SignWave — Breaking Communication Barriers with AI

Inspiration

Communication is something many people take for granted, but for individuals who rely on American Sign Language (ASL), interacting with people who do not understand sign language can present challenges.

Our team wanted to explore how artificial intelligence could help bridge this communication gap. We envisioned SignWave as an assistive communication tool that recognizes supported hand signs, converts them into text, improves the readability of that text, and generates natural-sounding speech.

Our goal was to make communication more accessible while exploring how different AI technologies could work together to solve a real-world problem.

What It Does

SignWave is a sign-to-speech prototype designed to turn recognized hand gestures into spoken communication.

The system is designed around four main components:

  1. Sign Recognition: Computer vision recognizes a limited set of supported hand signs and converts them into text.
  2. AI Text Processing: Google Gemini helps clean up recognized text into more natural English while preserving the intended meaning.
  3. Language Translation: The translation component is designed to convert finalized text into other languages.
  4. Text-to-Speech: ElevenLabs converts the processed text into realistic, natural-sounding audio.

The long-term vision is to allow a person who communicates through sign language to interact more easily with someone who does not understand it, including during video calls.

How We Built It

We divided our team into two groups to develop different parts of the system simultaneously.

Group A — Sign Recognition

Group A focused on the computer vision side of SignWave, using Python and exploring MediaPipe to identify supported hand gestures and produce text.

Group B — AI Language Processing and Speech

Group B focused on transforming recognized text into spoken communication.

We used:

  • Python to build and connect the different components.
  • ElevenLabs API to generate natural-sounding speech from text.
  • Google Gemini API to process and improve text

Built With

  • elevenlabs
Share this project:

Updates

Submission history