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

Transformer models using segment-based processing have been an effectivearchitecture for simultaneous speech translation. However, such models create acontext mismatch between training and inference environments, hinderingpotential translation accuracy. We solve this issue by proposing ShiftableContext, a simple yet effective scheme to ensure that consistent segment andcontext sizes are maintained throughout training and inference, even with thepresence of partially filled segments due to the streaming nature ofsimultaneous translation. Shiftable Context is also broadly applicable tosegment-based transformers for streaming tasks. Our experiments on theEnglish-German, English-French, and English-Spanish language pairs from theMUST-C dataset demonstrate that when applied to the Augmented MemoryTransformer, a state-of-the-art model for simultaneous speech translation, theproposed scheme achieves an average increase of 2.09, 1.83, and 1.95 BLEUscores across each wait-k value for the three language pairs, respectively,with a minimal impact on computation-aware Average Lagging.

Shiftable Context: Addressing Training-Inference Context Mismatch in Simultaneous Speech Translation