Free audio to text on a Mac.
Free means free here: no word caps, no watermarked exports, no upload-then-pay. Four routes, ranked by how much effort each asks of you, and an honest note on where free stops being enough.
Four genuinely free ways to convert audio to text on a Mac: Apple’s Voice Memos transcribes its own recordings on macOS 15 with Apple silicon; MacWhisper’s free tier transcribes common audio and video formats locally; Spokenly’s free tier does unlimited local transcription and offline dictation; and whisper.cpp is MIT-licensed and runs Whisper models on the command line, subtitles included. All four process audio on the Mac, so nothing is uploaded and there is no usage meter.
Last verified 2026-07-24
What free usually means online
Search this and most results are web converters where free means a trial: a word cap, a length limit, an export you can’t use, or a sign-up before you see the text. They also want the upload, which for an interview or a client call is the part worth avoiding.
Everything below is free without that bargain, because the transcription happens on your Mac. The models that made this possible are open, and the hardware to run them is the machine you already own.
The four routes
Ranked by effort. All process audio on the Mac. None need an account, though MacWhisper’s free download goes through a zero-cost checkout that asks for an email.
| Route | Effort | Handles | Catch |
|---|---|---|---|
| Apple Voice Memos | None, it’s installed | Its own recordings | macOS 15 or later on Apple silicon, and not in every country |
| MacWhisper free tier | One download | MP3, WAV, M4A, OGG, OPUS, MOV, MP4 | Some models, including the fastest and most accurate, are Pro-only |
| Spokenly free tier | One download | Files, plus unlimited offline dictation | Cloud engines need your own API keys, but the local models are free |
| whisper.cpp | Build it yourself | FLAC, MP3, OGG, WAV, and SRT or VTT output | Command line, and you manage the model files |
If the audio is already in Voice Memos
Start here, because there is nothing to install. On macOS 15 or later with Apple silicon, the Voice Memos app can show a transcription of a recording directly. Apple’s own caveat: audio transcription isn’t available in all countries or regions. It won’t label speakers or export subtitles, and it only works on recordings inside the app, but for reading back what you said on a walk it’s the whole answer.
For any other file: the free app tiers
MacWhisper’s free tier transcribes the common audio and video formats using local models, and it’s the route most people should take: one download, drag the file in, read the text. Their own listing notes some models are fully available only to Pro users, so the free tier is a good model rather than their best one. Spokenly’s free tier is the other strong option, with unlimited local transcription and unlimited offline dictation included, which makes it the better pick if you also want to talk into text fields. Both keep the audio on your Mac. Neither meters you.
The DIY route, if you like a terminal
whisper.cpp is an MIT-licensed C and C++ implementation of Whisper whose README calls Apple Silicon a "first-class citizen", optimized via ARM NEON, the Accelerate framework, Metal, and Core ML. Free forever, no app involved.
- 01
Clone the repository and build it with cmake, following the README’s build steps.
- 02
Download a model file. The smaller ones are faster and less accurate; the larger ones are the reverse. This is the tradeoff you now own.
- 03
Point whisper-cli at your audio. It reads FLAC, MP3, OGG, and WAV directly; anything else converts in one ffmpeg command first.
- 04
Read the output, or ask for subtitles with the SRT and VTT output flags. It takes several files in one invocation, so a shell script covers a backlog.
Where free stops being enough
Free routes cover the transcription itself completely, and more of the surrounding work than paid vendors like to admit: whisper.cpp writes SRT and VTT subtitles and takes a whole list of files at once, and MacWhisper’s free tier exports subtitles too. The genuine gap is narrower. Speaker labels on a single-microphone recording is the clearest one: MacWhisper keeps automatic speaker recognition in Pro, Apple’s transcription doesn’t label speakers at all, and whisper.cpp’s diarize flag splits stereo channels rather than telling voices apart. After that it’s convenience: how much terminal you want in your day, and whether an AI agent can drive the whole job for you.
Edict is a one-time purchase with a 7-day trial of the full app, nothing held back and no card asked for. If the free routes above do what you need, they’re the right answer and this page would rather you use them. The tool-by-tool comparison sets out who wins what, and the how-to guide covers the mechanics whichever route you pick.
Common questions
What’s the best free audio to text converter for Mac?
MacWhisper’s free tier for general files, or Voice Memos if the recording is already in that app. Spokenly’s free tier is the pick if you also want free offline dictation. All three run locally, so there’s no upload and no usage limit.
Is there a free audio to text tool with no word limit?
Yes, and that’s the advantage of on-device tools: there’s no meter to hit because no service is counting. MacWhisper’s and Spokenly’s free tiers, and whisper.cpp, all transcribe as much audio as you have time for.
Can I convert audio to text for free without uploading it?
That’s exactly what every route on this page does. The transcription runs on your Mac, so the file never leaves it, which is what you want for an interview or a client call, where the recording belongs to more people than you.
Is free transcription accurate?
Accuracy comes from the model and your audio quality. Three of these routes run Whisper- and Parakeet-family models, the same lineage the paid tools and many cloud services use, and Apple runs its own. Price enters only where a free tier gates the better models, as MacWhisper’s does. Names and jargon trip every model, so skim the result either way.
What about free web transcription sites?
They work, and some are genuinely useful. The trade is that your audio goes to their servers, and the free tier usually caps minutes or export formats. If the recording is sensitive or long, the on-device routes above avoid both problems.
Apple Silicon (M1 or newer) · macOS 15 Sequoia or later · 7-day full trial · one-time purchase