Babagaboosh is a Python application enabling verbal conversations with OpenAI's GPT-4o model. It's designed for users who want to interact with advanced AI through speech, offering a straightforward way to engage in spoken dialogue.

How It Works

The application orchestrates a pipeline involving speech-to-text, AI language processing, and text-to-speech. It captures user audio via the microphone, transcribes it using Microsoft Azure TTS, sends the text to GPT-4o for a response, and then converts the AI's reply into speech using Elevenlabs. This multi-service approach allows for natural, spoken interactions with the AI.

Quick Start & Requirements

• Install: pip install -r requirements.txt
• Prerequisites: Python 3.9.2, API keys for Microsoft Azure TTS, Elevenlabs, and OpenAI. OpenAI requires a minimum $5 credit balance for GPT-4o access.
• Optional: OBS Studio (v28+) and the Move OBS plugin for animated visuals synchronized with speech.
• Setup: Requires configuring environment variables with API keys. Detailed setup guide available in the README.

Highlighted Details

• Supports GPT-4o for advanced conversational AI.
• Integrates Microsoft Azure TTS for speech-to-text and Elevenlabs for text-to-speech, offering customizable AI voices.
• Optional OBS integration allows for dynamic visual responses synchronized with spoken output.

Maintenance & Community

• Developed by DougDoug. Credit is appreciated but not required.
• A video explanation is available: https://www.youtube.com/watch?v=vYE1rkIMj9w

Licensing & Compatibility

• The project is freely available for use without explicit licensing mentioned in the README.

Limitations & Caveats

• Requires paid access to OpenAI's GPT-4o model and API keys for all integrated services.
• OBS integration is optional and requires specific OBS versions and plugins.
• Users must ensure API keys are correctly set as environment variables.