Voice dictation, finally done right on Linux.
Wispr Flow and Superwhisper don't run on Linux. PromptStream does, natively on Wayland and X11. Hold a key, speak, release: about a second later clean text lands wherever your cursor is. Speaking is 3 to 4 times faster than typing.
Built for Linux, not ported to it.
Most dictation apps skip Linux entirely, and the open-source scripts that exist tend to break on Wayland. PromptStream is engineered around how modern Linux desktops actually work.
Native on Wayland
Text is typed through the desktop's RemoteDesktop portal, the one route that reaches native Wayland windows. No xdotool hacks, no root daemons, no switching back to X11.
Real hold-to-talk
Hold a key, talk, let go. The hotkey works the same on Wayland and X11, and defaults to Right Alt so GNOME's Super-key overview never gets in the way. Remap it to anything.
GNOME and KDE Plasma
Allow typing once when your desktop asks, and PromptStream remembers it across restarts. Lives in your tray where your desktop provides one.
Plays nice with PipeWire
Music and videos duck automatically while you speak, on PipeWire and PulseAudio, and come back the moment you let go.
One AppImage, any distro
Ubuntu, Fedora and most other distros: download, chmod +x, run. No dependencies to chase, no Python environment to babysit.
No GPU, no model downloads
Local Whisper setups want gigabytes of models and a decent GPU. PromptStream returns clean text in about a second on any laptop.
The dictation apps Linux never got.
The two best-known dictation apps leave Linux users out. PromptStream fills that gap.
| PromptStream | Wispr Flow | Superwhisper | |
|---|---|---|---|
| Linux | Yes | No | No |
| Wayland | Yes, native | No | No |
| macOS | Yes | Yes | Yes |
| Windows | Coming soon | Yes | Yes |
| Pro plan | $7/month | $15/month | $8.49/month |
Platform support per each vendor's own documentation, September 2026.
Everything in one quiet window.
Pick your mic, choose a hotkey, flip on AI cleanup. PromptStream sits in your tray or menu bar and stays out of the way until you hold to talk.
How it works
Hold the key
Hold Right Alt on Linux (Right ⌘ on Mac), or remap it to any key or shortcut. A small pill appears with a live waveform, so you know it's listening.
Say it like you'd say it
Ums, false starts, changing your mind mid-sentence — all fine. Speak naturally; PromptStream handles the mess.
Release. It's typed.
About a second later, clean text is pasted right where your cursor is: Slack, Gmail, Notion, VS Code, your AI chat, anywhere.
Long dictation? Tap the hotkey instead of holding to lock recording on. Esc cancels anytime.
Everything you need. Nothing you don't.
Cleanup that respects you
Filler words, false starts, and self-corrections disappear. Punctuation and capitalization are fixed. Your meaning and tone are never rewritten.
Works in every app
The text lands wherever your cursor is: Slack, Gmail, Notion, VS Code, Obsidian, ChatGPT and Claude in your browser. No copy-paste dance, no switching windows.
Fast, and it never loses your words
Powered by Groq inference — transcription typically returns in ~1 second. And if anything goes wrong mid-flight, your raw words are kept in history and on your clipboard. Dictation never fails silently.
13 languages, auto-detected
English, French, Arabic, German, Spanish, Italian, Portuguese, Dutch, Turkish, Hindi, Japanese, Korean, and Chinese. If typing in English is slower than speaking it, PromptStream closes the gap.
Personal dictionary
Teach PromptStream the names, brands, and jargon it must get right. Once learned, always spelled correctly.
Searchable history
Everything you dictate is saved locally and searchable, with one-click copy. It never leaves your machine.
Your setup, your rules
Pick your microphone, remap the hotkey to any key, and watch the live waveform confirm every word is being heard.
Easy on your wrists
If typing hurts — RSI, wrist pain, long days at the keyboard — dictation is real relief. Write your emails, messages, and docs without touching the keys.
Quiet about your words, too.
Start in 30 seconds
A free account and you're dictating — no credit card, no API keys, nothing to configure.
History stays on your machine
Your dictation history is stored locally. We can't read it; nobody else can either.
Audio is never stored
Your audio is processed transiently for transcription and immediately discarded. It's never saved, anywhere.
Simple pricing
Monthly
- ✓Unlimited dictation
- ✓AI cleanup on every transcript
- ✓Personal dictionary and local history
- ✓Priority support
Yearly
Save 30%Works out to under $5 a month
- ✓Unlimited dictation
- ✓AI cleanup on every transcript
- ✓Personal dictionary and local history
- ✓Priority support
Wispr Flow is $15/mo. superwhisper is $8.49/mo. PromptStream is $7 — and $59 a year works out to under $5 a month.
Questions
Is PromptStream a Wispr Flow alternative for Linux?
Yes. Wispr Flow and Superwhisper don’t support Linux. PromptStream brings the same hold-a-key, speak-anywhere dictation to Linux, natively on Wayland and X11. Pro is $7/month or $59/year, versus Wispr Flow’s $15/month or $144/year.
Does it work on Wayland?
Yes. PromptStream types through the desktop’s RemoteDesktop portal, which GNOME and KDE Plasma implement, so text reaches native Wayland apps and not just XWayland ones. The first time, your desktop asks you to allow it, and the choice is remembered. X11 sessions work too.
Which Linux distros and desktops are supported?
Any x64 distro that runs an AppImage, including Ubuntu and Fedora, on GNOME or KDE Plasma. On desktops without the RemoteDesktop portal (Sway, Hyprland, XFCE, Cinnamon), your text lands on the clipboard ready to paste with Ctrl+V, so nothing is lost.
Why does the Linux app need keyboard access?
Wayland gives apps no way to see a held key, so PromptStream reads the keyboard directly to detect your hotkey. You grant this once with a udev rule (or by joining the input group). Keystrokes are only compared against your hotkey. They are never stored or sent anywhere.
Which platforms does it run on?
Linux x64 via AppImage, on Wayland or X11, and macOS 13 or later (Apple Silicon only). Windows is not supported yet.
Can I change the hotkey?
Yes. Right Alt on Linux and Right ⌘ on macOS are just the defaults. Set any key or shortcut you like. Hold it to dictate, tap it to lock recording on, Esc to cancel.
Does PromptStream need an internet connection?
Yes. Your audio is sent for transcription, processed transiently, and never stored. That is why it needs no GPU and no model downloads, and why text comes back in about a second.
Where does my audio go?
Audio is processed transiently for transcription and immediately discarded. It is never stored. Your dictation history stays on your machine.
Does it work in every app?
Anywhere you can place a text cursor and paste with Ctrl+V (⌘V on Mac): Slack, Gmail, Notion, VS Code, Obsidian, AI chats, form fields in your browser. Linux terminals that paste with Ctrl+Shift+V aren’t supported yet; there the text stays on your clipboard.
Is there a free plan?
Yes: 20,000 words a week, free forever, with nothing to configure. Need more? Pro is $7/month or $59/year for unlimited dictation.
Which languages are supported?
13 languages with auto-detect: English, French, Arabic, German, Spanish, Italian, Portuguese, Dutch, Turkish, Hindi, Japanese, Korean, and Chinese.
What about Windows?
Windows is coming soon; we are focused on Linux and macOS Apple Silicon for now. Email support@promptstream.app and we’ll tell you the moment it ships.
Say it once. It's typed.
Free plan included. Dictating on Linux within a minute of downloading.
Linux AppImage (x64) · Wayland & X11 · macOS 13+ (Apple Silicon only)