🎙️ Lazy Speech To Text Converter

Forked from webrtc-speech-to-text, with a Vue 3 frontend, multi-vendor transcription (Whisper, Google, Azure, Baidu, Xunfei), audio recording, and file management.

Transform speech to text effortlessly with WebRTC and AI

---

✨ Features

Feature	Description
🎤 Real-time Streaming	WebRTC-based audio capture with low latency
🌍 99+ Languages	Powered by Whisper AI - works offline
🔒 Privacy First	Local processing - your audio never leaves your machine
📱 Cross-platform	Works on Chrome, Firefox, and Safari
🎛️ Flexible Options	Record only, transcribe only, or both
📊 Visual Feedback	Real-time audio waveform visualization
🔐 User Authentication	Simple login system for access control

---

🚀 Quick Start

# 1. Install Whisper (one-time)
pip install whisper-ctranslate2

# 2. Build the project
make

# 3. Run the server
./webrtc-transcriber

# 4. Open browser
open http://localhost:9070

That's it! No cloud accounts, no API keys, no configuration needed.

See Runbook for prerequisites, troubleshooting, and alternative vendors.

---

🏗️ Architecture

┌─────────────┐     WebRTC      ┌─────────────────┐     Audio      ┌──────────────┐
│   Browser   │ ◄─────────────► │  Go Server      │ ─────────────► │  Whisper AI  │
│  (Vue 3)    │                 │  (Pion WebRTC)  │                │  (or Cloud)  │
└─────────────┘                 └─────────────────┘                └──────────────┘
       │                               │                                  │
       │    DataChannel               │                                  │
       ◄──────────────────────────────┼──────────────────────────────────┘
              (Transcription Results)

Layer	Technology
Backend	Go + Pion WebRTC + Opus
Frontend	Vue 3 + TypeScript + Vite + Tailwind CSS
Transcription	Whisper (default), Google, Azure, Baidu, Xunfei

See Architecture for component diagrams, call chains, and module dependencies.

---

⚙️ Usage

./webrtc-transcriber [options]

  --vendor string     whisper | google | azure | baidu | xunfei | recorder (default "whisper")
  --model string      tiny | base | small | medium | large (default "small")
  --language string   en | zh | ja | auto | ... (default "auto")
  --output string     Output directory (default "recordings")
  --http.port string  HTTP port (default "9070")

Copy env.example to .env for cloud vendor credentials and account configuration.

See Data & API for full HTTP API contracts and Conventions for configuration details.

---

📚 Documentation

The doc/ directory contains a comprehensive Project Knowledge Base built with Sphinx + MyST Markdown.

cd doc/
pip install -r requirements.txt   # One-time setup
make html                          # Build HTML documentation
make serve                         # Build with live reload

Document	Description
Project Overview	Purpose, boundaries, tech stack, deployment model
Repository Map	Directory structure, entry points, naming conventions
Architecture	Component diagram, call chains, module dependencies
Workflows	Real-time transcription, file management, vendor selection
Data & API	HTTP API contracts, DataChannel messages, Go interfaces
Conventions	Code style, error handling, config management
Runbook	Build, run, debug, and troubleshoot
Testing	Test strategy, coverage targets, critical path checklist
AI Guide	Orientation guide for AI assistants

Vendor Setup Guides

• Whisper Setup · Azure Setup · Baidu Setup · Xunfei Setup · Recorder Setup

---

⚠️ Disclaimer

This project is a proof of concept and should not be deployed in production without implementing proper security measures.

📄 License

MIT - see LICENSE for details.

---

Made with ❤️ by Walter Fan