nalemi/Lisa_s-Persona-Community-Edition

Lisa_s (Persona Community Edition) 🎙️✨

A fully offline, standalone AI companion with integrated voice interaction and zero censorship.

Lisa_s is a native Windows desktop roleplay companion. Unlike standard WebUI setups that require running complex local servers, Docker containers, or multi-tab browser pipelines, Lisa_s is bundled as a self-contained portable Windows application with its own native graphical interface.

---

🚀 Download & Community Links

The complete bundled application (including engine, voice models, and quantized GGUF) is hosted externally for seamless one-click downloading:

• ☕ Get it on Ko-fi (Free Download): Download Lisa_s Community Edition on Ko-fi
• 💾 Direct Google Drive Host: Access provided via the Ko-fi download page.
• ☕ Support the Project: If you enjoy chatting with Lisa, tips and community memberships on Ko-fi directly fund larger base-model integrations and voice upgrades!

---

✨ Key Features

• 100% Offline & Private: Zero API keys, no telemetry, and no cloud dependency. Everything runs directly on your local CPU/GPU hardware.
• Full Voice-to-Voice Loop:
  • STT (Speech-to-Text): Integrated local Whisper engine listens via your default microphone.
  • TTS (Text-to-Speech): Ultra-fast local Piper synthesis generates low-latency speech.
  • Text Fallback: Full keyboard input and chat log visualization built directly into the UI.
• Uncensored Roleplay Core: Driven by an abliterated Qwen2.5-3B-Instruct quantization (Q5_K_M.gguf). Free from artificial refusals, lecturing, or moral gatekeeping during creative roleplay sessions.
• Truly Portable Desktop Application: Single .exe deployment. No Python environment setup, no Conda, and no terminal commands required.

---

💻 System Requirements

Component	Minimum	Recommended
OS	Windows 10 (64-bit)	Windows 10 / 11 (64-bit)
RAM	8 GB	16 GB
Storage	~5 GB free space	SSD recommended for fast model loading
Hardware	Modern multi-core CPU (Intel/AMD)	Dedicated GPU (VRAM helps, but CPU-only runs smoothly for 3B)
Peripherals	Microphone & Speakers/Headphones	Dedicated headset for clean voice input

---

🛠️ Quick Start Guide

1. Download: Download the ZIP archive from the (https://ko-fi.com/s/b19d752b96).
2. Extract: Extract the ZIP folder anywhere on your PC (e.g., Desktop or C:\Persona\).
3. Launch: Run the executable directly.
4. Talk or Type: Speak through your microphone or type into the input field to interact with Lisa.

Note on Windows SmartScreen:
Because this is a portable, independent community release without an expensive commercial code-signing certificate, Windows Defender / SmartScreen may display an "Unknown Publisher" prompt on the very first launch. Simply click "More info" and then select "Run anyway".

---

🛡️ Model & Technical Specifications

• Base Architecture: Qwen 2.5 (3 Billion Parameters)
• Alignment: Abliterated / De-censored instruction tuning
• Format: GGUF (Q5_K_M medium-high precision quantization)
• Speech Recognition: Whisper (local lightweight footprint)
• Speech Synthesis: Piper TTS (neural offline speech)
• Interface: Native PyQt desktop application

---

🤝 Community & Feedback

Have suggestions for voice tuning, personality prompts, or expanded model weights? Leave feedback on the community channels or reach out via Ko-fi!