Myna in Ubuntu 2026: Is voice-controlled system management ready for office work?

MarGib August 16, 2026

🌐 🇵🇱 Polski · 🇬🇧 EN

Canonical introduces Myna – a new open-source project designed to revolutionize interaction with Ubuntu via voice. Is it a ready-to-use tool for daily work, or still an experiment for early adopters? We examine its capabilities, how it works, and its limitations in 2026.

Interfejs Ubuntu z wizualizacją fal dźwiękowych i tekstem dyktowanym głosem — Illustration of Myna in action in Ubuntu 2026: system control via voice.

What is Myna and why is Canonical building it?

Myna is a project developed by Canonical, the company behind Ubuntu, aimed at enabling operating system control via voice. Officially announced in June 2026, Myna is not just another dictation tool, but a comprehensive solution that integrates with the GNOME interface, GTK applications, and core system functions. Its primary goal is to increase accessibility – especially for users with motor disabilities – and improve productivity by automating daily tasks.

Canonical treats Myna as a priority project. Mark Shuttleworth, the company's CEO, called it "a breakthrough in open-source accessibility" during the Ubuntu Summit 2025. Unlike commercial solutions such as Dragon naturallyspeaking, Myna is fully open-source and runs locally, without the need for cloud connectivity. This is a key advantage for users who value privacy.

What technologies power Myna?

Myna does not use external APIs like Google Speech-to-Text or OpenAI's Whisper. Instead, Canonical has developed a proprietary AI model in collaboration with the Ubuntu AI Lab. The model is based on a transformer architecture similar to the one used in Whisper, but optimized for integration with the operating system.

Hardware requirements and operation

Minimum requirements: 4-core processor, 8 GB RAM.
Recommended: 8+ cores, 16 GB RAM, GPU acceleration (NVIDIA CUDA or AMD ROCm).
Offline operation: The model processes speech locally, which eliminates latency and the risk of data leaks.
System integration: Myna runs as a system service, using pipewire to capture audio from the microphone and integrating with the GNOME Shell.

It is worth noting that, for now, Myna does not support Wayland – it only works in an X11 session. Canonical has announced support for Wayland in Ubuntu 26.10, which is set to debut in October 2026.

What can Myna do in practice?

Myna offers two main features: text dictation and system control. Here is what you can do with it right now:

Text dictation

Real-time speech-to-text processing in applications such as LibreOffice, Geary, or Firefox.
Formatting support: commands like *"new paragraph"*, *"bold word"*, or *"semicolon"* are recognized.
Available languages: English (US/UK), Spanish, German, French, Italian, Portuguese (BR), Polish, Russian, Japanese, and Chinese (Mandarin).

In tests conducted by OMG! Ubuntu in June 2026, Myna achieved ~85% accuracy for English under optimal conditions. Unfortunately, in noisy environments or with heavy accents, accuracy dropped to ~60%.

System control

Launching applications: *"Launch Firefox"*, *"Open system settings"*.
Interface navigation: *"Minimize window"*, *"Switch to workspace 2"*.
Experimental terminal commands: *"Run sudo apt update"* (disabled by default for security reasons).

However, Myna does not work in all applications. For now, it does not support Qt-based programs (e.g., KDE Plasma, VLC) or the GNOME Terminal. Users must use alternatives like Tilix or Konsole.

How does it handle Polish?

Polish is one of the supported languages, but with limited accuracy. In tests, Myna struggled with:

Verb conjugation (e.g., *"idę"* vs *"idziemy"*).
Recognizing proper nouns and homophones (e.g., *"morze"* vs *"może"*).
Specialized vocabulary (e.g., medical or programming terminology).

Canonical has announced improvements in accuracy for Slavic languages by the end of 2026.

How to install Myna on Ubuntu?

Myna is not yet available in the default Ubuntu 24.04 LTS repositories. Here are three ways to install it (as of July 2026):

1. Snap (recommended by Canonical)

sudo snap install myna --edge

The package --edge represents the unstable version, available since June 15, 2026.

2. PPA (unofficial, community-maintained)

sudo add-apt-repository ppa:myna-dev/ppa
sudo apt update
sudo apt install myna

Note: The PPA is not supported by Canonical.

3. Source code (for advanced users)

Available on GitHub, but requires manual compilation.

Limitations and challenges

Myna is a promising project, but it still has many limitations that may discourage users looking for a ready-to-use solution for daily work.

Accuracy and performance

In a quiet room, Myna achieves ~85% accuracy for English, but in noisy environments, it drops to ~60%.
On lower-end hardware (e.g., Intel Celeron), there are 1-2 second delays between speaking and text appearing.
RAM usage: ~1.5 GB during active dictation.

Security

Myna does not store recordings – processing happens locally.
Experts point out the risk of attacks, such as malicious command injection via recorded audio. Canonical plans to introduce sandboxing in future versions.

Lack of integration with some applications

Myna does not work in Qt applications (e.g., KDE Plasma, VLC).
Limited terminal support: only works in Tilix and Konsole (with limitations).

Is Myna ready for office work?

Early tester opinions are mixed. Here is what users and experts are saying:

"Myna is a step in the right direction, but it is not yet ready for daily use. It lacks the precision and fluidity offered by commercial solutions (e.g., Dragon naturallyspeaking)."

— OMG! Ubuntu, June 2026

Users on Reddit praise the ease of installation and GNOME integration, but complain about the lack of documentation and issues with Bluetooth microphones.
Phoronix noted in performance tests that Myna consumes fewer resources than Whisper.cpp, but is less accurate.

Who is Myna useful for right now?

People with motor impairments (e.g., RSI, paralysis).
Users needing quick note dictation (e.g., students, journalists).
Developers testing terminal commands (with sudo permissions disabled).

Who is it not for yet?

People working in noisy environments (e.g., open-plan offices).
Users requiring 100% accuracy (e.g., lawyers, doctors).
People using Qt applications or Wayland.

Myna development roadmap – what's next?

Canonical has announced an ambitious roadmap for Myna. Here is what is planned for the coming months and years:

Ubuntu 26.10 (October 2026)

Support for Wayland.
Integration with LibreOffice and Thunderbird.
Improved accuracy for Polish and other Slavic languages.

2027 and beyond

Myna Pro: Commercial version with a cloud API (translation, text summarization).
Support for Qt applications and Steam (voice-controlled gaming).
Myna for IoT: Version for embedded devices (e.g., Raspberry Pi).
Expansion of language support to over 30, including Ukrainian, Arabic, and Hindi.

Canonical emphasizes that Myna is not an experimental project, but a long-term investment. The company has hired AI teams from the University of Cambridge and the Warsaw University of Technology to improve language models and system integration.

Myna vs the competition – how does it compare?

Myna is not the only speech recognition tool, but it is the only one natively integrated with Ubuntu. Here is a comparison with other solutions:

Tool	Type	Accuracy (Eng.)	Polish support	Offline	Ubuntu integration
Myna	Open-source	~85%	Yes (limited)	Yes	Native
Whisper (OpenAI)	Open-source	~92%	Yes	Yes*	None
Vosk	Open-source	~88%	Yes	Yes	None
Dragon (Nuance)	Commercial	~95%	No	Yes	None

*Requires manual compilation.

Myna performs well in terms of system integration and privacy, but still lags behind the competition in terms of accuracy. Its greatest advantage is ease of use – it requires no configuration or additional tools.

Summary: Is it worth trying Myna?

Myna is a promising project that could revolutionize how we use Ubuntu. However, in July 2026, it is still an early beta that requires many improvements before it becomes a ready-to-use tool for daily work.

Advantages of Myna

Native integration with Ubuntu and GNOME.
Offline operation, no need to connect to the cloud.
Support for multiple languages, including Polish.
Easy installation via Snap or PPA.

Disadvantages of Myna

Limited accuracy, especially in noisy environments.
No support for Wayland and Qt applications.
Issues with recognizing Polish and other Slavic languages.
High hardware requirements for smooth operation.

If you are a Linux enthusiast and want to test new technology, Myna is worth trying. However, if you are looking for a ready-to-use solution for office work, it is better to wait for further updates or consider alternatives like Whisper or Vosk.

Canonical has announced that Myna will be gradually improved, and we will see its full potential only in 2027. It is worth following the project's development, as it may change the way we use computers.

If you are interested in the topic of AI integration with operating systems, check out our article on Sakana AI and its open-source models or read about autonomous AI agents that could revolutionize our work.