YakGPT offers a locally-run, hands-free web interface for interacting with OpenAI's GPT models, targeting users who want a faster, more private, and voice-enabled ChatGPT experience. It bypasses official UIs for direct API interaction, enhancing speed and user control over data.

How It Works

This project leverages a Next.js frontend architecture, enabling direct browser-to-API communication for ChatGPT interactions. It integrates speech-to-text via Azure and OpenAI Whisper, and text-to-speech via Azure and Eleven Labs, facilitating a hands-free, voice-driven chat experience. All application state is managed locally in localStorage, ensuring no external analytics or data collection.

Quick Start & Requirements

• Install dependencies and run: yarn install && yarn build && yarn start
• Prerequisites: Git, Yarn (or npm/pnpm), modern web browser.
• API Keys: OpenAI (GPT-3.5/GPT-4), Eleven Labs (optional for TTS).
• Docker: docker run -it -p 3000:3000 yakgpt/yakgpt:latest
• Demo: https://yakgpt.vercel.app/

Highlighted Details

• Hands-free operation via microphone integration.
• Direct API connection for faster responses.
• Local state storage for enhanced privacy.
• Supports GPT-3.5 and GPT-4 via OpenAI API.
• Speech-to-text (Azure, Whisper) and Text-to-speech (Azure, Eleven Labs) integration.

Maintenance & Community

• Project is open for issues and feedback.

Licensing & Compatibility

• MIT License. Permissive for commercial use and integration with closed-source projects.

Limitations & Caveats

The Docker image is currently limited to amd64 architecture; custom builds are required for others. GPT-4 API access is necessary for GPT-4 functionality.