Optimizing Segmentation Strategies for Simultaneous Speech Translation
Yusuke Oda, Graham Neubig, Sakriani Sakti, Tomoki Toda, Satoshi Nakamura · 2014
In this paper, we propose new algorithms for learning segmentation strategies for si-multaneous speech translation. In contrast to previously proposed heuristic methods, our method finds a segmentation that di-rectly maximizes the performance of the machine translation system. We describe two methods based on greedy search and dynamic programming that search for the optimal segmentation strategy. An experi-mental evaluation finds that our algorithm is able to segment the input two to three times more frequently than conventional methods in terms of number of words, while maintaining the same score of auto-matic evaluation.1 1