How to transcribe audio on Mac free

The easiest free way to transcribe audio on a Mac is MacWhisper — a free app that runs OpenAI’s Whisper model locally on your machine. Drag in any audio or video file and accurate text appears in seconds, with no cloud upload, no account, and no subscription. If you recorded in Voice Memos on an Apple Silicon Mac, macOS can even transcribe it automatically with nothing installed at all.

That’s the short answer. The longer answer is that there are actually several genuinely free ways to turn audio into text on a Mac in 2026 — and which one you should use depends on what kind of audio you’ve got, how old your Mac is, and how much you care about privacy. Let’s walk through all of them, from “zero effort” to “slightly nerdy but powerful.”

Method 1: MacWhisper (The One to Use)

Back in 2022, OpenAI released Whisper — a shockingly accurate speech-to-text model — completely free. The catch was you needed to be comfortable with the Terminal to use it. MacWhisper fixed that: it wraps Whisper in a friendly Mac app where the entire workflow is “drag file in, get text out.”

Here’s the whole process:

Step 1: Download MacWhisper from the developer’s official site (it’s by indie dev Jordi Bruin — avoid lookalike download sites).

Step 2: Open the app and pick a Whisper model to download. Smaller models are faster, larger ones are more accurate. The medium options are the sweet spot for most people.

Step 3: Drag in your audio or video file — MP3, M4A from Voice Memos, WAV, video files, whatever. You can even paste a YouTube URL and transcribe the video directly.

Step 4: Hit transcribe. On an Apple Silicon Mac, text starts appearing almost immediately. Copy it out, or export as a text file or even subtitles.

Why this is our top pick: everything happens on your Mac. Your audio never touches the internet, which matters a lot if you’re transcribing meetings, interviews, or anything sensitive. It supports 100+ languages. And the free version covers the core experience — the paid Pro upgrade (a one-time purchase, refreshingly, not a subscription) adds power features like batch transcription, speaker labels, and recording system audio straight from Zoom or Teams calls. For occasional transcription, you may never need it.

Method 2: Voice Memos’ Built-In Transcription

Here’s the option a surprising number of Mac users don’t know they already have: macOS Sequoia added automatic transcription to the Voice Memos app. Record something (or sync a recording from your iPhone), open it in Voice Memos on your Mac, and there’s a transcript waiting for you. On-device, free, zero setup.

The fine print, though, is real:

It needs Apple Silicon. M1 or newer — Intel Macs don’t get the feature.

It only works on Voice Memos recordings. You can’t feed it a random MP3 from Finder or that podcast episode you downloaded. Recordings made in the app, that’s it.

Language support is limited compared to Whisper’s 100+, and export is basically “select the text and copy it” — there’s no one-click save-as-document.

So: if your audio was recorded in Voice Memos and you’re on a recent Mac, this is genuinely the fastest free path — nothing to install. The moment your needs go beyond that, jump back to MacWhisper. (This is very on-brand for Apple Intelligence-era features, as we noted in our roundup of the best AI apps for your Mac: convenient seasoning, not the full meal.)

Method 3: Live Captions for Real-Time Audio

Slightly different problem, same “it’s already on your Mac” energy: Live Captions is an accessibility feature (Apple Silicon Macs) that generates real-time captions for any audio playing on your system — a video call, a YouTube video, a livestream, whatever. Turn it on in System Settings under Accessibility.

It’s designed for accessibility rather than producing transcripts, so you can’t neatly export the text. But if your actual goal is “follow along with audio as text right now” rather than “get a document afterward,” it does the job for free and entirely on-device. Handy trick to have in your back pocket.

Method 4: Whisper in the Terminal (For the Brave)

If you liked the local-AI section of our Mac apps guide, here’s the DIY version: whisper.cpp, the open-source project that makes Whisper run fast on Apple Silicon. Install it via Homebrew, download a model, and transcribe from the command line. Completely free, no app, infinitely scriptable.

Why bother when MacWhisper exists? Automation, mostly. If you want to auto-transcribe every file dropped into a folder, wire transcription into a script, or add speaker-labeling tools on top, the command line is where that flexibility lives. For everyone else, this is strictly a “because it’s cool” option — and there’s zero shame in using the app instead. The app is literally this, with a nicer shirt on.

What About Free Online Transcription Sites?

Google “free transcription” and you’ll drown in web tools promising instant results. Here’s the honest assessment: most give you a tiny free allowance — often 10 minutes a month or a per-file cap — before the subscription pitch arrives. And every single one requires uploading your audio to someone else’s servers, which is exactly the thing the local options above let you avoid.

They’re not scams, mostly. Some are quite good. But when your Mac can run the same class of AI model locally, for free, with no limits and no privacy trade-off, uploading your audio to a stranger’s website to get 10 free minutes is a bad deal. The one scenario where web tools earn their keep is the next section.

A Warning for Intel Mac Owners

All the local-transcription magic in this article leans hard on Apple Silicon. On an M-series Mac, transcribing an hour of audio takes minutes. On an older Intel Mac, that same hour can take somewhere between 30 minutes and forever, because there’s no Neural Engine doing the heavy lifting — and the built-in Voice Memos transcription and Live Captions aren’t available at all.

MacWhisper still works on Intel — just slowly, and you’ll want the smaller models. If you’re on an older machine and transcribing hours of audio every week, this is the one case where a cloud transcription service’s speed might genuinely be worth paying for. Or, you know, it’s one more argument for that Apple Silicon upgrade you’ve been eyeing.

Getting Better Transcripts: Quick Tips

Pick the right model size. In MacWhisper, bigger models are noticeably more accurate on messy audio — accents, crosstalk, background noise. If a transcript comes back rough, re-run it with a larger model before blaming the recording.

Garbage in, garbage out. A phone in the middle of a conference table produces worse transcripts than any model can fix. Get the mic close to the speaker.

Use exports smartly. MacWhisper can export SRT subtitle files — which means “transcribe audio” quietly doubles as “free subtitles for my videos.” Underrated.

Clean up with a chatbot. Raw transcripts are full of ums and rambles. Paste one into ChatGPT or Claude with “clean this up and summarize it” and you’ve built yourself a free meeting-notes pipeline out of two free tools. (Both are covered in our Mac AI apps guide.)

Which Method Should You Use?

Ninety percent of you: download MacWhisper, grab a medium model, drag your file in, and get on with your day. It’s the rare free tool with no meaningful catch — local, private, accurate, and unlimited.

If your recording already lives in Voice Memos on an M-series Mac, skip the download and use the built-in transcript. If you need text from live audio as it plays, flip on Live Captions. And if you’re the type who has Homebrew installed and strong opinions about it, whisper.cpp will happily eat a folder of podcast episodes overnight.

The bigger picture is the same one we keep running into on this site: Apple Silicon plus open AI models means your Mac can now do locally — for free — what cost a per-minute subscription three years ago. Transcription just happens to be the category where that’s most obvious.

Want more of this? Check out our full guide to the best AI apps for your Mac, or the iPhone side of things in the best AI apps for iPhone.