Zobrazeno 1 - 10
of 17
pro vyhledávání: '"Ankur Bapna"'
End-to-end speech-to-speech translation (S2ST) without relying on intermediate text representations is a rapidly emerging frontier of research. Recent works have demonstrated that the performance of such direct S2ST systems is approaching that of con
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::8256e1b689fe46f847abaa17c546580f
http://arxiv.org/abs/2203.13339
http://arxiv.org/abs/2203.13339
Autor:
Takaaki Saeki, Heiga Zen, Zhehuai Chen, Nobuyuki Morioka, Gary Wang, Yu Zhang, Ankur Bapna, Andrew Rosenberg, Bhuvana Ramabhadran
This paper proposes Virtuoso, a massively multilingual speech-text joint semi-supervised learning framework for text-to-speech synthesis (TTS) models. Existing multilingual TTS typically supports tens of languages, which are a small fraction of the t
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::52500f6059f16130fbb660682cd9fdfb
Much of text-to-speech research relies on human evaluation, which incurs heavy costs and slows down the development process. The problem is particularly acute in heavily multilingual applications, where recruiting and polling judges can take weeks. W
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::b3c2078509e36e897d48e0ce5cd4abb0
Autor:
Alexis Conneau, Ankur Bapna, Yu Zhang, Min Ma, Patrick von Platen, Anton Lozhkov, Colin Cherry, Ye Jia, Clara Rivera, Mihir Kale, Daan van Esch, Vera Axelrod, Simran Khanuja, Jonathan Clark, Orhan Firat, Michael Auli, Sebastian Ruder, Jason Riesa, Melvin Johnson
We introduce XTREME-S, a new benchmark to evaluate universal cross-lingual speech representations in many languages. XTREME-S covers four task families: speech recognition, classification, speech-to-text translation and retrieval. Covering 102 langua
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::7d5112276248b7b6fa2427d9ed857587
Document-level neural machine translation (DocNMT) achieves coherent translations by incorporating cross-sentence context. However, for most language pairs there's a shortage of parallel documents, although parallel sentences are readily available. I
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::4969ea8b265a7546b3c3dc08a3d513ad
Publikováno v:
COLING
Large text corpora are increasingly important for a wide variety of Natural Language Processing (NLP) tasks, and automatic language identification (LangID) is a core technology needed to collect such datasets in a multilingual context. LangID is larg
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::25cfb0764a492e7b9e8ed9f2843fdbdc
http://arxiv.org/abs/2010.14571
http://arxiv.org/abs/2010.14571
Autor:
Ankur Bapna, Yuan Cao, Aditya Siddhant, Mia Xu Chen, Sneha Kudugunta, Orhan Firat, Naveen Arivazhagan, Yonghui Wu
Publikováno v:
ACL
Over the last few years two promising research directions in low-resource neural machine translation (NMT) have emerged. The first focuses on utilizing high-resource languages to improve the quality of low-resource languages via multilingual NMT. The
Autor:
Yonghui Wu, Zhifeng Chen, Eugene Weinstein, Tara N. Sainath, Seungji Lee, Anjuli Kannan, Arindrima Datta, Ankur Bapna, Bhuvana Ramabhadran
Publikováno v:
INTERSPEECH
Multilingual end-to-end (E2E) models have shown great promise in expansion of automatic speech recognition (ASR) coverage of the world's languages. They have shown improvement over monolingual systems, and have simplified training and serving by elim
Publikováno v:
EMNLP/IJCNLP (1)
Multilingual Neural Machine Translation (NMT) models have yielded large empirical success in transfer learning settings. However, these black-box representations are poorly understood, and their mode of transfer remains elusive. In this work, we atte
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::06bf1b6717501894c86c27917895fe79
http://arxiv.org/abs/1909.02197
http://arxiv.org/abs/1909.02197
Autor:
Ankur Bapna, Orhan Firat
Publikováno v:
NAACL-HLT (1)
Neural Networks trained with gradient descent are known to be susceptible to catastrophic forgetting caused by parameter shift during the training process. In the context of Neural Machine Translation (NMT) this results in poor performance on heterog
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::8ff63a850c17b43dcb3b5f161d2ee9a8
http://arxiv.org/abs/1903.00058
http://arxiv.org/abs/1903.00058