MacWhisper Setup: Models, Free vs Pro and Recording App Audio

Get it from the official site

The developer's site is macwhisper.com, and it warns that fake MacWhisper sites exist and can carry malware. It says purchases go only through its Gumroad page or the App Store, so download the free version with the Download Free button in the pricing table there.

There are two Mac builds. The version comparison explains the difference:

• MacWhisper, the direct download, takes a one-time licence key. Only this build can detect meetings and offer to record them, dictate into any text field, and use every AI provider.
• Whisper Transcription, on the Mac App Store, shares the same core features and is sold by subscription. It has fewer AI providers and doesn't accept licence keys.

If you want meeting recording, get the direct download. The iPhone and iPad app, also called Whisper Transcription, runs all local models for free.

Free or Pro

Per the pricing table on macwhisper.com, the free version:

• transcribes MP3, WAV, M4A, OGG, OPUS, MOV and MP4 files in 100 languages
• records from your Mac's microphone
• exports plain text and SRT or VTT subtitles
• dictates into other apps

MacWhisper Pro costs €64 once, with lifetime updates, and adds:

• recording audio from other apps, and recording meetings in Zoom, Teams, Webex and similar apps with automatic start and end detection
• automatic speaker recognition, which the Gumroad page limits to M-series Macs for local models
• YouTube and other media links, and batch transcription of many files
• translation of transcripts and subtitles
• more export formats, including Markdown, PDF and DOCX
• more AI providers, such as Anthropic and Google, and local AI models through Ollama and LM Studio

The Gumroad page, where the licence is sold, lists a few more Pro features that the table leaves out: WhisperKit models, and real-time captions with live translation from the microphone or system audio. It also says the free version supports the Whisper models, but some of them are only fully available in Pro. It doesn't say which.

If you only need to turn a few recordings into text, the free version is enough to try first. Pay when you need app audio, automatic speaker recognition with local models on an M-series Mac, or batches.

Transcribe your first file

1. Open the Manage Models screen and pick a model. With Pro on an M-series Mac, filter the list by WhisperKit and download one; the developer's WhisperKit note calls Turbo a good choice for many Macs, and the first load of a large model can take a few minutes. On the free version, or on an Intel Mac, download one of the other Whisper models instead. WhisperKit models don't run on Intel Macs at all.
2. Drag an audio file into MacWhisper, as the Gumroad page describes. The pricing table also lists MOV and MP4 video among the supported formats.
3. Wait for the transcript. Check names, numbers and technical terms against the recording.
4. Export as text, or as SRT or VTT if you want subtitles for a video. The free version's exports are listed in the pricing table on macwhisper.com.

Download the model while you're online. After that, MacWhisper's privacy page says transcription runs on your Mac.

On an Intel Mac, transcribing on the Mac itself is much slower. The developer's page on slow transcription suggests one of the cloud transcription providers instead, which means your audio is uploaded (see What leaves your Mac below).

Record app audio and meetings (Pro)

Recording sound from other apps is a Pro feature; the free version records your Mac's microphone. MacWhisper's meeting recording guide says recording needs Screen Recording and Microphone permission. Grant both in System Settings when macOS asks.

For a video, a lecture or anything else that isn't a meeting, the same guide says you can record any audio by hand:

1. Start playing the audio in the other app.
2. Choose Start Recording from the MacWhisper menu.
3. When you're done, stop the recording from the bottom-right corner of the MacWhisper main window.
4. Select the recording in the history sidebar and wait for it to transcribe. Recordings also stay in the Recordings tab.

To have MacWhisper notice meetings, open Settings, go to Record Meetings and turn on Automatic Meeting Notifications. It watches apps such as Zoom, Teams, Webex, Slack, Discord, WhatsApp, and browsers including Chrome and Safari. FaceTime isn't available because of macOS privacy restrictions. When a meeting starts, a Meeting Detected notification appears; click Record within 10 seconds or it goes away.

The developer still describes automatic meeting recording as unfinished and warns it can occasionally lose data. For a meeting that matters, start the recording yourself with Start Recording, as above.

When the call ends, stop the recording from the main window or the Meeting Ended notification. It appears in the history sidebar, and MacWhisper transcribes it when you select it. For recordings made this way, the meeting guide says the other attendees are grouped as one speaker. The Gumroad page also lists real-time meeting transcription with speaker recognition as a Pro feature, on supported WhisperKit and Parakeet models; the meeting guide doesn't describe it.

When the transcript comes out wrong

• The transcript ends in the same words repeated over and over. The developer's note on repetitions says a WhisperKit model avoids this, which needs Pro and an M-series Mac. Without them, email the developer at [email protected].
• You get "Transcription finished without any text." Go to Settings → Advanced and toggle Downmix to Mono, then try again, as the error page suggests.
• Chinese comes out in Traditional characters when you wanted Simplified, or the reverse. Add a prompt under Settings → Advanced, for example 以下是普通话的句子. The developer's Chinese troubleshooting page says this can help but won't always work.
• The file mixes two languages. The developer's page on files with more than one language says MacWhisper can't transcribe both at once and recommends transcribing the file twice, once per language, then combining the parts by hand.
• Transcription on an M-series Mac is slower than it should be. After a crash, MacWhisper can switch to a compatibility mode; check Settings → Advanced → WhisperKit and make sure Disable all GPU usage is off.
• Activation says your licence is disabled. The personal licence has a soft limit of 3 devices. When you move to a new Mac, remove the licence from the old one in Settings → Pro.

What leaves your Mac

MacWhisper's privacy page says transcription, including speaker identification, runs on your Mac by default once a model is downloaded. Cloud transcription uploads your audio to the provider. The translation service sends your transcript to its servers. AI prompts send the transcript to the AI provider you picked, unless you choose a local one such as Ollama or LM Studio, which runs on your Mac. Cloud transcription uses your own API key, and some providers cap files at 25 MB, according to the cloud transcription page.

If you need to read along as it plays

The meeting guide cited above describes recording first and transcribing the saved recording afterward. MacWhisper's Gumroad page lists real-time captions with live translation, from the microphone or system audio, among the Pro features.

Without Pro, a Mac with Apple silicon has Live Captions built in. Apple's Live Captions guide for Mac puts it in System Settings → Accessibility → Live Captions and says it isn't available in every language, country or region. It isn't available on Intel Macs. Captions and transcripts for podcasts on a Mac compares it with other apps.

Vocena captions the microphone and the Mac's system audio live, in a floating window, on macOS 15.5 or later. Live captions are free with no time limit; transcribing an imported recording is a paid feature.