Transcribe Audio Files on Your Mac: Utility and Statistics
YapToText transcribes any audio or video file on your Mac for free, offline, with no account and no upload. Drop the file on the Utility page and you get the text back. The speech model that ships inside the app, Whisper Large v3 Turbo, does the work on your Mac.
This page covers two parts of the app that are not about live dictation: Utility, a workbench for text jobs, and Statistics, which adds up what all the talking amounts to. Both run entirely on your Mac.
Transcribe an audio or video file
Drop an audio or video file on the Utility page and the app transcribes it. Drop in audio or video, get the text. Nothing is uploaded, because the transcription runs on your Mac.
Dictionary entries apply to every transcript automatically, so the names and terms you have taught the app come out right in a file transcript as well.
Once you have the text, you can run it through a mode in Utility the same way you would with pasted text. Raw Transcription gives you exactly what was said, with punctuation, and never touches the cleanup model. Clean Up removes filler words and false starts and fixes grammar. Note, Email, Message, and Code Comment each reformat the words for that kind of writing.
The menu bar quick panel can transcribe a file too. The same panel starts a dictation, switches mode or microphone, regenerates your last dictation as any mode, or inserts the last dictation again.
Dictate and transform inside the app
Utility is where text jobs happen inside the app. Dictate into a scratchpad instead of another app, paste anything and run it through a mode, drop an audio or video file to transcribe it, or run one of your AI Actions on a piece of text. The connectors between the boxes show how text flows from one stage to the next: dictate or paste, transform, then send onward.
Dictating into the scratchpad keeps the words inside the app instead of typing them at the cursor in whatever app is in front. It is the place to talk when the text is not headed anywhere yet.
Transforming means running text through a mode. Paste rough notes, pick Email, and the mode reformats them as an email. Every mode is editable, including the built-in ones, and you can create new modes with your own instructions, so the transform step does whatever you have told that mode to do.
What the Statistics page counts
The Statistics page charts your last 30 days, counts your words and your speaking time, and works out your real words per minute. It also estimates the time you saved compared to typing everything at 40 words a minute, and it keeps track of streaks.
Each bar in the chart is a day. Selecting a bar shows that day's exact count.
Pace is your real words per minute across what you have actually dictated, so it reflects how you talk, not a number you typed in.
Everything stays on your Mac
Every number on the Statistics page is computed on your Mac from your own history, at the moment you open the page. Nothing is tracked, and nothing leaves the machine.
File transcription works the same way. The audio is transcribed on your Mac, and the app makes no network calls unless you start a model download yourself. The details are on the privacy page, and the whole source is at github.com/ryleighnewman/YapToText.
Questions
Can YapToText transcribe a video file?
Yes. Drop in an audio or video file and you get the text. The transcription runs on your Mac.
Does transcribing a file need an internet connection?
No. The speech model ships inside the app, and the file is transcribed on your Mac. Nothing is uploaded.
Can a file transcript be cleaned up the way a dictation is?
Yes. Utility runs any text through a mode, so a transcript can go through Clean Up, Email, Note, or any mode you have made.
How does YapToText work out the time saved?
It compares the words you dictated against typing everything at 40 words a minute. The estimate is computed from your own history on your Mac.
Do my statistics get sent anywhere?
No. Every number is computed on your Mac from your own history when you open the page. Nothing is tracked and nothing leaves the machine.