Talk. It types.
Hold one key and say what you mean. Finished writing appears in whatever app you are already in, while you are still talking.
Free for 14 days, no card. $5/month after.
Hold the button below and watch it write this for you.
The part nobody else does
It writes while you are still talking
Every other dictation tool records you, waits for you to finish, and then makes you wait again. Yours is already on the page.
Nothing appears until you stop. Then you wait again for the whole thing at once.
Words land in the document as they are said. Letting go only tidies the last few.
A word locks in about two seconds after you say it. You watch the sentence build, so you catch a wrong name while you are still in the sentence, not after you have pasted it.
Why it beats plain dictation
The same sentence, written five different ways
Every other dictation app gives you back what you said. YapperTalk gives you back what you meant, in the shape you needed it. Pick a mode and watch the same mumble land differently.
“tell mike cant make it tuesday, maybe thursday, ask him”
Said out loud, at talking speed, with the ums and the backtracking left in.
Hi Mike, I can't make Tuesday. Would Thursday work for you?
A rambled gist becomes a message you can send as it stands.
A third of the price, because we own the engine
The well-known dictation apps rent a speech API and pass you the bill. We run our own model on our own servers, so the low price is structural rather than a launch discount.
Your voice goes to our engine, and stops there
Audio is transcribed on hardware we run, using weights we own. It is held in memory for the length of the request, never written to disk, and never used to train anything. We do not rent a speech API by the minute, which is also why the price below is what it is.
Rewriting works on text, never on your audio
Straight Transcription never leaves our engine at all. Ask for a summary or a polish and that transcript, never the recording, goes to a named processor under contract terms forbidding retention and training. Every one of them is listed in the privacy policy, because a marketing page and a legal page that disagree is worse than either being blunt.
It learns how you actually work
The mode you pick in a given app becomes the mode it uses there. Names, product words and acronyms you correct once stop coming back wrong. Voice snippets expand a phrase into a paragraph you use constantly.
One key, everywhere, no integrations
As far as macOS is concerned this is a very fast keyboard, so it works in any app with a cursor. Nothing to connect, no per-app plugin, no list of supported tools to check before you buy.
About sixty seconds
Getting it is the easy part
Download it
One file. No account, no email, no waiting for a link. It is free to download and free to try.
Double-click the installer
It sets itself up and asks for the microphone once. There is no multi-gigabyte model to wait for, because transcription happens on our engine.
Hold Right ⌘ and talk
In any app, anywhere. A microphone appears in your menu bar and that is the last piece of setup you will ever see.
Free for 14 days, no card. $5/month after, renewing until you cancel. macOS 13 or newer.
Windows is in progress and not released yet.
It types into everything
There is no integration list, because there are no integrations. Your Mac thinks it is a keyboard, so it works wherever a cursor blinks.
You think faster than you type
Not a productivity claim, just arithmetic. Most people type around 45 words a minute and talk at more than four times that.
A 2,000 word day is about 44 minutes of typing, or about 10 minutes of talking. The rest of that hour goes back in your pocket.
The rest of it
Where does my voice actually go?+
To our own speech engine, running on hardware we rent and weights we own. The audio is held in memory for the length of the request, turned into text, and dropped. It is never written to disk on our side, never kept, and never used to train a model.
Does anything get sent to an AI company?+
Transcripts deserve a precise answer. Transcription itself happens entirely on our engine. If you invoke a rewrite mode, that text goes to a named processor under zero-retention, no-training contract terms. Your audio never does, and Straight Transcription does not use that path at all. Every processor is listed in our Privacy Policy.
How is this different from the well-known dictation apps?+
Two things. It types into your app while you are still speaking instead of pasting a block when you stop, and it has five rewrite modes rather than one, including briefs written specifically for AI coding agents. It also costs $5 instead of $15, because we run our own engine rather than renting one by the minute.
What do I need to run it?+
A Mac on macOS 13 or newer and an internet connection, since transcription happens on our engine rather than on your machine. Setup takes about a minute and there is no large model to download. Windows is in progress and is not released yet.
What does it cost, and how does the trial work?+
14 days free with no card, then $5 a month for unlimited dictation. The founding price stays at $5 for as long as you stay subscribed.
Does it handle names and jargon it has never heard?+
Correct a word once and it stops coming back wrong. There is a custom dictionary for product names and acronyms, and voice snippets that expand a short phrase into a paragraph you use often.
Can it edit text I have already written?+
Select the text, hold Option and Right Command, and say the change you want. The edit replaces the selection in place.
Rolling it out to a team?
Central billing, seats managed in one place, and a real person to talk to during the rollout. Tell us the seat count and we will come back with a number.
Security review ahead of you? Our data processing addendum is published in full, subprocessors and all, and we sign it on request.
Stop typing what you could
have just said
↓Download for MacFree for 14 days, no card. $5/month after, renewing until you cancel.
Cancel anytime from your account page, in two clicks. See the Terms and Privacy Policy.