Robust Speech-to-Intent Systems for Conversational Agents KU Leuven
Conversational agents with spoken interaction are composed of multiple modules: a speech recognizer (speech-to-text), a natural language processor (text-to-semantics), a dialog manager (generator of semantic representation next dialog turn), natural language generation (semantics-to-text) and speech synthesis (text-to-speech). The goal of the thesis is to eliminate the text representation and map speech directly to intent with minimal amounts ...