Splitting Ill-formed Input for Robust Multi-lingual Speech Translation

摘要

This paper proposes a method for robust speech translation by splitting spoken-language input which may include many ill-formed expressions. The proposed method splits input into well-balanced translation units based on a score for a substructure. The complete transla- tion result is formed by concatenating the partial translation results of each split unit. The splitting is performed during left-to-right parsing, and does not degrade translation efficiency.