A spoken-based question answering system for train route service using the frame-based approach

A frame-based approach is a simple yet highly effective for a Question Answering (QA) system. In this research, a prototype of speech based QA system that provides information about train route service and departure schedule is developed based on that technique. There are three main components of the system, the automatic speech recognizer (ASR) that transcribes user speech input into sequence of words query, a QA system that translates the transcribed text into a query and search for relevant answer, and the automatic speech synthesizer (ASS) that output the relevant answer using speech. The ASR and ASS were developed based-on the hidden markov model for its acoustic model, and the n-gram model for its language model, while some techniques such as frame-based, keyword spotting, data fetching, and canned answer were employed in the QA system.