T-modules: Translation modules for zero-shot cross-modal machine translation

PA Duquenne, H Gong, B Sagot, H Schwenk - arXiv preprint arXiv …, 2022 - arxiv.org
We present a new approach to perform zero-shot cross-modal transfer between speech and
text for translation tasks. Multilingual speech and text are encoded in a joint fixed-size …

Structured literature review of published research on indirect translation (2017–2022)

H Pięta, L Ivaska, Y Gambier - Perspectives, 2023 - Taylor & Francis
Drawing on a structured literature review, this article offers a meta-analysis of published
research on indirect translation in different domains between 2017 and 2022. The article first …

Revamping multilingual agreement bidirectionally via switched back-translation for multilingual neural machine translation

H Lu, H Huang, S Ma, D Zhang, F Wei… - arXiv preprint arXiv …, 2022 - arxiv.org
Despite the fact that multilingual agreement (MA) has shown its importance for multilingual
neural machine translation (MNMT), current methodologies in the field have two …

Leveraging Mandarin as a Pivot Language for Low-Resource Machine Translation between Cantonese and English

KY Suen, R Chow, AYS Lam - Proceedings of the Seventh …, 2024 - aclanthology.org
Cantonese, the second most prevalent Chinese dialect after Mandarin, has been relatively
overlooked in machine translation (MT) due to a scarcity of bilingual resources. In this paper …

End-to-end Training and Decoding for Pivot-based Cascaded Translation Model

H Cheng, M Zhang, L Li, Q Liu, Z Zhang - arXiv preprint arXiv:2305.02261, 2023 - arxiv.org
Utilizing pivot language effectively can significantly improve low-resource machine
translation. Usually, the two translation models, source-pivot and pivot-target, are trained …

Sentence Embeddings for Massively Multilingual Speech and Text Processing

PA Duquenne - 2024 - theses.hal.science
Representation learning of sentences has been widely studied in NLP. While many works
have explored different pre-training objectives to create contextual representations from …