Question · Dictation in your apps
What is the best speech to text software for Linux?
There is no single best app, because your display session decides which typing method works. On Ubuntu 26.04 with GNOME on Wayland, Handy is the choice to start with, since its setup guide was tested on that configuration. On an X11 session, nerd-dictation is the command-line option to try, and Canonical's Myna is an opt-in option to test.
Updated October 17, 20264 min read8 cited facts, each traced to its source
Neither Ubuntu nor Linux Mint ships a finished dictation feature that we could verify, so the best speech to text software for Linux is an open-source app that you install. Which app is best depends on your display session. On Ubuntu 26.04 with GNOME on Wayland, start with Handy. On an X11 session, start with nerd-dictation.
On Ubuntu 26.04 with GNOME on Wayland, Handy has a tested setup guide
Handy is the app we would install first on Ubuntu 26.04 with GNOME on Wayland, because its maintainers wrote a setup guide for that exact configuration. The guide was tested on Ubuntu 26.04.1 LTS with GNOME on Wayland and Handy 0.9.8 1. A guide tested on your setup tells you more than a feature list does, so use it as your starting point.
The typing step is the part that depends on your session. The older typing tool, xdotool, does not work on Wayland, and its own README says it is unclear whether it ever could 2. Follow the guide's steps, because setting names change between releases. Our walkthrough of those steps is in the guide to speech to text on Linux.
On an X11 session, nerd-dictation is the command-line option to try
nerd-dictation is a command-line tool, and its default input method is xdotool 3. That default suits an X11 session. It does not suit Wayland, for the reason in the section above, so the session type decides the choice.
Check yours before you install anything. Run echo "$XDG_SESSION_TYPE" in a terminal. A result of x11 fits nerd-dictation. A result of wayland means you should choose Handy or another tool that has a Wayland typing route.
Accuracy comes from the engine, and Parakeet and Whisper both have published benchmarks
Accuracy depends mostly on the speech engine, not the app around it. On the Open ASR Leaderboard's English short-form tests, NVIDIA's Parakeet TDT 0.6B v3 had an average word error rate of 6.32 percent 4. OpenAI's Whisper Large v3 had 7.44 percent in the same table 5. Both figures come from that one table, but the gap between them is small, so read it as rough.
These are benchmark results on test audio. Your microphone, your room and your accent can produce a different error rate, and the leaderboard does not test your setup. Word error rate explains how the figure is calculated, so you can measure your own result.
If your computer has no graphics card, the engine choice still matters. Handy offers a Parakeet V3 model that its README describes as CPU-optimized 6. Whisper also runs without a GPU, and the question on running Whisper without a GPU covers the speed trade-off.
Canonical's Myna is worth watching, but it is opt-in for now
Canonical, the company behind Ubuntu, is developing Myna, an app for voice typing. A September 2026 how-to guide says Myna is an opt-in feature in Ubuntu 26.10, and that the current Ubuntu 26.04 can also try it out 7. Linuxiac reports that the initial release targets Ubuntu Desktop on Wayland, with GNOME 8.
Treat Myna as untested for your work. Because the how-to describes it as opt-in, it is not on by default. If you want to try it, install it on a spare machine first, and do not make it your only dictation tool yet.
The right tool depends on your setup
| Your setup | Start with | Why |
|---|---|---|
| Ubuntu 26.04, GNOME, Wayland | Handy | Its setup guide was tested on this configuration |
| Any X11 session, you are comfortable in a terminal | nerd-dictation | Its default typing method is xdotool, which fits X11 |
| Ubuntu 26.04 or 26.10, and you want to test Canonical's option | Myna | Opt-in in 26.10, per a September 2026 how-to guide |
| You want a desktop app that runs Whisper on your computer | Vocalinux | Covered in the question on Whisper dictation on Linux |
The table is a starting point, not a ranking. We did not run every tool on a test machine. Try your choice on a short note before you rely on it.
What to do next
- Check your session type. Run
echo "$XDG_SESSION_TYPE"in a terminal. The answer isx11orwayland. - Pick a tool from the table that matches that answer, and follow its install steps from its own documentation.
- Test in two stages. Dictate one sentence into a text field and check that the words appear in the right window. Then dictate two minutes of ordinary speech and read the result before you send it anywhere.
- Re-test after upgrades. Session and input rules can change between Ubuntu or Mint releases, so repeat the test after each upgrade.
Two related questions may help. The question on controlling Linux by voice covers voice commands, which are a different job from dictation. The question on Dragon on Linux explains why we did not include Dragon.
Yap, which we make, runs on Mac, Windows and phones, not on Linux, so every tool on this page is made by someone else. If you only need dictation in a web browser, the free dictation page works without installing anything.
Evidence
Every number on this page, with where it comes from. Open a fact to see the exact words in its source.
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
Go deeper
Related questions
- Can you control Linux with voice commands?
- Does Dragon NaturallySpeaking work on Linux?
- How do you use Whisper for dictation on Linux?
- Can Chrome do speech to text offline?
- Can Google Docs transcribe an audio file?
- Can Microsoft Word transcribe an audio file?
- Can OneNote transcribe audio?
- Can Whisper run in the browser?