Voice-to-Text Electron App 🎤

Because typing is so 2023

What is this?

A dead-simple Electron app that turns your voice into text. Press a button, speak, and watch the magic happen.

# Run it
npm run electron:dev

# Ship it
npm run electron:dist

The Stack

Electron - Desktop wrapper that doesn't suck
Next.js 15 - React with the good stuff
shadcn/ui - Components that actually look good
Web Speech API - Built-in browser magic (works offline)
Deepgram - Premium speech recognition (optional)

Features

🎙️ One-click voice recording
📝 Real-time transcription
🔄 Dual speech engines (Web Speech + Deepgram)
🖥️ Native desktop experience
📱 Mobile-first responsive (because why not)

Why Electron?

Look, I know Electron gets hate, but sometimes you just need a desktop app that works everywhere. This thing boots fast, uses reasonable memory, and doesn't feel like a web page pretending to be an app.

The Good Stuff

// Automatic fallback between speech engines
const activeService = useWebSpeech || !DEEPGRAM_API_KEY 
  ? webSpeech 
  : deepgram;

// Press mic → speak → profit
<Button onClick={activeService.startListening}>
  <Mic />
</Button>

Performance Notes

Web Speech API works offline (no API calls)
Deepgram for production-grade accuracy
TypeScript because we're not animals
Proper error handling (shocking, I know)

Deployment

Cross-platform builds with electron-builder:

# macOS
npm run electron:dist

# Windows (from macOS)
npm run electron:dist -- --win

# Linux
npm run electron:dist -- --linux

The Real Talk

This took like 2 hours to build. Modern tooling is wild. The hardest part was dealing with pnpm workspaces and Electron playing nice together.

Voice recognition in 2025 is basically free and works great. Why are we still typing everything?

Built with ☕ and too much caffeine

Source Code | Issues