Speech-Enabled Hybrid Multilingual Translation for Mobile Devices
Krasimir Angelov, Björn Bringert, Aarne Ranta · 2014
This paper presents an architecture and aprototype for speech-to-speech translationon Android devices, based on GF (Grammatical Framework). From the user’spoint of view, the advantage is that thesystem works off-line and yet has a leansize; it also gives, as a bonus, grammatical information useful for languagelearners. From the developer’s point ofview, the advantage is the open architecture that permits the customization of thesystem to new languages and for specialpurposes. Thus the architecture can beused for controlled-language-like translators that deliver very high quality, whichis the traditional strength of GF. However,this paper focuses on a general-purposesystem that allows arbitrary input. It covers eight languages.