a4f-local

A unified, professional Python wrapper for various reverse-engineered AI provider APIs, designed to be OpenAI-compatible and easy to use.

Key Features

• Unified API: Call multiple unofficial AI providers with a single, OpenAI-like interface.
• Modular: Easily extend with new providers and capabilities.
• Supports Text-to-Speech (TTS): Initial implementation includes OpenAI.fm reverse-engineered TTS.
• OpenAI-Compatible: Accepts and returns data in the same format as OpenAI's official API.
• Simple to Use: Designed for developers of all skill levels.

Installation

Install directly from PyPI:

pip install -U a4f-local

Usage Example

from a4f_local import A4F

client = A4F()

try:
    audio_bytes = client.audio.speech.create(
        model="tts-1",  # Model name (currently informational)
        input="Hello from a4f-local!",
        voice="alloy"   # Choose a supported voice
    )
    with open("output.mp3", "wb") as f:
        f.write(audio_bytes)
    print("Generated output.mp3")
except Exception as e:
    print(f"An error occurred: {e}")

Supported Voices

The following voice names are supported (mapped to OpenAI's official voices):

• alloy
• echo
• fable
• onyx
• nova
• shimmer

Use one of these as the voice parameter in your TTS requests.

How It Works

• You interact with the A4F client just like OpenAI's Python SDK.
• The client automatically discovers available providers and their capabilities.
• When you call a method (e.g., client.audio.speech.create()), it routes the request to the appropriate provider.
• The provider translates the OpenAI-compatible request into its own API format, calls the API, and translates the response back.

Roadmap

• More Providers: Support for chat, image generation, and other AI capabilities.
• Configuration: Easier ways to configure API keys and provider preferences.
• Async Support: Async versions of API calls.
• Better Error Handling: More informative error messages and exceptions.

License

See the LICENSE file for details. This software is not open source in the traditional sense. Please review the license terms carefully before use.

Disclaimer

This package uses reverse-engineered, unofficial APIs. These may break or change at any time. Use at your own risk.

More Information

More detailed documentation, tutorials, and examples will be published here and on PyPI.

For now, refer to the example above and the source code for guidance.