Redefining Machine Simultaneous Interpretation: From Incremental Translation to Human-Like Strategies

TL;DR

This paper introduces adaptive actions in simultaneous machine translation using large language models, enabling real-time restructuring and omission to improve translation quality and reduce delay, bringing machine interpretation closer to human strategies.

Contribution

It extends traditional SiMT policies with new adaptive actions and develops a latency-aware evaluation pipeline, demonstrating improved performance on multiple language benchmarks.

Findings

Enhanced semantic translation metrics

Lower latency compared to baselines

Improved fluency and fidelity balance

Abstract

Simultaneous Machine Translation (SiMT) requires high-quality translations under strict real-time constraints, which traditional policies with only READ/WRITE actions cannot fully address. We extend the action space of SiMT with four adaptive actions: Sentence_Cut, Drop, Partial_Summarization and Pronominalization, which enable real-time restructuring, omission, and simplification while preserving semantic fidelity. We adapt these actions in a large language model (LLM) framework and construct training references through action-aware prompting. To evaluate both quality and word-level monotonicity, we further develop a latency-aware TTS pipeline that maps textual outputs to speech with realistic timing. Experiments on the ACL60/60 English-Chinese, English-German and English-Japanese benchmarks show that our framework consistently improves semantic metrics and achieves lower delay compared to reference translations and salami-based baselines. Notably, combining Drop and Sentence_Cut leads to consistent improvements in the balance between fluency and latency. These results demonstrate that enriching the action space of LLM-based SiMT provides a promising direction for bridging the gap between human and machine interpretation.