Voice Dictation for Obsidian
Daily notes, literature notes, the thought you had walking to the kettle. Speak into the open note with a global hotkey — no plugin, no community-plugin trust decision, no API token. Your vault stays exactly where it is.
No plugin in your vault
Nothing is added to .obsidian/plugins, nothing runs inside Obsidian, and nothing can read your notes. The client types into the focused field like a keyboard would.
Plain text, which is the point
You get punctuated prose in your Markdown file. Links, tags and headings stay yours to type — the syntax is faster written than spoken.
Linux, macOS and Windows
The same client and the same account on all three. Most dictation tools in this category skip Linux entirely, which is where a lot of Obsidian vaults live.
How it works with Obsidian
Install the desktop client
A native binary for Windows, macOS or Linux, linked to your account once with a device code. Obsidian is not modified and does not need to know it exists.
Open the note and place the cursor
Today’s daily note, a literature note, a canvas card, the quick switcher. Templater, Dataview and everything else in your vault behave exactly as before.
Hold your hotkey and think out loud
Punctuated text arrives in the note as you speak. It is written to the file by Obsidian, the same as anything you type — there is no separate sync path.
Where the audio goes, said plainly
If you chose Obsidian partly because your notes are local files that no company can read, you should know exactly what this changes. Your vault stays local and we never see it. The client has no access to your notes, your file system or your sync.
What does leave the machine is the audio you dictate, while you are dictating it. It is streamed to our servers, transcribed, and the text comes back. That is how the recognition works and there is no local model option — being straightforward about it matters more than the sale. If it is a dealbreaker, it is a real one, and Superwhisper runs Whisper entirely on your own Mac. We would rather point you there than have you find out later.
For everyone else the trade is a familiar one: server-side recognition is faster, does not heat your laptop, does not need a multi-gigabyte model download per machine, and works the same on a thin Linux laptop as on an M4.
What it does not do
- Audio is transcribed on our servers, not on your machine. Local-only transcription is not an option we offer, and for some Obsidian users that is decisive.
- It does not create or link notes. No [[wikilinks]], no tags, no templates and no graph edits — it types the words you say into the note you have open.
- Markdown syntax is not spoken. Saying "hash hash heading" types those words. Headings, links and callouts stay keyboard work, which is faster anyway.
- There is no mobile client, so dictating into Obsidian on a phone is outside what this does. Obsidian Mobile’s own keyboard dictation covers that case.
Common questions
Does it install a plugin into my vault?+
No. Nothing is written into .obsidian, no community plugin is involved, and there is no plugin to trust or audit. The desktop client runs outside Obsidian and types into whatever text field has focus, which is why it also works in your browser, terminal and email client.
Do my notes leave my machine?+
Your notes never leave your machine — the client cannot read them. The audio you dictate does leave: it is streamed to our servers for transcription and the text is streamed back. If you need transcription that runs entirely locally, Superwhisper does that on macOS and is the honest recommendation.
Can it write wikilinks and tags for me?+
No. It produces punctuated plain prose. Links, tags, headings and callouts are still typed, which is genuinely faster than saying "open square bracket open square bracket" out loud.
Does it work with Obsidian Sync or a git vault?+
Yes, because it changes nothing about how your files are written. Obsidian saves the text exactly as if you had typed it, and whatever sync you use picks it up in the normal way.
Does it work on Linux?+
Yes, there is a native Linux client. That is unusual here: Wispr Flow, Superwhisper and MacWhisper have no Linux support at all, and a large share of Obsidian users are on Linux.
On a Linux desktop, see the Linux client. Weighing local against server transcription? Superwhisper vs Wispr Flow is that argument without us in it.
Last reviewed . Competitor pricing and platform support change; check theirs before buying.