Assignment 2:
- Use a camera and microphone to determine when someone is
moving towards the door and quickly speak a loud
warning before the door opens outwards and crushes someone
walking the corridors outside. Since the simple systems that you
built for the last assignment are not very accurate can you ask
the user a quick yes/no query and recognize yes/no
answers. You need not use festival for generating speech but
have pregenerated speech/questions ready to go through the
sound-card. You may consider using multiple cameras and multiple
machines to speed up the system.
You are to improve the systems from your previous assignment and
will need to
- Use a camera (or cameras) for motion tracking
- Get user-input from a microphone to discern intent
- Speed up the system using pre-canned speech-sound files or
multiple machines or both.
The assignment objectives are to:
- Overcome limitations of your motion tracker.
- Learn how to recognize simple speech input from Sphinx.
- Use festival to generate sound files for pre-generated speech
- Use "play" to play sound files
- ...
I will ask you to come up with test scenarios. It would be good if
each of your systems had different strengths and weaknesses.
Turning in the assignment
Please turn in the assignment by demoing your working project to
me. Also turn in your code and instructions on installing,
compiling, and running your program via email. You will demo
your program during class time on March 10.
Good Luck!
Sushil Louis
Last modified: Wed Feb 25 09:57:04 PST 2004