Alice

Say 'Hi' to Alice 👋

Alice is an open-source AI assistant for your desktop, offering voice interaction, text-to-speech, and image recognition. It leverages powerful language and vision APIs to enable natural conversations, assist with everyday tasks, and interpret visual inputs with deep, context-aware intelligence.

This project aims to create a personal AI assistant that feels truly present, intuitive, responsive, and emotionally engaging. Alice isn’t just another soulless chatbot, but a thoughtfully crafted companion designed to feel as real as today’s technology allows, while staying both performance and cost-efficient.

Features

• 🗣️ Voice recognition
  Speak to Alice using voice with fast Voice Activity Detection (VAD), transcribed via OpenAI's gpt-4o-transcribe or Groq's whisper-large-v3 API.

• 💬 Text-to-speech
  Alice responds with natural-sounding speech using OpenAI's TTS API.

• 🖼️ Image recognition & Image generation
  Capture screenshots and let Alice interpret them using OpenAI's Vision API. Generate images using OpenAI gpt-image-1.

• 📝 Memory system
  All Alice interactions are stored in a local database, allowing her to remember past conversations and facts.

  • Thoughts: Past conversation fragments embedded in a local Hnswlib vector database for quick context inspiration.
  • Memories: Long-term structured facts stored in a local database, retrievable on demand via memory tools.
  • Summarization: Past messages are automatically summarized and served to Alice for longer context awareness.
  • User emotion awareness: Summarization includes a user emotion summary, enabling more empathetic replies.

• 🎞️ Animated assistant appearance
  Several video states (standby, thinking, speaking) are used to make interactions lively.

• 🪪 Advanced system prompt
  Alice comes with an advanced system prompt that allows a seamless user experience and tool usage. The system prompt can be updated in settings at any time.

• ⏹️ Interruptible responses
  Yes, kinda Realtime API but much cheaper to use ;) Alice can cancel her response mid-stream when interrupted with voice input, including both text and TTS pipelines.

• ⚙️ Function calling
  Data retrieval:

  • Web search
  • Google calendar integration
  • Gmail integration
  • Torrent search
  • Current date and time awareness

  Computer and software use:

  • MCP support
  • Image generation
  • Open applications
  • Open URLs
  • Read/write clipboard content
  • Add torrents to the download client

• 📃 Flexible settings
  Configure Alice for your needs, including:

  • Model
  • System prompt
  • Temperature
  • Top P
  • Max History Messages
  • Summarization Message Count
  • Summarization Model
  • Summarization System Prompt
  • Microphone/Audio toggle, take screenshot hot keys
  • Available tools
  • MCP servers
  • Google integration

Download

Get the latest version of Alice for your operating system:

Use Setup Instructions to create all needed API references and set up your assistant.

Technologies Used

• Vue.js (Frontend framework)
• Electron (Desktop app framework)
• Pinia (App state management)
• OpenAI API (Responses API, TTS, Vision, Image generation, etc ..)
• Groq API (Whisper-large-v3: fast speech-to-text)
• Hnswlib (Vector database for thought retrieval)
• VAD (Voice Activity Detection)
• ChatGPT 4o (Native image generation: Alice image)
• Kling 1.6 Pro (fal.ai) (Image-to-video animation: Alice's animated states)

Tools:

• Jackett (Torrent search aggregation)
• qBittorrent (Torrent client for downloads)

Getting Started With Development

1. Clone the repository:
   git clone https://github.com/pmbstyle/Alice.git
2. Install dependencies:
   npm install
3. Use Setup Instructions to create all needed API references and set up your environment variables in a .env file.
   Note: You can use the example .env.example file as a reference.
4. Run the development server:
   npm run dev

Local build for production

Create app-config.json file in the root directory with the following content to use Google integration (optional):

{
  "VITE_GOOGLE_CLIENT_ID": "",
  "VITE_GOOGLE_CLIENT_SECRET": ""
}

1. Build the Electron app:
   npm run build

2. Install the app from the release folder.

Community links

• ArchLinux port

Contributing

I welcome your ideas, suggestions, and pull requests! Whether it’s a bug fix, a feature idea, or just a creative thought — feel free to open an issue or a PR. Your input helps make Alice smarter and more helpful for everyone 💚