University of Guelph
#
CIS*6650 Automatic Speech Recognition Fall 2012 Instructor: Fangju Wang##
Assignment 3 (20%)In this assignment, you are required to develop a spoken dialogue system (SDS) using Sphinx-4 as its speech recognizer.
An SDS is a computer system that interacts with its human user in a spoken natural language, usually on a turn-by-turn basis. An SDS interprets human speech and responds based on its interpretations. A major advantage of an SDS is that speech input and output enable the user to communicate with it in a more natural way, and allow it to be used when keyboard and screen input/output are impossible or inconvenient.
An SDS has a speech interpreter and a dialogue manager. They are responsible for recognizing and interpreting user input, and controlling system output respectively. If the system uses speech to respond, a text-to-speech (TTS) synthesizer is needed to generate spoken output. In the SDS developed for this assignment, system speech output is not required.
You need to choose an application field for your SDS. Possible fields include tourism, education, equipment trouble shooting, and so on.
Requirements:
-
The system should be able to answer at least ten questions, in arbitrary orders.
-
You are required to expanding the existing acoustic model and language model to include some new words, for example, names of cities, restaurants, stores, products, and so on.
-
Write a brief report of your system, including its application, system architecture, the expansion of the acoustic model and language model, and whatever you consider important.
-
Submit your source code (not including the Sphinx-4 code) and your report.
-
Demo your system to the instructor.
###Due time: 24:00pm, Dec 3, 2012, (Monday)
###Demo time: Dec 4 and 5, 2012, (Tuesday and Wednesday)