Skip to content

YOUTUBE_AUTOMATION_SETUP.md

Latest commit

 

History

History
303 lines (229 loc) · 8.31 KB

YOUTUBE_AUTOMATION_SETUP.md

File metadata and controls

303 lines (229 loc) · 8.31 KB
Raw

YouTube Channel Automation - Complete Setup Guide

🎯 What This Replicates

This replicates the N8N workflow from the video "How I Automate AI Faceless Channels with N8N" that:

  • Finds viral YouTube videos automatically
  • Generates 3-act story scripts using AI
  • Creates b-roll images
  • Generates voiceovers
  • Renders complete videos
  • Tracks everything in a database

Cost per video: Under $0.30 (as shown in the video)

📋 Required API Keys & Services

⭐ Required:

  1. OpenRouter API Key (REQUIRED)
    • Purpose: AI script generation (3-act stories)
    • Get it: https://openrouter.ai/keys
    • Cost: ~$0.01-0.05 per script
    • Set as: OPENROUTER_API_KEY

🔧 Recommended:

  1. YouTube Data API Key (Recommended)

💡 Optional (Free Alternatives Available):

  1. ElevenLabs API Key (Optional - Free TTS available)

    • Purpose: High-quality voiceovers
    • Get it: https://elevenlabs.io/
    • Cost: Free tier available, then ~$0.01-0.05 per voiceover
    • Free Alternative: Use gTTS (Google Text-to-Speech) - completely free
    • Set as: ELEVENLABS_API_KEY

  2. Image Generation API (Optional - Free options available)

    • Options:
      • Hugging Face Inference API (Free tier)
      • Stability AI (Free tier available)
      • Replicate API (Pay-per-use)
    • Cost: Free tier or ~$0.01-0.10 per image
    • Set as: STABILITY_API_KEY or HUGGINGFACE_API_KEY

  3. Video Rendering (Free options available)

    • Options:
      • FFmpeg (Free, local) ⭐ Recommended
      • NCA Toolkit (Free, self-hosted)
      • RenderAPI (Paid, cloud)
    • Cost: Free if using FFmpeg locally
    • Set as: RENDER_API_KEY (only if using cloud rendering)

🚀 Quick Setup

Step 1: Install Dependencies

pip install aiohttp gTTS

For video rendering (FFmpeg):

Step 2: Configure API Keys

Option A: Environment Variables (Recommended)

# Windows PowerShell
$env:OPENROUTER_API_KEY="your-openrouter-key"
$env:YOUTUBE_API_KEY="your-youtube-key"  # Optional
$env:ELEVENLABS_API_KEY="your-elevenlabs-key"  # Optional

Option B: Config File

Create workflow_config.json:

{
  "api_keys": {
    "openrouter": "your-openrouter-key-here",
    "youtube": "your-youtube-api-key",
    "elevenlabs": "your-elevenlabs-key",
    "stability": "your-stability-key",
    "render_api": "your-render-api-key"
  },
  "workflows": {
    "storage_path": "./workflows/data"
  }
}

Step 3: Load Workflows

# Load main orchestrator
python -m workflow_engine.main load --workflow-id youtube_main --file workflow_engine/examples/youtube_automation/main_orchestrator.json

# Load individual workflows
python -m workflow_engine.main load --workflow-id find_viral --file workflow_engine/examples/youtube_automation/find_viral_videos.json
python -m workflow_engine.main load --workflow-id generate_script --file workflow_engine/examples/youtube_automation/generate_script.json
python -m workflow_engine.main load --workflow-id generate_images --file workflow_engine/examples/youtube_automation/generate_images.json
python -m workflow_engine.main load --workflow-id generate_voiceover --file workflow_engine/examples/youtube_automation/generate_voiceover.json

Step 4: Run Workflows

# Run main workflow (orchestrates everything)
python -m workflow_engine.main run --workflow-id youtube_main

# Or run individual workflows
python -m workflow_engine.main run --workflow-id find_viral --data '{"api_key": "your-youtube-key"}'

📊 Cost Breakdown (Per Video)

Based on the video's analysis:

Component Service Cost
Script Generation OpenRouter (GPT-4) $0.01-0.05
Images (6 images) Free APIs $0.00
Voiceover gTTS (free) $0.00
Video Rendering FFmpeg (local) $0.00
Total $0.01-0.05

With paid services:

  • ElevenLabs voiceover: +$0.01-0.05
  • Cloud rendering: +$0.10-0.50
  • Total with paid: Still under $0.30 per video

🎬 Workflow Process

  1. Find Viral Videos

    • Searches YouTube for trending videos
    • Filters by view count criteria (10k/1day, 50k/7days, 200k/30days)
    • Checks for duplicates
    • Saves to database

  2. Generate Script

    • Fetches video transcript
    • Uses AI to create 3-act story structure
    • Adds random character names
    • Saves script

  3. Generate Assets

    • Images: Creates 6 b-roll images (can run in parallel)
    • Voiceover: Converts script to audio
    • Both run simultaneously for speed

  4. Render Video

    • Combines images + voiceover
    • Creates final MP4 video
    • Updates status in database

🔧 Free Alternatives Setup

Free Text-to-Speech (gTTS)

The workflow supports free TTS automatically. No API key needed!

# Already integrated - just don't set ELEVENLABS_API_KEY
# The workflow will use free TTS automatically

Free Image Generation

Option 1: Hugging Face (Free)

  1. Get API key: https://huggingface.co/settings/tokens
  2. Set: HUGGINGFACE_API_KEY
  3. Update image node to use Hugging Face API

Option 2: Local Stable Diffusion

  • Run Stable Diffusion locally (free)
  • Update image node to call local API

Free Video Rendering (FFmpeg)

Setup FFmpeg:

  1. Install FFmpeg (see Step 1)
  2. Create a simple rendering script
  3. Update workflow to call FFmpeg instead of cloud API

Example FFmpeg command:

ffmpeg -loop 1 -i image1.jpg -i voiceover.mp3 -c:v libx264 -tune stillimage -c:a aac -b:a 192k -pix_fmt yuv420p -shortest output.mp4

📁 Database Structure

All data is stored in workflows/data/ as JSON files:

  • videos.json - Video tracking with status
  • scripts.json - Generated scripts
  • voiceovers.json - Voiceover metadata
  • images.json - Image metadata

Example video record:

{
  "id": "video_123",
  "video_id": "dQw4w9WgXcQ",
  "url": "https://youtube.com/watch?v=...",
  "status": "pending",
  "views": 150000,
  "created_at": "2024-01-01T12:00:00",
  "updated_at": "2024-01-01T12:05:00"
}

🎨 Customization

Adjust Viral Criteria

Edit find_viral_videos.json:

{
  "min_views_1day": 5000,    // Lower threshold
  "min_views_7day": 25000,
  "min_views_30day": 100000
}

Change AI Model

Use cheaper model in generate_script.json:

{
  "model": "openai/gpt-3.5-turbo"  // Cheaper than GPT-4
}

Add More Images

Edit generate_images.json to add more image nodes or increase count.

⚠️ Important Notes

  1. Image Generation: Currently returns descriptions. For actual images, integrate with:

    • Stability AI API
    • Hugging Face Inference API
    • Replicate API
    • Or run Stable Diffusion locally

  2. Video Rendering: Requires:

    • FFmpeg installed locally, OR
    • Cloud rendering API configured

  3. Database: Uses file-based JSON storage. For production, consider:

    • SQLite (simple)
    • PostgreSQL (scalable)
    • NoCoDB (as mentioned in video)

  4. YouTube Transcripts: May require:

    • YouTube Data API for some videos
    • Or use youtube-transcript-api Python library (free)

🆘 Troubleshooting

No API keys found:

  • Set environment variables or add to workflow_config.json
  • Check key names match exactly

Image generation not working:

  • Integrate actual image API (see notes above)
  • Or use local Stable Diffusion

Video rendering fails:

  • Install FFmpeg
  • Or configure cloud rendering API
  • Check file paths are correct

Database errors:

  • Ensure workflows/data/ directory exists
  • Check write permissions

📚 Additional Resources

  • Workflow Examples: workflow_engine/examples/youtube_automation/
  • Full Documentation: workflow_engine/README.md
  • Main Setup Guide: WORKFLOW_ENGINE_SETUP.md

🎉 You're Ready!

With just an OpenRouter API key, you can start automating your YouTube channel! All other services have free alternatives.

Minimum setup:

  1. Get OpenRouter key
  2. Install dependencies
  3. Load workflows
  4. Run!

The system will use free alternatives for everything else automatically.