YouTube Voice Cloner

Transform YouTube and TikTok videos into custom AI-generated speech using Dia TTS voice synthesis technology.

What does it do?

This application takes a YouTube or TikTok video URL and creates a new audio file with the same content but spoken in a different AI-generated voice. Here's how it works:

1. Download: Extracts audio from YouTube/TikTok videos
2. Analyze: Processes the original audio for voice characteristics
3. Synthesize: Generates new speech using Dia TTS voice cloning technology

Features

• ✅ Multi-platform support: YouTube and TikTok videos
• ✅ Voice cloning: Creates custom voice models from source audio
• ✅ Web interface: Simple, responsive UI built with HTMX
• ✅ Real-time processing: See progress as your audio is generated
• ✅ Audio player: Listen to results directly in the browser
• ✅ Docker deployment: Ready for cloud deployment (Google Cloud Run)

Requirements

System Dependencies

1. yt-dlp - Video/audio downloader

   # macOS
   brew install yt-dlp
   
   # Ubuntu/Debian
   sudo apt install yt-dlp
   
   # Or via pip
   pip install yt-dlp

2. TailwindCSS - For building styles

   npm install -g tailwindcss

3. TEMPL - For templating in go

   go install github.com/a-h/templ/cmd/templ@latest

Environment Variables

Create a .env file with:

PORT=8080
FAL_KEY=your_fal_ai_api_key_here
DATABASE_URL=your_postgres_connection_string

Quick Start

1. Clone the repository

   git clone https://github.com/henrik392/youtube-voice-go.git
   cd youtube-voice-go

2. Install dependencies

   go mod download

3. Build and run

   make build
   make run

4. Open your browser Navigate to http://localhost:8080

Development Commands

# Build the application (generates templates + CSS + binary)
make build

# Run the application
make run

# Start with live reload (installs air if needed)
make watch

# Run tests
make test

# Start PostgreSQL database container
make docker-run

# Stop database container
make docker-down

# Clean build artifacts
make clean

Video Limitations

• Length: 30 seconds to 10 minutes
• Optimal: 1-5 minutes with clear audio
• Format: Supports any format that yt-dlp can process

Architecture

cmd/
├── api/           # Main application entry point
└── web/           # Web handlers and templates
internal/
├── database/      # PostgreSQL integration
├── elevenlabs/    # Voice synthesis API client
├── server/        # HTTP server setup
└── youtube/       # Video processing logic

Technology Stack

• Backend: Go with Chi router
• Frontend: HTML templates (templ) + HTMX + TailwindCSS
• Database: PostgreSQL
• Audio Processing: yt-dlp + ffmpeg
• AI Voice: Dia TTS (fal.ai) API

Deployment

Docker

make docker-build
docker run -p 8080:8080 yt-voice

Google Cloud Run

gcloud run deploy --image=europe-north1-docker.pkg.dev/youtube-to-voice/youtube-to-voice-repo/youtube-to-voice-image:tag1

How It Works

1. URL Validation: Checks if the provided URL is from YouTube or TikTok
2. Audio Extraction: Downloads and converts video to MP3 (max 3 minutes)
3. Reference Processing: Prepares the original audio as reference for voice cloning
4. Text Processing: Formats the target text for Dia TTS processing
5. Voice Synthesis: Uses Dia TTS to generate speech with the cloned voice in one step
6. Delivery: Serves the final audio file through the web interface

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run tests: make test
5. Submit a pull request

License

This project is for educational and personal use. Please respect content creators' rights and fal.ai's terms of service.