Speech-to-text for Obsidian: dictate notes 5× faster
Dictate directly into Obsidian on Mac. SpeechFlow inserts clean, punctuated text at your cursor in any note — no plugin, no cloud storage. Free to start.
Obsidian is where serious thinkers build their second brain — daily notes, zettelkasten slips, literature reviews, half-formed ideas that need to land now. But the keyboard is the bottleneck. You think at 150-180 words per minute and type at 40-60. SpeechFlow closes that gap: hold Control, speak, release — clean punctuated text appears at your cursor in any Obsidian note. No plugin, no paste, nothing stored.
The problem: capturing fleeting notes before they vanish
Obsidian's power is the speed at which you can connect ideas — but every second spent typing is a second the thought is drifting. Daily notes pile up with half-finished bullets. Zettel drafts sit shallow because committing to long-form prose feels heavy. Literature notes end up as copy-pasted quotes instead of your own synthesis, because writing it out takes too long.
Apple's built-in dictation can place raw text in Obsidian, but it leaves every “um”, every false start, and zero punctuation. You end up editing more than you dictated. That's not a speed-up, it's a transfer of effort.
How SpeechFlow works in Obsidian
SpeechFlow is a native macOS app (Apple Silicon, ~50 MB). It operates at the system cursor level, so it slots into Obsidian — desktop app or any note — without any plugin or vault configuration. The flow is three steps:
- Click into the Obsidian note where you want to write.
- Hold Control and speak naturally — full sentences, tangents, all of it.
- Release. A cleanup LLM strips fillers, adds punctuation, adapts tone, and inserts the finished text exactly where your cursor was.
It works in the editor, in callouts, in canvas cards, and in the quick-capture modal. Markdown formatting you speak (“heading level two”, “bullet point”) is not interpreted — the output is clean prose you then structure yourself, which is exactly how most Obsidian workflows prefer it. For a broader look at how the underlying technology works, see the AI dictation guide.
What to dictate in your Obsidian vault
| Note type | What to dictate |
|---|---|
| Daily note | Morning intentions, end-of-day review, anything fleeting — talk for 60 seconds instead of staring at a blank daily template. |
| Zettel / atomic note | Speak the core idea and one or two supporting observations. Faster to link later than to never write it down. |
| Literature note | Summarise what you just read in your own words while the book is still open — SpeechFlow cleans the summary instantly. |
| Journaling | Brain-dump the day in plain speech; get a readable paragraph without editing a single word. |
| Project / meeting notes | Recap a meeting or planning session live, then structure the blocks at your own pace. |
Why SpeechFlow fits the Obsidian ethos
Obsidian users chose a local, plain-text, privacy-first vault for a reason. SpeechFlow is built around the same values:
- No cloud lock-in — SpeechFlow is a native app, not a browser extension or SaaS widget. Your vault never moves.
- Zero data retention — nothing is stored on a SpeechFlow server. In BYOK mode, your audio goes straight from your microphone to the AI provider you chose (OpenAI, Gemini or Groq) and nowhere else.
- Works offline for capture — hold and speak even when you have no internet; the cleanup step fires when you release and a connection is available.
- Plain text output — SpeechFlow inserts clean prose, so Obsidian's markdown remains yours to shape. No proprietary format, no hidden metadata.
If you use other tools alongside Obsidian, the same shortcut works everywhere — see how it fits Notion or how writers use it to write a book by voice on Mac.
FAQ
Does SpeechFlow require an Obsidian plugin or community plugin?
No. It inserts text at the macOS system cursor, so Obsidian receives it as normal keystrokes. Nothing to install inside your vault.
Will it store my notes or voice recordings?
No. SpeechFlow has zero data retention. In BYOK mode your audio goes directly to your chosen provider (OpenAI, Gemini or Groq) — no SpeechFlow server ever touches it.
Is there a free plan?
Yes — 2,500 words per week with no credit card required. Pro is €10/month or €70/year; BYOK is a one-time €69 for lifetime access.
Does the cleanup LLM add markdown syntax automatically?
No. The output is clean prose. If you want a heading or a list, you add the markdown yourself — which keeps your Obsidian formatting style fully in your hands.
Does it work with Obsidian Sync or local vaults?
Yes. SpeechFlow doesn't touch your vault files directly — it types at the cursor, so it works identically whether your vault is local, on iCloud, or synced with Obsidian Sync.
Start capturing ideas at the speed you think. Try SpeechFlow free — 2,500 words a week, no card needed.