Project Ananya - AI Waifu Core

Live Preview / GitHub
PythonGradiofaster-whisperCoqui TTSKokoro TTSGroqOllamaOpenAIChromaDB

An interactive, voice-driven AI companion with modular LLM, advanced TTS, real-time Gradio UI, and persistent memory. Persona: Ananya, a culturally-aware friend who speaks English & Hinglish.

Project Ananya is a modular framework for building personalized AI companions. Features multi-backend LLM (Groq, Ollama, OpenAI), dual TTS (Coqui XTTS v2, Kokoro), real-time Gradio UI, long-term memory with ChromaDB, and a dynamic 3D avatar. Persona, voice, and avatar are fully customizable via YAML and config files.

  • Multi-backend LLM: Groq, Ollama, OpenAI
  • Dual TTS: Coqui XTTS v2 (voice cloning), Kokoro (local, fast)
  • Real-time Gradio UI with streaming mic/audio
  • Persistent memory (ChromaDB vector store)
  • Dynamic 3D avatar (VRM, lip-sync)
  • Persona & voice fully configurable
2025 — Built by Hitesh Kumar