Myna speech-to-text reaches technology preview on Ubuntu Desktop
Ubuntu has made Myna available to the wider community as a technology preview. Myna is an optional desktop dictation feature: place the cursor in a text field, press a configured shortcut, speak, and the tool transcribes into the application in use. This is not a voice assistant for controlling the desktop. Its current purpose is dictation, turning spoken words into text.
The key detail is where recognition happens. Ubuntu says speech recognition runs on the user’s own machine. After a model is downloaded, transcription does not require an Internet connection and audio is not uploaded or stored remotely. That makes Myna relevant to people who draft notes, email, or documents and prefer not to send audio to a cloud service.
What the source actually says
Myna is an optional Ubuntu 26.10 feature, while the preview is available for Ubuntu 24.04 LTS, Ubuntu 26.04 LTS, and Ubuntu 26.10. Support is limited to amd64 and amd64v3 systems using GNOME on Wayland. It should therefore not be described as a finished solution for every Linux desktop, CPU architecture, or display session.
The software supports several recognition engines, including Parakeet, Whisper, and FunASR. They differ in language support, accuracy, and hardware requirements. Ubuntu does not claim that one engine wins in every case. A useful evaluation uses the real language, accent, microphone, and document type a person works with, rather than a short polished demonstration.
Supported NVIDIA GPUs can enable a GPU-accelerated engine, and Ubuntu plans support for additional accelerators. Performance will consequently vary across machines. Older laptops, CPU-only systems, and noisy microphones may feel very different from a well-equipped test system.
Why it matters
Local speech-to-text addresses two recurring concerns: privacy and connectivity. People can dictate while offline or in restricted networks, and sensitive audio does not need to leave the device for recognition. For IT teams, this reduces questions about a third-party audio-processing service. Local processing is not the entire security story, however. The resulting text still enters the target application, clipboard, file, or synchronization system already used by the organization.
The technology-preview label matters as well. Ubuntu explicitly says Myna remains under active development and users should expect rough edges. This is the right stage for a controlled pilot and specific bug reports, not for making it mandatory in a critical workflow. Errors involving names, numbers, negations, or specialist terms can change meaning completely, so text should always be reviewed before it is sent or relied on.
Practical trial checklist
Ubuntu’s installation path is to add the canonical-desktop-team/myna PPA, install myna-config, then open Myna Settings from Activities and complete setup. After configuring a shortcut, test it in an ordinary text field. Installing the package alone does not prove the environment is suitable; confirm GNOME, Wayland, CPU architecture, model storage, and hardware first.
- Test with the microphone, language, local accent, and applications actually used each day.
- Include names, numbers, punctuation, and specialist terminology in sample passages.
- Measure responsiveness from the shortcut to displayed text, especially on battery-powered or CPU-only systems.
- Review where the output text is saved, synchronized, or shared; local transcription does not govern the destination app.
- Do not treat preview output as the sole record for technical commands, minutes, or high-accuracy material.
Ubuntu is specifically seeking feedback on transcription quality, responsiveness, language and accent support, application compatibility, setup, and onboarding. A useful report identifies the Ubuntu version, hardware, selected engine, destination application, and a redacted example of the issue. Canonical develops Myna openly and provides its code, issue tracker, and discussions on GitHub.
Conclusion
Myna offers a promising local dictation option for Ubuntu Desktop, with offline use and on-device audio processing as practical benefits. It remains a preview with platform limits and quality that needs validation on each machine. The sensible approach is a small controlled trial, careful review of output, and concrete feedback that helps the project improve.
Source: Ubuntu Blog – Myna speech-to-text is now available as a technology preview.