coral

• 

  render page

• 

  timeline editor

• 

  dashboard

• 

  upgrade page

• 

  Landing page

Inspiration

India's creators speak in Hinglish, Hindi, and many regional languages, but most captioning tools are primarily designed for English. When creators mix Hindi and English in the same sentence, getting accurate captions becomes frustrating. We built Coral to bridge this gap and make captioning easier for Indian creators.

What it does

Coral is an AI-powered captioning platform built for multilingual video creators. It transcribes speech, supports Hinglish captions, intelligently segments text, and transforms captions into engaging animations. Creators can customize their captions and export finished videos in one workflow.

How we built it

We built Coral using Google AI Studio and Gemini for AI-powered transcription and caption generation. We use FFmpeg for audio extraction and video processing, structured AI-generated data for caption segmentation, and an HTML5 Canvas rendering pipeline to create animated captions and export video

Challenges we ran into

One of our biggest challenges was handling Hinglish, where speakers naturally switch between Hindi and English within the same sentence. Accurate transcription alone wasn't enough. Captions also needed proper segmentation, synchronization, and readable timing. We also faced challenges with video rendering performance, processing large files, and making exports reliable across devices.

Accomplishments that we're proud of

We learned that building an AI product involves much more than connecting an AI model to an interface. Accuracy, latency, synchronization, rendering performance, affordability, and user experience all matter. We also learned the importance of building around the real language patterns of the people we're serving.

Built With

• aistudio
• node.js
• vite