Abstract
End-to-End speech translation usually leverages audio-to-text parallel data to train an available speech translation model which has shown impressive results on various speech translation tasks. Due to the artificial cost of collecting audio-to-text parallel data, the speech translation is a natural low-resource translation scenario, which greatly hinders its improvement. In this paper, we proposed a new adversarial training method to leverage target monolingual data to relieve the low-resource shortcoming of speech translation. In our method, the existing speech translation model is considered as a Generator to gain a target language output, and another neural Discriminator is used to guide the distinction between outputs of speech translation model and true target monolingual sentences. Experimental results on the CCMT 2019-BSTC dataset speech translation task demonstrate that the proposed methods can significantly improve the performance of the End-to-End speech translation system.- Anthology ID:
- 2020.autosimtrans-1.2
- Volume:
- Proceedings of the First Workshop on Automatic Simultaneous Translation
- Month:
- July
- Year:
- 2020
- Address:
- Seattle, Washington
- Venue:
- AutoSimTrans
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 10–14
- Language:
- URL:
- https://aclanthology.org/2020.autosimtrans-1.2
- DOI:
- 10.18653/v1/2020.autosimtrans-1.2
- Cite (ACL):
- Xuancai Li, Chen Kehai, Tiejun Zhao, and Muyun Yang. 2020. End-to-End Speech Translation with Adversarial Training. In Proceedings of the First Workshop on Automatic Simultaneous Translation, pages 10–14, Seattle, Washington. Association for Computational Linguistics.
- Cite (Informal):
- End-to-End Speech Translation with Adversarial Training (Li et al., AutoSimTrans 2020)
- PDF:
- https://preview.aclanthology.org/auto-file-uploads/2020.autosimtrans-1.2.pdf