🎵 AI-powered karaoke video creator using Docker Spleeter for ML-based vocal separation and OpenAI Whisper for transcription.

Automatically download trending music videos and convert them to karaoke format every day!

# Quick start - Set up daily automation
./setup_daily_karaoke.sh

# Or run manually now
python3 daily_karaoke_generator.py

# Test your setup
./test_setup.sh

📚 Complete Daily Karaoke Guide →

1. ✅ Downloads trending music videos daily from YouTube
2. ✅ Separates vocals from instrumentals using Docker Spleeter
3. ✅ Creates karaoke videos with instrumental-only tracks
4. ✅ Organizes output in timestamped folders
5. ✅ Generates comprehensive reports and logs
6. ✅ Automatic cleanup of old runs

Perfect for: Daily karaoke content creation, music libraries, entertainment venues

• 🤖 Docker Spleeter integration - True ML-based vocal/instrumental separation
• 🗣️ OpenAI Whisper transcription - Accurate speech-to-text with word-level timing
• 📝 Synchronized subtitles - Word highlighting with professional typography
• 🎚️ Multiple vocal levels - 0%, 5%, 10%, 15%, 25%, 50%, 75%
• 🌐 Bilingual support - Original language + transliteration
• ☁️ Cloud deployment - Google Cloud Run ready
• ⏰ Daily automation - Scheduled runs via cron or systemd

• Python 3.9+
• Docker (for Spleeter audio separation)
• FFmpeg
• yt-dlp (for video downloads)
• OpenAI API key (optional, for Whisper transcription)

# 1. Install dependencies
pip install -r requirements.txt
pip install yt-dlp

# 2. Pull Spleeter Docker image
docker pull researchdeezer/spleeter:3.8-2stems

# 3. Test setup
./test_setup.sh

# 4. Set up daily automation (optional)
./setup_daily_karaoke.sh

Create a .env file:

# Optional: For Whisper transcription
export OPENAI_API_KEY="your-openai-api-key"

# Optional: For cloud storage
export BUCKET_NAME="your-gcs-bucket"

# Optional: Custom port for web service
export PORT=8080

# Run daily karaoke generation manually
python3 daily_karaoke_generator.py

# Or use the wrapper script
./run_karaoke_now.sh

Output will be saved to: karaoke_daily_runs/run_YYYYMMDD_HHMMSS/

# Start processing service
docker-compose up processing

# Run in background
docker-compose up -d processing

# View logs
docker-compose logs -f processing

# Test Spleeter directly
mkdir -p input output
docker-compose --profile tools run spleeter separate -i /input/video.mp4 -o /output -p spleeter:2stems

# Multiple vocal levels (0%, 5%, 10%, 15%, 25%, 50%, 75%)
python create_multi_vocal_karaoke.py

# Low vocal levels (5%, 10%, 15%)
python create_low_vocal_karaoke.py

# Download & process YouTube videos
python download_and_create_karaoke.py

1. Set up Google Cloud Project

gcloud config set project YOUR_PROJECT_ID
gcloud services enable run.googleapis.com cloudbuild.googleapis.com storage.googleapis.com

2. Create Storage Bucket

gsutil mb gs://soniq-karaoke-videos

3. Deploy with Cloud Build

gcloud builds submit --config=cloudbuild.yaml

Health Check

curl https://soniq-karaoke-HASH-uc.a.run.app/health

Process Video

curl -X POST https://soniq-karaoke-HASH-uc.a.run.app/process \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://www.youtube.com/watch?v=VIDEO_ID",
    "vocal_levels": [0.0, 0.25, 0.5]
  }'

Response:

{
  "job_id": "uuid",
  "title": "Video Title",
  "videos": [
    {
      "vocal_level": 0,
      "url": "https://storage.googleapis.com/bucket/file.mp4",
      "filename": "karaoke_0_vocal.mp4"
    }
  ]
}

YouTube URL → yt-dlp → Docker Spleeter → OpenAI Whisper → FFmpeg → Cloud Storage
                ↓              ↓             ↓           ↓
            Video File    Vocal/Instrumental  Subtitles   Karaoke Video

1. Download - Extract video from YouTube URL
2. Separate - Use Docker Spleeter for ML-based audio separation
3. Transcribe - OpenAI Whisper for word-level timestamps
4. Subtitle - Create synchronized ASS subtitles with highlighting
5. Mix - Combine vocals/instrumentals at specified levels
6. Render - Generate final karaoke video with FFmpeg
7. Upload - Store in Google Cloud Storage

• Python Spleeter - ML audio separation (replaced Docker-in-Docker)
• OpenAI Whisper - Speech transcription
• yt-dlp - YouTube video downloading
• FFmpeg - Video/audio processing
• Flask - Web API framework
• Google Cloud Run - Serverless deployment
• Google Cloud Storage - Video storage

✅ Trigger "soniqpush" active - pushes to main branch automatically deploy to Cloud Run!

Created karaoke videos with various vocal levels:

• punjaban_karaoke_0_vocal.mp4 - Pure instrumental
• punjaban_karaoke_25_vocal.mp4 - Light vocal guide
• punjaban_karaoke_50_vocal.mp4 - Balanced mix
• punjaban_karaoke_75_vocal.mp4 - Strong vocal guide

Perfect for different karaoke preferences and skill levels!

MIT License - See LICENSE file for details.# GitHub Auto-Deployment Test Fri 22 Aug 2025 15:28:44 EDT