Labrish

Where every voice has a story, and every story finds its voice.

Comment

Inspiration

Labrish was born from a deep appreciation for the Caribbean’s rich oral traditions and the unique energy of island conversation. Growing up surrounded by animated storytelling, expressive dialects, and vibrant community “labrish”—the Jamaican term for lively chit-chat—we saw an opportunity to blend this cultural heritage with modern technology. Our goal was to create a digital gathering place where anyone, anywhere, could experience the warmth, humor, and authenticity of Caribbean voices, while also making storytelling accessible and interactive for a global audience.

What it does

Labrish is a web app that empowers users to chat, narrate, and listen to stories using lifelike AI-generated voices: • Real-Time AI Voice Chat: Connect and converse with others using expressive AI voices, with support for Caribbean accents and global languages. • Storytelling Studio: Write your own stories and have them narrated by AI, or train the app to use your own voice for a personal touch. • Cultural Preservation: Celebrate and share Caribbean oral traditions, making them interactive and accessible for a worldwide audience. • Personalization: Choose from a library of voices or create your own, ensuring every story feels unique and authentic. • Seamless Integration: Built on ElevenLabs’ powerful API, Labrish delivers high-quality, real-time audio experiences across devices.

How we built it

• Tech Stack: Labrish is a modern web app built with a mobile-first, responsive design philosophy. We leveraged ElevenLabs’ API for text-to-speech and voice cloning, using both REST and WebSocket streaming for real-time chat and narration. • Voice Integration: We experimented with different ElevenLabs models (Turbo v2 for speed, Multilingual v2 for language variety), customizing voice characteristics for stability, emotion, and similarity to real Caribbean accents. • Storytelling Studio: Users can write stories, train their own AI voice, or select from a library of expressive Caribbean voices. We used SSML (Speech Synthesis Markup Language) to add pauses, emphasis, and natural phrasing to narrated stories. • Frontend Design: Inspired by 2025 design trends, we incorporated bold, multi-tonal Caribbean color palettes, expressive typography, and playful animations. The UI features bento box layouts, custom illustrations, and both dark/light modes for accessibility. • Real-Time Features: We implemented WebRTC and ElevenLabs’ streaming API for low-latency, real-time audio conversations, ensuring users could chat and share stories seamlessly. • Performance & Scalability: Caching strategies and chunked audio streaming were used to optimize performance and minimize API calls, especially for high-volume or enterprise scenarios.

Challenges we ran into

• Voice Authenticity: Achieving natural-sounding Caribbean accents with AI voices required extensive tuning. We had to experiment with different models, voice settings, and SSML tags to avoid monotony or mispronunciation. • Latency and Streaming: Real-time audio streaming introduced latency and connection management challenges. Keeping WebSocket connections open and optimizing chunk sizes helped reduce delays, but required careful backend engineering. • API Limitations and Errors: We encountered rate limits, occasional audio glitches, and format mismatches. Robust error handling and retry logic were essential for a smooth user experience. • Design Balance: Merging bold Caribbean visuals with a clean, modern interface for global users was a creative challenge. We iterated on layouts and color schemes to achieve both authenticity and universal appeal. • Scalability: Supporting high volumes of real-time audio interactions required thoughtful caching, queueing, and monitoring strategies to maintain performance and reliability.

Accomplishments that we're proud of

• Authentic Voice Experience:

Successfully created expressive, natural-sounding Caribbean AI voices that resonate with both regional and global users. • Seamless Real-Time Features: Integrated ElevenLabs’ API for real-time chat and storytelling, delivering fast, interactive audio experiences. • Cultural Impact: Built a platform that celebrates and preserves Caribbean oral traditions, making them accessible to a new generation of storytellers worldwide. • User-Centered Design: Achieved a balance between bold, culturally authentic visuals and intuitive, modern usability. • Scalable Architecture: Engineered Labrish to handle high user volumes and real-time communication, ensuring reliability and performance at scale. • Personalization and Accessibility: Enabled users to train their own voices and choose from a diverse library, making storytelling more personal and inclusive.

What we learned

Building Labrish taught us the power and complexity of voice technology. We explored the nuances of Caribbean dialects and the technical challenges of making AI voices sound authentic and expressive. Integrating the ElevenLabs API opened our eyes to the possibilities of real-time, multilingual audio synthesis, as well as the importance of fine-tuning voice settings for natural storytelling. We also gained valuable insights into user experience design, balancing bold Caribbean aesthetics with modern, global usability.

What's next for Labrish

Labrish is a celebration of Caribbean storytelling, powered by TTS API AI. The project pushed us to innovate at the intersection of culture and technology, overcoming challenges in voice synthesis, real-time communication, and design. We’re proud to offer a platform where every voice, Caribbean or global, can share its story and be heard. Labrish is more than just an app, it’s a celebration of culture, technology, and the universal power of storytelling.

Built With

Share this project:

Updates