A week or so ago, Nyle DiMarco, the model/actor/deaf activist, visited my school and enlightened our students about how his experience as a deaf person was in shows like Dancing with the Stars (where he danced without hearing the music) and America's Next Top Model. Many people on those sets and in the cast could not use American Sign Language (ASL) to communicate with him, and so he had no idea what was going on at times.
What it does
SpeakAR listens to speech around the user and translates it into a language the user can understand. Especially for deaf users, the visual ASL translations are helpful because that is often their native language.
How we built it
We utilized Unity's built-in dictation recognizer API to convert spoken word into text. Then, using a combination of gifs and 3D-modeled hands, we translated that text into ASL. The scripts were written in C#.
Challenges we ran into
The Microsoft HoloLens requires very specific development tools, and because we had never used the laptops we brought to develop for it before, we had to start setting everything up from scratch. One of our laptops could not "Build" properly at all, so all four of us were stuck using only one laptop. Our original idea of implementing facial recognition could not work due to technical challenges, so we actually had to start over with a completely new idea at 5:30PM on Saturday night. The time constraint as well as the technical restrictions on Unity3D reduced the number of features/quality of features we could include in the app.
Accomplishments that we're proud of
This was our first time developing with the HoloLens at a hackathon. We were completely new to using GIFs in Unity and displaying it in the Hololens. We are proud of overcoming our technical challenges and adapting the idea to suit the technology.
What we learned
Make sure you have a laptop (with the proper software installed) that's compatible with the device you want to build for.
What's next for SpeakAR
In the future, we hope to implement reverse communication to enable those fluent in ASL to sign in front of the HoloLens and automatically translate their movements into speech. We also want to include foreign language translation so it provides more value to deaf and hearing users. The overall quality and speed of translations can also be improved greatly.