Resonant + Google Docs
Rough drafts,
spoken straight into Docs.
Google Docs has a built-in voice typing feature. It's famously clunky — Chrome only, cloud-only, tripped up by any technical term, no cleanup of filler words. Resonant is the local replacement: faster, cleaner, and it works with any browser.
Place your cursor, press a key, speak the paragraph. Clean text lands in the doc. No audio leaves your Mac.
Used by professionals at
Typed vs. dictated
The rough draft,
already on the page.
Proposal first draft
“intro paragraph about why this matters”
“The core problem the foundation has been trying to solve for the last six years is that early-stage climate adaptation projects in coastal regions almost never reach the phase where they can attract institutional capital. The gap isn't technical and it isn't political — it's a 12 to 18 month window between proof of concept and bankable pilot where no existing funding vehicle fits. This proposal describes a bridge facility designed specifically for that window, and why we believe this particular foundation is uniquely positioned to anchor it.”
Meeting notes, in the doc
“david - said we need to push deadline. agreed??”
“David raised the timeline concern again. His view is that if we hold the current April 30 deadline, the engineering team will ship something that passes QA but doesn't actually solve the customer's workflow problem, which is the whole reason we took the project on. He proposed pushing internal delivery to May 14 in exchange for two additional rounds of user testing. Priya agreed in principle, Mark wanted to check with the account team first. Action: Priya to confirm with the account team by EOD Thursday and update the shared timeline.”
Section with technical vocabulary
“section on the auth flow, kinda complicated”
“The authentication flow uses a short-lived JWT issued by our edge worker after the initial OAuth handshake with the identity provider. The JWT carries the subject, a rotating session ID, and the subscription tier, and it's signed with an asymmetric key whose public half is distributed through a JWKS endpoint. Downstream services — the Convex backend in particular — validate the token by fetching JWKS, which means we can rotate signing keys without redeploying the services themselves. This section documents the flow end-to-end so a new engineer can trace a request from the browser to the database.”
The insight
Staring at a blank page
isn't writing.
Talking is.
The hardest part of any document is the first pass. You know roughly what you want to say — you've been turning it over in your head all morning — but the moment the cursor blinks, everything seizes up. You type a sentence, delete it, type half of another, tab over to check something, come back, type nothing for a minute.
The truth is that your brain already has the draft. It's just that your fingers are a terrible bottleneck for getting it out. Voice closes the gap. You speak the thing the way you'd say it to a colleague, clean text lands in the doc, and now you have something to edit instead of something to start.
Google Docs' own voice typing understood this — but only halfway. It's been stuck at the 2016 level of accuracy, locked to Chrome, and it treats your audio as something to upload. A local alternative with modern neural models is a different kind of tool entirely.
Chrome only. Audio streams to Google's servers. Any technical term, proper noun, or number gets mangled. No removal of filler words. Stops working if the tab loses focus. Not available inside a Docs cursor on mobile.
Usable once, then you quietly go back to typing.
Works in every browser and every app. Local transcription with current neural models — accurate on jargon, names, and numbers. Filler words and restarts cleaned automatically. The clean text lands in the Docs cursor as plain keystrokes.
The feature Google Docs should have had by now.
Where it fits
Six moments inside
a Google Doc.
Rough drafts from scratch
The blank page is the hardest part. Speaking a first draft gets the thinking out of your head and onto the doc, where you can actually see it. Editing a rough draft is ten times easier than writing one.
Proposal and brief writing
Proposals reward a narrative voice and concrete specifics. Typing pushes you toward bullet points and hedged language. Speaking gives you the full argument in one take — exactly as you'd pitch it in a room.
Meeting notes in real time
Capture decisions, positions, and action items while the meeting is still happening. Voice is fast enough to keep up without making you a stenographer. The doc is complete before the call ends.
Long-form documentation
Technical docs, onboarding guides, internal wikis. The kind of writing where you know what to say but typing it out feels like a tax. Voice turns a two-hour doc into a 20-minute one.
Comment replies and reviews
Left-margin comment threads in a collaborative doc are their own kind of email. Speaking replies lets you give substantive feedback without breaking the editing flow of the person you're reviewing for.
Outlines that become drafts
Speak the outline, then speak the paragraphs under each heading. What would have been a week of staring becomes an afternoon of talking. The clean text is already in the doc, ready to edit.
Architecture
Local transcription.
Even Google doesn't hear it.
Google Docs' native voice typing sends your audio to Google's servers. That's a different posture than typing, where only the final text is transmitted. For unpublished proposals, client strategy, research notes, or any document under embargo, that difference matters.
Resonant runs modern speech models entirely on your Mac. No audio ever leaves the device — not to us, not to Google, not to anyone. The transcription happens in real time on your hardware and the clean text is what lands in your doc.
Google Docs still syncs the final text, same as ever. Your collaborators see the same document they'd see if you had typed it. The only difference is how fast you got it there and what happened to your voice along the way.
Free. Local. Works in every browser.
Get the draft out of
your head. Today.
Speak directly into the Docs cursor. Clean text, cleaner drafts, no cloud audio.
Requires macOS 14+ · Apple Silicon
Related