English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Research in multilingual speech-to-text translation is topical. Having asingle model that supports multiple translation tasks is desirable. The goal ofthis work it to improve cross-lingual transfer learning in multilingualspeech-to-text translation via semantic knowledge distillation. We show that byinitializing the encoder of the encoder-decoder sequence-to-sequencetranslation model with SAMU-XLS-R, a multilingual speech transformer encodertrained using multi-modal (speech-text) semantic knowledge distillation, weachieve significantly better cross-lingual task knowledge transfer than thebaseline XLS-R, a multilingual speech transformer encoder trained viaself-supervised learning. We demonstrate the effectiveness of our approach ontwo popular datasets, namely, CoVoST-2 and Europarl. On the 21 translationtasks of the CoVoST-2 benchmark, we achieve an average improvement of 12.8 BLEUpoints over the baselines. In the zero-shot translation scenario, we achieve anaverage gain of 18.8 and 11.9 average BLEU points on unseen medium andlow-resource languages. We make similar observations on Europarl speechtranslation benchmark.

Improved Cross-Lingual Transfer Learning For Automatic Speech Translation