Repolode
Submit
dnhkng

dnhkng/GLaDOS

A local voice assistant with a GLaDOS-style custom voice, camera vision, persistent memory, tools and a browser console.

View on GitHub
Stars
5,750
Forks
446
Open issues
10
Last commit
10 days ago

What is it?

GLaDOS is a voice assistant built around the personality core from Valve's Portal series. It supports talking through a local microphone or a browser, with a GLaDOS voice and an animated avatar that reacts to spoken emotion directions.

The system is split into independent cores for emotion, memory, vision, system health and research. Their observations feed a shared context, and an Autonomy Core decides when an update is worth a spoken response. Conversation can run on a local llama.cpp server using Gemma 4 E4B with direct audio input, or on an extended profile that uses Parakeet transcription with Ollama or an OpenAI-compatible API.

The project also includes a web console for controlling devices and inspecting memory, context and routing decisions, MCP tool integration (including Home Assistant), and an optional OpenAI-style speech endpoint that serves the GLaDOS voice.

Key features

  • 01Voice conversation with Gemma 4 E4B direct audio or Parakeet transcription
  • 02Browser console for microphone, camera, memory and routing inspection
  • 03Camera-based scene description with timestamped frames
  • 04Persistent memory of conversation summaries and facts
  • 05MCP tool integration, including Home Assistant
  • 06Optional speech API serving the GLaDOS voice in MP3, WAV or OGG

Use cases

  • Building a voice assistant with a custom character voice on a local GPU
  • Controlling smart home devices through Home Assistant via voice
  • Experimenting with multi-agent cores that share context and decide when to speak
  • Generating GLaDOS-voiced speech through a local API

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/dnhkng/GLaDOS.git
  2. 2

    Enter the project folder

    cd GLaDOS
  3. 3

    Run the installer (CUDA, ROCm or CPU selected automatically)

    python scripts/install.py
  4. 4

    Start the web console and open http://127.0.0.1:8050/ (requires a separately running llama-server model)

    uv run --no-sync glados webapp --config configs/glados_webapp_config.yaml

Things to know

  • Needs a separately installed CUDA llama-server and about 16 GB-class GGUF model files to run the default conversation model.
  • Requires a GPU for the reference setup; latency and memory depend heavily on your hardware.
  • macOS support is experimental and the AMD installer path is Linux-only.

Topics

#voice-assistant#llm#speech-synthesis#llama-cpp#mcp#home-assistant#computer-vision#python

FAQ

What is GLaDOS?
A local voice assistant with a GLaDOS-style custom voice, camera vision, persistent memory, tools and a browser console. GLaDOS is a voice assistant built around the personality core from Valve's Portal series. It supports talking through a local microphone or a browser, with a GLaDOS voice and an animated avatar that reacts to spoken emotion directions.
Is GLaDOS free to use? What license does it have?
GLaDOS is open source under the MIT license. Check the license terms before using it in commercial products.
Is GLaDOS still maintained?
Actively maintained. The last commit to GLaDOS was on Oct 11, 2026.
What can I use GLaDOS for?
Building a voice assistant with a custom character voice on a local GPU Controlling smart home devices through Home Assistant via voice Experimenting with multi-agent cores that share context and decide when to speak Generating GLaDOS-voiced speech through a local API
How do I install GLaDOS?
Clone the repository: git clone https://github.com/dnhkng/GLaDOS.git Enter the project folder: cd GLaDOS Run the installer (CUDA, ROCm or CPU selected automatically): python scripts/install.py Start the web console and open http://127.0.0.1:8050/ (requires a separately running llama-server model): uv run --no-sync glados webapp --config configs/glados_webapp_config.yaml
Where can I get GLaDOS?
The source code and installation instructions are on GitHub: https://github.com/dnhkng/GLaDOS

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects