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

The paper describes BUT's English to German offline speech translation(ST)systems developed for IWSLT2021. They are based on jointly trained AutomaticSpeech Recognition-Machine Translation models. Their performances is evaluatedon MustC-Common test set. In this work, we study their efficiency from theperspective of having a large amount of separate ASR training data and MTtraining data, and a smaller amount of speech-translation training data. Largeamounts of ASR and MT training data are utilized for pre-training the ASR andMT models. Speech-translation data is used to jointly optimize ASR-MT models bydefining an end-to-end differentiable path from speech to translations. Forthis purpose, we use the internal continuous representations from theASR-decoder as the input to MT module. We show that speech translation can befurther improved by training the ASR-decoder jointly with the MT-module usinglarge amount of text-only MT training data. We also show significantimprovements by training an ASR module capable of generating punctuated text,rather than leaving the punctuation task to the MT module.

The IWSLT 2021 BUT Speech Translation Systems