# How to use Myna local dictation on Ubuntu > Canonical's Myna preview can dictate into a focused app on supported Ubuntu GNOME Wayland desktops. Follow the documented PPA setup and check its limits. By BIG CHANGE Editorial Published: 2026-10-15T21:21:15.837Z Updated: 2026-10-15T21:21:15.837Z Canonical: https://bigchange.ai/blog/ubuntu-myna-local-speech-to-text-guide ![Conceptual charcoal illustration of one person speaking toward an unbranded laptop with a blank screen; the keyboard is untouched.](https://bigchange.ai/api/media/file/myna-local-dictation-concept-v1.png) AI-generated conceptual illustration by BIG CHANGE; no Myna interface, observed session or transcription result is depicted. Canonical made Myna available as a technology preview on October 12, 2026, three days before the October 15 release of Ubuntu 26.10. The optional dictation tool takes speech from a microphone, recognizes it on the computer and inserts text into the application with the active text field. Canonical says it also works on two earlier Ubuntu releases. This guide follows Canonical's published setup instructions; BIG CHANGE has not installed or tested Myna. ## Check your desktop before installing The [preview announcement](https://ubuntu.com/blog/myna-speech-to-text-technology-preview-available) lists Ubuntu 24.04 LTS, 26.04 LTS and 26.10, on amd64 or amd64v3, using GNOME on Wayland. It does not claim support for every Ubuntu flavor, desktop session or processor architecture. You need a working microphone, a text field in an application and an internet connection for installation and the initial model download. Canonical says transcription needs no internet connection after a model is downloaded. Ubuntu 26.10's [October 15 release announcement](https://ubuntu.com/blog/canonical-releases-ubuntu-26-10-stonking-stingray) calls Myna an opt-in feature. That release date should not be confused with the October 12 technology preview or taken to mean Myna is enabled by default. ## Install and dictate, following Canonical's preview instructions 1. On a supported Ubuntu GNOME Wayland desktop, add Canonical's Myna PPA in a terminal: ```bash sudo add-apt-repository ppa:canonical-desktop-team/myna ``` 2. Install the configuration application: ```bash sudo apt install myna-config ``` 3. Open **Myna settings** from Activities and follow its setup. This is where the preview guides you through the required model and dictation configuration. Canonical lists Parakeet, Whisper and FunASR as supported recognition engines, with differing language, accuracy and hardware trade-offs. It does not publish one universal best choice in the preview announcement. A supported NVIDIA GPU may offer a GPU-accelerated engine; additional accelerators are planned. 4. Place the cursor in an ordinary text field, press the hotkey configured during setup and speak. Canonical says Myna gives visual feedback during recognition and inserts the transcription into the application in use. Check the inserted words and correct errors before sending or saving the text. The completion check is visible text in the field you selected. This is a check for a reader to perform, not a result BIG CHANGE observed. The preview post does not specify a universal hotkey, model download size, memory requirement, transcription accuracy or supported-app list, so the settings shown on your installed version are the authority for those choices. ## If the first attempt does not produce text Start with the setup path Canonical documents: confirm that you are on one of the listed Ubuntu versions and architectures in a GNOME Wayland session, reopen Myna settings and finish setup, including the model download. Then confirm a microphone is available, put the cursor in a normal text field, and use the hotkey shown in your settings. These are checks against the stated prerequisites and input flow, not a tested fault diagnosis or a guarantee that a particular fix will work. Try another ordinary text field if one application does not accept dictation. Canonical specifically asks preview users to report application compatibility, setup problems, responsiveness, language and accent behavior. If the issue persists, use the [Myna issue tracker](https://github.com/canonical/myna/issues) and include the Ubuntu version, desktop session, engine, application and the step that failed, without posting private dictated text. ## Privacy, human checks and limits Canonical says speech recognition runs on the machine, and that audio is neither uploaded nor stored. Model download and installation still need network access. Once Myna inserts text into an application, that application's own saving, syncing and sharing behavior applies to the text. Review dictated material, especially names, numbers and sensitive content, before using it. Myna is a dictation preview under active development. It is not a voice assistant and cannot control the desktop through spoken commands. Its engines vary in language support, accuracy and hardware demands; Canonical has not provided a general accuracy or speed guarantee in the preview post. The PPA installation instructions state no Myna subscription or usage charge, but the announcement gives no complete cost schedule or hardware/resource estimate. Users supply the computer, microphone, network connection for setup and local processing capacity. ## The big change Canonical has moved Myna from a project announcement to an installable, optional desktop dictation preview. On the supported Ubuntu GNOME Wayland systems, its documented flow sends microphone speech to local recognition and places the resulting text in the focused application. That gives users a direct way to try offline transcription after setup, while leaving accuracy, app compatibility and machine-specific performance to be assessed on their own systems. Ubuntu 26.10 arrived on October 15; Myna's technology preview was announced on October 12 and remains a preview across the listed releases. ## Sources & further reading - [Myna speech-to-text is now available as a technology preview](https://ubuntu.com/blog/myna-speech-to-text-technology-preview-available), Canonical, October 12, 2026. Primary source for supported systems, PPA commands, setup, local processing and preview limits. - [Canonical releases Ubuntu 26.10 Stonking Stingray](https://ubuntu.com/blog/canonical-releases-ubuntu-26-10-stonking-stingray), Canonical, October 15, 2026. Confirms the separate OS release date and Myna's opt-in status. - [Myna project README](https://github.com/canonical/myna), Canonical. Describes the PPA as the end-user installation route and warns that features may change before a stable release. ## Sources - [Myna speech-to-text is now available as a technology preview](https://ubuntu.com/blog/myna-speech-to-text-technology-preview-available) — Canonical's dated preview instructions specify supported systems, installation, setup, dictation, local recognition and limits. Product and privacy claims are Canonical's. - [Canonical releases Ubuntu 26.10 Stonking Stingray](https://ubuntu.com/blog/canonical-releases-ubuntu-26-10-stonking-stingray) — Confirms the separate Ubuntu release date and Myna's opt-in status. - [Myna project README](https://github.com/canonical/myna) — Canonical's end-user PPA route and development warning.