Keynote speech 3: Toward simultaneous, natural and multimodal speech-to-speech translation

Spoken language technologies enable to support natural oral communication. Language barrier between different language speaking people is one of the biggest communication problems for human being. Speech-to-speech translation had been studied in order to break the language barrier. In this talk, I would like to introduce brief history of the speech-to-speech translation research and state-of-the-art speech translation system. Then I would introduce new challenges toward a simultaneous speech-to-speech translation including simultaneous incremental machine translation, joint-optimization of ASR and MT modules, Speech-to-speech translation preserving para/non-linguistic information and multimodal speech translation with a talking face.