A Next.js web application that helps you communicate when you can't use your voice. Type text and let AI speak for you using Google's Gemini Text-to-Speech API.
Perfect for situations like laryngitis, voice rest, or any time you need to communicate without speaking. I initially vibe-coded this quickly on Google AI Studio, then tweaked it a bit more. The voice is meant to sound vaguely like my own (not very well, but still), but you can choose lots of other voices or change the prompt as needed (different voices may need different prompting) to make it sound a bit more like you. Try it at https://vocalrest.vercel.app (but please don't abuse it, the default API quota for TTS is quite small)!
- Text-to-Speech: Convert typed text to natural-sounding speech using Gemini TTS
- Multiple Voice Options: Choose from 7 different voices (Orus, Umbriel, Algenib, Sadaltager, Charon, Fenrir, Puck) with varying tones and styles
- Quick Phrases: One-click buttons for common phrases like "Thank you", "Yes, please", "No, thanks", and more
- Speech History: View and replay your recent spoken phrases (last 30 items)
- Audio Caching: Automatically caches generated audio in IndexedDB for instant playback of repeated phrases
- Dark Mode UI: Modern, accessible dark theme interface
- Keyboard Shortcuts: Press Enter to speak (Shift+Enter for new line)
Prerequisites: Node.js
-
Install dependencies:
npm install
-
Create a
.env.localfile in the root directory and set your Gemini API key:GEMINI_API_KEY=your_gemini_api_key_here -
Run the development server:
npm run dev
-
Open http://localhost:3000 in your browser.
npm run build
npm startThis will build the app and run it in production mode on http://localhost:3000. Use this to test the production build locally before deploying.
- Next.js 15 - React framework
- React 19 - UI library
- TypeScript - Type safety
- Tailwind CSS - Styling
- Google Gemini API - Text-to-speech generation
- IndexedDB - Client-side audio caching
- Web Audio API - Audio playback
MIT