Inspiration

Dumb people cannot speak an oral language or have some degree of speaking ability but prefer not to talk because atypical voices often draw negative or unwanted attention. Using sign language, these individuals communicate. But not everyone knows ASL who will understand them. Due to COVID-19, if a Dumb person somehow tried to speak, they can't because they're wearing masks. So, what's the resolution?

What it does

Speakable, where we tried to create a user-friendly web application in which a Dumb person can record their sign language hand gestures, name them and it will convert them to Sound.

How we built it

Frontend:

  • JavaScript
  • HTML/CSS

Backend:

  • Python

Frameworks:

  • TensorFlow and TensorFlow.JS
  • Open CV

Challenges we ran into

To implement model training on client-side and choosing which Deep Learning algorithm to be used. Also, how to convert live text to speech.

Accomplishments that we're proud of

We're able to train the model on the client-side with a good accuracy without losing frame-rate.

What we learned

We learned how to shift TensorFlow from Python to JavaScript. There are soo many use cases!

What's next for Speakable

To improve the UI/UX by shifting the code base to React.JS. Also, create a Profile Section for each user and able to store their trained gestures and not training again on Cloud. Lastly, add more voice options with live translation!

Built With

Share this project:

Updates