Multilingual deep bottle neck features: a study on language selection and training techniques

Markus Müller, Sebastian Stüker, Zaid Sheikh, Florian Metze, Alex Waibel

Abstract
Previous work has shown that training the neural networks for bottle neck feature extraction in a multilingual way can lead to improvements in word error rate and average term weighted value in a telephone key word search task. In this work we conduct a systematic study on a) which multilingual training strategy to employ, b) the effect of language selection and amount of multilingual training data used and c) how to find a suitable combination for languages. We conducted our experiment on the key word search task and the languages of the IARPA BABEL program. In a first step, we assessed the performance of a single language out of all available languages in combination with the target language. Based on these results, we then combined a multitude of languages. We also examined the influence of the amount of training data per language, as well as different techniques for combining the languages during network training. Our experiments show that data from arbitrary additional languages does not necessarily increase the performance of a system. But when combining a suitable set of languages, a significant gain in performance can be achieved.
Anthology ID:
2014.iwslt-papers.15
Volume:
Proceedings of the 11th International Workshop on Spoken Language Translation: Papers
Month:
December 4-5
Year:
2014
Address:
Lake Tahoe, California
Editors:
Marcello Federico, Sebastian Stüker, François Yvon
Venue:
IWSLT
SIG:
SIGSLT
Publisher:
Note:
Pages:
257–264
Language:
URL:
https://aclanthology.org/2014.iwslt-papers.15
DOI:
Bibkey:
Cite (ACL):
Markus Müller, Sebastian Stüker, Zaid Sheikh, Florian Metze, and Alex Waibel. 2014. Multilingual deep bottle neck features: a study on language selection and training techniques. In Proceedings of the 11th International Workshop on Spoken Language Translation: Papers, pages 257–264, Lake Tahoe, California.
Cite (Informal):
Multilingual deep bottle neck features: a study on language selection and training techniques (Müller et al., IWSLT 2014)
Copy Citation:
PDF:
https://preview.aclanthology.org/emnlp-22-attachments/2014.iwslt-papers.15.pdf
PDF Search