Learning to Recognize Speech or Handwriting


Sponsor Prof. Chris Atkeson
cga@cc.gatech.edu
Area Intelligent Systems

Problem
It has become clear that in both speech and handwriting recognition some form of adaptation to a specific user improves performance. The Newton PDA learns to improve its handwriting recognition, and most speech recognition systems have a training phase. In this project you will explore learning to recognize either speech or handwriting.

Your goal is to create an expert program that recognizes your speech or handwriting. It can use templates, rules, or any other representation you prefer. The first step in this project is to figure out how to get data to work with, and how to process that data. I have systems for acquiring speech and handwriting, and some example programs that process that information. You can also surf the net to find other free software to assist you, including software for machine learning. The performance level and the time you put in on this project are up to you.