ZcChat is an AI desktop pet designed to emulate the experience of Galgame characters, offering long-term memory, expressive character art, voice interaction, and computer control capabilities. It targets users seeking an interactive and visually engaging virtual companion.

How It Works

ZcChat utilizes a Galgame-style approach with static character art (立绘) that changes expressions and poses, differing from Live2D animation to reduce production costs and allow for more distinct visual states. It integrates with LLM frameworks like Letta (formerly MemGPT) for long-term memory and personality development, and supports various TTS and ASR tools for realistic voice interaction.

Quick Start & Requirements

• Install: Download and install from the Release page. Avoid non-ASCII characters in installation paths.
• Prerequisites:
  • LLM backend (Letta or OpenAI compatible API).
  • Optional: VITS-based TTS (e.g., vits-simple-api), Whisper ASR (e.g., whisper-asr-webservice), or Baidu Speech Recognition.
• Setup: Requires downloading character assets and configuring LLM/voice backends.
• Links: Video Tutorial, Character Creation Docs, Letta Docs

Highlighted Details

• Employs Galgame-style character art for expressive animations and easier asset creation.
• Integrates Letta for advanced AI memory and personality management.
• Supports voice interaction with wake-word detection and interruption capabilities.
• Allows for computer control via LLM prompts.

Maintenance & Community

• Open-source project seeking contributions via Pull Requests and bug reports via Issues.
• Discussions forum available for sharing character templates.
• Related projects include LogChat (AI chat client) and vits-simple-api.

Licensing & Compatibility

• The project's primary license is not explicitly stated in the provided README text. Dependencies may have their own licenses.

Limitations & Caveats

• Voice wake-word and interruption functionality are sensitive to environmental noise and require careful configuration of audio energy thresholds.
• The project's code quality is noted by the maintainer as potentially difficult to navigate for new contributors.