Build the translation path
Created Flask services in Python for grammar parsing, synonym matching, and fallback mapping when an exact vocabulary match is unavailable.
CASE STUDY · 2025
Silent Bridge.
Turn spoken language into a sign language conversation.
A cross platform speech and text translator for Indian Sign Language, pairing a Flask language pipeline with a custom animated 3D avatar.
01THE PROJECT
OVERVIEW
What it is
and who it helps.
Silent Bridge is a cross platform app that translates speech and text into Indian Sign Language through a custom 3D avatar. It brings together mobile product work, natural language processing, and a hand built animation library so a sentence can become a sequence of signs rather than a string of isolated words.
50+Custom Blender animations
~200 msReported response time
202611066813 APublished Indian application
THE CHALLENGE
Spoken English and Hindi do not map cleanly word for word to Indian Sign Language. The system had to interpret input, account for grammar, map terms to available signs, and keep the avatar responsive while rendering a sequence of detailed hand and facial animations.
HOW I APPROACHED IT
From the first
interaction to the edge cases.
Give the output a face and hands
Modeled and animated a custom avatar in Blender, creating more than 50 ISL animations for use in the cross platform Flutter app.
Keep voice input responsive
Integrated Firebase ML voice handling and optimized the translation and animation path to around 200 ms response time.
OUTCOME
A sentence becomes a sequence of signs.
The work combined mobile, NLP, and 3D animation into a single accessibility focused prototype. The associated Indian patent application was published on 17 July 2026 under the title “Sign Language Communication and Personalized Training Device.”
02TOOLS USED
BUILT WITH