SilentBridge

Bridging Silence with AI-Powered Sign Language Translation.

Created on 8th March 2025

•

Problem:

466M+ deaf/mute individuals globally face communication barriers due to language gaps (e.g., ISL vs. ASL, regional variations).
Existing tools lack bidirectional communication (sign ⇔ text/speech) and multilingual support (e.g., Hindi, Marathi).

Solution:
SilentBridge is a web app that:

Teaches Sign Language: Interactive lessons for ISL, ASL, and BSL with regional language support.
Translates in Real-Time:
- Sign-to-Text/Speech: Converts gestures to text (e.g., ISL → Hindi) and spoken language.
- Text-to-Sign: Converts typed text to sign language animations.
Supports Regional Languages: Translates to 10+ Indian languages (Hindi, Marathi, Tamil, etc.).

Impact:

Education: Helps deaf students learn sign language and communicate with teachers.
Healthcare: Enables deaf patients to explain symptoms to doctors.
Public Services: Assists deaf citizens in banks, government offices, and businesses.

Gesture Variability:
- Different users perform gestures differently, making it hard to train a robust model.
- Solution: Collected a diverse dataset with multiple angles and lighting conditions.
Real-Time Performance:
- Ensuring smooth performance at 30 FPS was challenging.
- Solution: Used lightweight models (MediaPipe, TensorFlow.js) and optimized code.
Multilingual Support:
- Integrating multiple APIs (Google Translate, Web Speech) while maintaining simplicity.
- Solution: Used modular code and pre-built libraries for seamless integration.

Technologies used

Firebase

Google Translate API

Tailwind CSS

React.js

Tensorflow.Js

Web Speech Api

Teachable machine

Google Mediapipe