Plugins¶

Voice¶

Lightweight, natural-sounding TTS with voice cloning support. Drop in a short audio sample and UNI speaks in that voice. CPU-only, no GPU required.

Expressive local TTS with style control and voice cloning. Drop in a sample and UNI speaks in any voice. Requires an Nvidia GPU.

Open source wake word detection. Choose from the built-in triggers or train a custom model on your own wake phrase.

Wake word detection by Picovoice. Lightweight and highly accurate with minimal CPU usage. Not available on ARM64 systems.

Enabled by default. A typical voice assistant sphere that pulses with voice. Multiple color schemes available via settings.

Animated 2D characters with reactive expressions. Emotions and gestures shift in real time with the conversation.

Evaluate math expressions with precision. Language models aren't great at arithmetic, so this plugin handles the numbers.

Set timers and alarms by voice. A visual countdown shows on screen and UNI notifies you the moment time runs out.

Play sound effects during conversations: reactions, memes, whatever you want. Add your own clips to the soundboard.

Search the web, read pages, and bring back answers from multiple sources. The results feed right into your conversation.

Current conditions and multi-day forecasts at your configured location. Pushes weather cards to your connected devices.

Control Spotify by voice. Play, skip, search, and queue tracks without leaving the conversation. Requires a Spotify account.

Connect to Model Context Protocol servers running on your host. Any tools they provide become available to UNI automatically.

Keeps UNI aware of the current date and time so it can reason about schedules, deadlines, and time of day.

Remembers facts about you across sessions: preferences, names, past topics. Everything stored locally and encrypted.

UNI can read your recent emails and summarize or act on them. Ask about what came in today, or let it flag what matters.

UNI comes with a Python-based SDK for building plugins.

All you need is a UNI installation, some Python knowledge and a code editor.