Simultaneous Interpretation Corpus Construction by Large Language Models in Distant Language Pair
This addresses the problem of limited SI corpora for researchers and developers in machine translation, enabling more effective low-latency translation systems, though it is incremental as it builds on existing data conversion methods.
The paper tackles the challenge of constructing simultaneous interpretation (SI) corpora for training Simultaneous Machine Translation (SiMT) systems by proposing a method using Large Language Models to convert existing speech translation data into interpretation-style data, resulting in reduced latencies while maintaining quality comparable to offline-trained models.
In Simultaneous Machine Translation (SiMT) systems, training with a simultaneous interpretation (SI) corpus is an effective method for achieving high-quality yet low-latency systems. However, it is very challenging to curate such a corpus due to limitations in the abilities of annotators, and hence, existing SI corpora are limited. Therefore, we propose a method to convert existing speech translation corpora into interpretation-style data, maintaining the original word order and preserving the entire source content using Large Language Models (LLM-SI-Corpus). We demonstrate that fine-tuning SiMT models in text-to-text and speech-to-text settings with the LLM-SI-Corpus reduces latencies while maintaining the same level of quality as the models trained with offline datasets. The LLM-SI-Corpus is available at \url{https://github.com/yusuke1997/LLM-SI-Corpus}.