Jointly Trained Transformers Models for Spoken Language Translation

conference paper

English

End-to-End and cascade (ASR-MT) spoken language translation (SLT) systems are reaching comparable performances, however, a large degradation is observed when translating the ASR hypothesis in comparison to using oracle input text. In this work, degradation in performance is reduced by creating an End-to-End differentiable pipeline between the ASR and MT systems. In this work, we train SLT systems with ASR objective as an auxiliary loss and both the networks are connected through the neural hidden representations. This training has an End-to-End differentiable path with respect to the final objective function and utilizes the ASR objective for better optimization. This architecture has improved the BLEU score from 41.21 to 44.69. Ensembling the proposed architecture with independently trained ASR and MT systems further improved the BLEU score from 44.69 to 46.9. All the experiments are reported on English-Portuguese speech translation task using the How2 corpus. The final BLEU score is on-par with the best speech translation system on How2 dataset without using any additional training data and language model and using fewer parameters.

Spoken Language Translation, Transformers, Joint training, How2 dataset, Auxiliary loss, ASR objective, Coupled decoding, End-to-End differentiable pipeline.

VYDANA, H.; KARAFIÁT, M.; ŽMOLÍKOVÁ, K.; BURGET, L.; ČERNOCKÝ, J.

6. 6. 2021

IEEE Signal Processing Society

Toronto, Ontario

978-1-7281-7605-5

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

7513

7517

5

https://www.fit.vut.cz/research/publication/12522/