When to Use Extra Context: Evidence-Grounded Terminology Adaptation for Simultaneous Speech Translation

2026-07-20

AI Summary

This work addresses the low accuracy of terminology translation in simultaneous interpretation of technical lectures by proposing the EGTA framework. EGTA constructs a document-level terminology memory bank and dynamically selects compact candidate terms during streaming decoding, adaptively constraining the decision spaces of both speech recognition and translation exclusively to these candidates. This approach explicitly attributes contextual gains to the recovery of specific terms and can be integrated into cascaded, end-to-end, and purely generative simultaneous interpretation systems without requiring full-model fine-tuning. Experimental results demonstrate that EGTA significantly improves BLEU and XCOMET-XL scores, as well as recall for named entities and acronyms on the MCIF-dev benchmark, and further validates cross-dataset terminology recall gains on ACL60/60-dev.

Abstract

Extra context is valuable for simultaneous speech translation of technical talks, but injecting the entire document context into every streaming segment is often too coarse. Through diagnostic experiments, we find that context gains mainly come from paper-specific terminology recovery rather than uniform semantic enhancement. We therefore propose EGTA, an Evidence-Grounded Terminology Adaptation framework that builds a document terminology memory, selects compact candidate terms conditioned on the current streaming state, and adapts ASR/speech-side and decoder-side decision spaces using only the selected terms. EGTA can be instantiated in cascaded, end-to-end, and generation-only SimulST settings without full-model fine-tuning. We evaluate EGTA on an ACL technical-talk SimulST evaluation suite consisting of MCIF-dev and ACL60/60-dev. On MCIF-dev, EGTA-RG improves BLEU by +1.05/+0.59, XCOMET-XL by +0.019/+0.006, named-entity recall by +79\%/+73\% relative, and acronym recall by +0.099/+0.171 on En$\rightarrow$Zh and En$\rightarrow$De. Across MCIF-dev latency settings, EGTA consistently improves XCOMET-XL, named-entity recall, and acronym recall. External validation on ACL60/60-dev further shows consistent terminology-recall gains without additional fine-tuning. Shuffled-memory controls and activation audits provide evidence that the improvements are tied to paper-specific evidence alignment rather than generic context prompting.

Problem

• simultaneous speech translation
• extra context
• terminology adaptation
• technical talks
• context utilization

Innovation

• terminology adaptation
• simultaneous speech translation
• evidence-grounded context
• streaming translation
• document-level memory