π Inspiration
The inspiration behind Translato AI came from real-world communication barriers β travelers lost in translation, students struggling with foreign study material, and global teams miscommunicating due to language gaps. While translation tools exist, most lack multimodal support and real-time accessibility in one place. I envisioned a unified, AI-powered solution that could seamlessly translate text, images, and voice with a friendly UX.
π§ What it does
Translato AI is a full-featured web app that enables real-time, multilingual translation using AI across:
- π€ Text-to-text translation
- πΌοΈ Image-to-text-to-translation (OCR + LLM)
- ποΈ Voice-to-text-to-translation-to-speech
Users can:
- Input text or upload images/audio
- Choose source and target language
- Instantly receive translated text or voice
- Enjoy a clean UX with role-based usage limits (Free vs Pro)
- Access the app on any device via Netlify
It empowers users from all walks of life β travelers, remote workers, students β to communicate effortlessly in any language.
π οΈ How we built it
Built primarily using Bolt.new, with a Node.js/React frontend and powerful AI integrations:
π§° Tech Stack
- Frontend: React, TypeScript, Tailwind CSS, Vite
- Logic & Routing: Bolt.new
- AI Services:
- Google Gemini API for text translation & OCR
- Google Cloud Vision API for image-to-text
- ElevenLabs API for text-to-speech
- Auth & DB: Supabase (PostgreSQL, RLS, Triggers, Auth)
- UX Tools: React Dropzone, Framer Motion, Toasts, reCAPTCHA
- Security: DOMPurify, Validator.js, HTTPS enforcement, API key protection
- Hosting: Netlify (
https://translatoai.netlify.app) - Role Management: Role-based usage limits via Supabase (RevenueCat integration planned)
π§ Challenges we ran into
- Chaining multimodal APIs: Stitching audio β text β translation β voice in real-time while handling latency and async issues.
- Prompt tuning: Getting accurate translations from Gemini required precise prompt engineering and language-aware fallback flows.
- OCR inconsistencies: Image quality and handwriting produced edge cases that needed graceful fallbacks.
- API cost control: Implementing smart rate limiting and usage tracking for Free/Pro users without external billing SDKs yet.
- Voice integration: Managing audio formats (MP3/WAV), TTS generation delays, and safe streaming of results.
π Accomplishments that we're proud of
- Successfully delivered real-time, multimodal AI translation in a single unified platform
- Created a secure, production-ready UX using Supabaseβs RLS, reCAPTCHA, and frontend sanitization
- Used 3+ major AI APIs (Gemini, ElevenLabs, Vision) within Bolt.new β fully compliant with hackathon requirements
- Published on Netlify with live demo and Bolt badge
- Designed a scalable architecture, with future migration path to Spring Boot microservices already in mind
π What we learned
- Mastered the integration of large language models (LLMs) and multimodal AI in real-world apps
- Learned how to build production-ready projects on Bolt.new, including routing, state management, and API orchestration
- Gained deeper insights into prompt engineering, rate limiting, and error fallback design
- Understood the nuances of user experience across audio/image/text input flows
- Strengthened experience with security (XSS prevention, file validation, API key safety)
π What's next for Translato Ai: language translator web app
- π Integrate RevenueCat to unlock subscription billing & dynamic plan upgrades
- π± Mobile app with offline translation mode
- π¬ Real-time multilingual chat between users
- π§Ύ Exportable transcripts of translated voice sessions
- π Admin dashboard to monitor usage and API costs
- π Auto language detection and smarter prompt templating
- π― Launch Translato AI as a public SaaS with free & premium tiers
Translato AI is more than a hackathon project β it's a global communication tool powered by AI, ready to grow.
Built With
- elevenlabs
- gemini
- geminiapi
- javascript
- netlify
- npm
- react
- router
- supabase
- tailwind
- typescript
- vite
Log in or sign up for Devpost to join the conversation.