Batch convert audio files to .wav, preserve folder structure, clean up after, and optionally analyze them using GPT-4o to auto-rename with musical key and tempo.

• 🔍 Recursively scans downloads/ for .m4a, .opus, and .wav files
• 🔄 Converts to .wav using ffmpeg, preserving folder structure
• 🚀 Parallel processing for speed
• 🧼 Automatically clears converted files from downloads/ (except .gitkeep)
• 💬 Clean CLI with progress logging and friendly output
• 🧠 GPT-4o integration to rename tracks by key + BPM (uses analyze-key-gpt4o.py)
• 🧪 --dry-run mode to preview changes without touching files
• 🎨 Fancy ASCII banner with figlet (fallback included)

ai-audio-tool/
├── downloads/               # Drop your audio files in here
│   └── .gitkeep             # Keeps directory in Git
├── output/                  # Converted files show up here
├── main.py                  # Main batch conversion + GPT automation
├── analyze-key-gpt4o.py     # GPT-based key/BPM analyzer and renamer
├── requirements.txt         # Python dependencies
├── .env                     # Stores your OpenAI API key
└── README.md                # You're reading this

# macOS
brew install ffmpeg

# Ubuntu
sudo apt install ffmpeg

# macOS
brew install figlet

# Ubuntu
sudo apt install figlet

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

requirements.txt:

python-dotenv
openai
tqdm

Create a .env file in the project root:

OPENAI_API_KEY=your-api-key-here

You can get your key from: https://platform.openai.com/account/api-keys

Drop .m4a, .opus, or .wav files into downloads/ (folders are fine):

python main.py

This will:

1. Convert all supported files in downloads/ to .wav
2. Save the converted files to output/, preserving subfolders
3. Clean up the downloads/ directory (excluding .gitkeep)
4. Run analyze-key-gpt4o.py to:
   • Send filenames to GPT-4o
   • Get key + BPM metadata
   • Rename each .wav file to:
     Track Name (Key - BPM).wav

Preview all actions without converting or deleting anything:

python main.py --dry-run

This is useful for testing file paths and GPT output before running for real.

After conversion, analyze-key-gpt4o.py:

• Scans output/ for .wav files
• Sends a cleaned list of filenames to GPT-4o
• Parses the response to extract musical metadata
• Renames matched files with their key + BPM

Example:

Track Name: The Weeknd - Blinding Lights  
Key: F# Minor  
BPM: 171

➡️  output/The Weeknd - Blinding Lights (F# Minor - 171 BPM).wav

If no confident match is found, it skips the file and logs it.

• Instrumentals and obscure tracks may not return metadata from GPT
• GPT relies on accurate filenames — try to include artist and title
• .gitkeep ensures empty downloads/ and output/ dirs are preserved in Git
• You can comment out GPT automation in main.py if not needed

Built with love and ffmpeg, OpenAI, tqdm, and way too many .m4a files.

figlet m4a-wav

MIT License.