUS:85054248119
SN - 9781538646588
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 5939
EP - 5943
BT - 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018
Y2 - 15 April 2018 through 20 April 2018
ER -