Mimic AI Desktop Assistant

A privacy-first, AI-powered desktop assistant with voice synthesis, persistent memory, and local model inference. Mimic AI runs entirely on your local machine using Ollama for language models and supports Browser TTS and Qwen3-TTS for voice synthesis.

Table of Contents

• Features

• System Requirements

• Installation

• Quick Start

• Architecture Overview

• Data Flow

• Component Interaction

• Configuration

• Usage Guide

• Troubleshooting

• Privacy & Security

Features

Core Capabilities

• Local AI Inference: Uses Ollama for running LLMs locally (Llama, Mistral, etc.)

• Voice Synthesis: Browser TTS (no setup) and Qwen3-TTS (AI voice cloning)

• Persistent Memory: Conversation memory with per-persona context

• Web Search: Privacy-focused SearXNG integration

• Wake Word Detection: Hands-free activation with custom wake words

• Vision Support: Image analysis with vision-capable models

• Audio Watermarking: All generated audio is invisibly watermarked for identification

Persona System

• Create multiple AI personas with unique personalities

• Each persona has independent memory and voice profiles

• Voice creation with adjustable parameters (pitch, speed, emotion, etc.)

• Avatar customization with dynamic 3D visuals

Avatar Personality System (New!)

• Automatic Trait Extraction: Derives avatar personality traits (energy, playfulness, expressiveness, curiosity, empathy, formality) from personality prompts

• Voice Parameter Derivation: Automatically adjusts voice parameters based on extracted personality traits

• Dynamic Emotional State: Avatar tracks and displays emotional states during conversations

• Height-Based Camera: Camera intelligently positions based on avatar model height for optimal viewing

• Procedural Vocalizations: AI-generated non-verbal sounds (giggles, sighs, hums) using Qwen TTS (cached, max 10 per persona)

Persona Learning System (New!)

• Interaction Tracking: Records conversation history and user preferences per persona

• Emote Learning: Tracks preferred emotes and animations based on user feedback

• Relationship Metrics: Calculates rapport, familiarity, and trust scores over time

• Favorite Topics: Automatically identifies and remembers user's favorite conversation topics

• Emotional History: Maintains emotional state history for more natural interactions

Licensing

• 7-Day Free Trial: Full functionality for 7 days from first launch

• Subscription: $5/month via Patreon for continued use after trial

• Support Development: Subscribe to help fund ongoing development

• Subscribe: https://www.patreon.com/c/MimicAIDigitalAssistant

Memory Management

• File Memories: Upload text files for the AI to reference

• Conversation Memories: Automatic extraction of important conversation points

• Memory Controls: Per-persona memory management with delete/recall capabilities

• Summary Generation: Automatic conversation summarization

System Requirements

Minimum Requirements

• OS: Windows 10/11, macOS 12+, or Linux

• RAM: 8GB (16GB recommended)

• Storage: 5GB free space

• GPU: Optional but recommended for faster inference

Software Dependencies

• Python: 3.10, 3.11, or 3.12 (3.13+ not supported due to PyTorch compatibility)

• Node.js: 18.x or 20.x LTS

• Ollama: Latest version from ollama.com

• Docker Desktop: Optional, for SearXNG web search

VRAM Requirements (Voice Synthesis)

• Browser TTS: No VRAM required (uses system voice)

• Qwen3-TTS 0.6B: ~3GB VRAM

• Qwen3-TTS 1.7B: ~6GB VRAM

Attribution

Character Model

• Default Avatar: PeePhanthong (VRoid Hub)

Animations

• VRoid Project by pixiv Inc.

• 八ツ橋まろんのお店 (Yatsuhashi Maron's Shop)

• Kannaku @ Nekokoya

See ATTRIBUTION.md for full credits and license terms.

License

Proprietary License with 7-Day Trial - See LICENSE.txt for details

Copyright (c) 2026 Mimic AI. All Rights Reserved.

Support

• Patreon: https://www.patreon.com/c/MimicAIDigitalAssistant

• Issues: Use GitHub issue tracker

• Documentation: See QUICKSTART.md