Canonical has released early testing versions of Myna, a new local AI-powered dictation tool for Ubuntu that enables offline speech-to-text transcription using models like Whisper and Parakeet.
Key Points
- Myna components, including the orchestrator snap and desktop settings app, are available for Ubuntu 26.04 LTS and 26.10.
- Users can choose between AI models like OpenAI’s Whisper or NVIDIA’s Parakeet, which require 1-2GB of disk space each.
- The tool operates entirely offline, ensuring that audio data remains on the user's device without cloud processing.
- Ubuntu 26.10 includes a preinstalled GNOME Shell extension that provides an on-screen heads-up display (HUD) for transcription status.
- The software is currently in an experimental state, requiring manual installation via command-line snaps and PPA repositories.