Inspiration
I wanted to create an automated tool to help folks in business/school with their oral presentations, as public speaking is a pervasive fear even among professionals. Due to lack of availability a good speaking coach may not always be at hand, so this automated tool can provide a easy and fast way to improve one's public speaking.
What it does
After some calibration, the user practices their speech/report/etc. in front of the Kinect, which tracks the user's joints and facial positions to determine when the user starts looking away from the audience for too long, loses their posture, etc. Afterwards, the user can view how well they did and what they need to improve on.
How I built it
Visual Studio (C#) with Kinect libraries
Challenges I ran into
- Limitations of the Kinect (resorted to doing calculations with individual 3D points)
- Calibration (sometimes face is not properly tracked)
Accomplishments that I'm proud of
- Simple, yet feature-complete, solo hackathon project that I believe is the prototype for something that can reach a wide audience with one of their greatest fears
What I learned
- Various C# tricks, how to work with the Kinect, and the importance of making state diagrams of user scenarios to help track my development progress
What's next for SpeakerCoach
After fine-tuning the parameters/make better algorithms for tracking common public speaking flaws (in addition to adding some voice recognition API), this tool can really stand out as an on-the-fly automated coach.
Log in or sign up for Devpost to join the conversation.