Inspiration
Communication should never be a barrier. Uni-voice was inspired by the everyday challenges faced by deaf and hard-of-hearing individuals during live conversations, lectures, and meetings. While many speech tools exist, few are designed with accessibility as the primary focus or provide a seamless, real-time experience. We wanted to build a simple yet powerful tool that promotes inclusive communication and gives users confidence in spoken environments.
What it does
Uni-voice is an accessibility-first web application that enables real-time voice-to-text and text-to-voice communication. It allows hearing-impaired users to instantly read spoken conversations and respond using synthesized voice output. Users can securely log in and access their transcription history, making the tool useful beyond one-time interactions.
How we built it
Uni-voice was built using a modern web stack with accessibility and performance in mind:
- Frontend: React with a clean, distraction-free interface and adjustable font sizes
- Speech Processing: Real-time speech-to-text and text-to-speech powered by AI models
- Authentication & Database: Supabase for secure user authentication and transcript storage
- Deployment: Hosted on Vercel for fast and reliable access
Supabase plays a core role by managing user sessions and storing user-specific transcripts with proper data isolation.
Challenges we ran into
- Handling real-time audio streams while keeping transcription latency low
- Managing partial and final transcription updates without cluttering the UI
- Designing an interface that remains readable and intuitive for different accessibility needs
- Balancing feature scope within the hackathon timeline
Accomplishments that we're proud of
- Building a fully functional real-time transcription system
- Creating an accessibility-focused UI with adjustable font scaling
- Securely storing user-specific transcription history
- Successfully integrating authentication and backend services in a short timeframe
What we learned
- Accessibility should be considered from the start, not added later
- Real-time applications require careful state and resource management
- Backend platforms like Supabase significantly speed up development without sacrificing best practices
- Small, well-designed features can have meaningful real-world impact
What's next for Uni-voice
- Multilingual transcription and voice output
- Offline or low-bandwidth support
- More accessibility customization options
- Integration with educational and workplace platforms
Built with
- TypeScript
- React
- Supabase (Authentication & PostgreSQL Database)
- AI Speech APIs (Speech-to-Text & Text-to-Speech)
- Web Audio API
- Vercel
- HTML5 & CSS3
Built With
- gemini-api
- react
- supabase
- typescript
- vercel


Log in or sign up for Devpost to join the conversation.