Skip to content
Petal

How-toMac5 min read

Transcribe Audio File on Mac for Free and Offline

Transcribe audio file on Mac for free and offline: drop the file on Petal, pick a file model, and copy the transcript. Includes a MacWhisper comparison.

Published by Optimal Apps, the maker of Petal.

On this page

To transcribe audio file on Mac recordings for free, drop the file on the Petal menu bar icon. Petal runs a speech model on your Mac, copies the transcript to the clipboard, and saves it in history. Nothing uploads, so it also works offline. For better accuracy, give audio files their own slower model in Settings > Router. This guide shows the steps, explains which models suit files, and compares the flow with MacWhisper.

The steps match Petal 2.8.1 on macOS 26. Optimal Apps publishes Petal, which is free and open source.

Quick checklist

Check Where Why
Petal setup is complete First launch Petal asks you to finish setup before it transcribes a file
A speech model is downloaded Settings > Models The model does the transcription
The file is a supported audio type Finder Petal rejects folders and unknown types
One file at a time Drop target Petal says "Drop one file at a time" for more
No recording is in progress Menu bar Petal says "Finish the current one first"

Step 1: Install Petal and download a model

  1. Download Petal from GitHub Releases or the Petal for Mac page, and drag it to Applications.
  2. Open Petal and complete the setup. Choose a speech model when it asks.
  3. Wait for the download to finish.

Petal needs an Apple silicon Mac.

Step 2: Pick a model for audio files

Dictation and files have different needs. A dictation must be fast, because you wait for the paste. A file can wait for a more accurate result. Petal lets you set the two models apart.

  1. In Settings > Models, download a second model that is more accurate, such as Whisper Large V3 Turbo or Qwen3 ASR 1.7B.
  2. Open Settings > Router. When two or more models are downloaded, a Speech Models section appears.
  3. Keep Dictation on your fast model.
  4. Set Audio Files to the accurate model. The default is Same as Dictation.

Petal describes it like this: "Dropped and chosen files can use a slower, more accurate model."

About heavy models

Each model card shows a speed meter, a quality meter, and the download size. The large models trade speed for accuracy:

Model Size Speed Quality Notes
Parakeet 110M ~455 MB 5/5 2/5 English, near-instant; good for dictation
Parakeet 0.6B V3 ~485 MB 5/5 4/5 25 European languages
Whisper Large V3 Turbo ~1.1 GB 2/5 5/5 99 languages; a good file model
Qwen3 ASR 1.7B ~2.5 GB 3/5 5/5 30 languages, automatic language detection
Voxtral Realtime 4B ~3.2 GB 4/5 5/5 13 languages; the largest download

The meters are Petal's own ratings, not a benchmark. A heavy model uses more memory and takes longer to load. On a Mac with little memory, a heavy model for every dictation makes the paste slow. That is the reason to keep it for files, where a longer wait is fine. Petal keeps the memory need of its default model low, so it runs on base Macs.

Step 3: Transcribe audio file on Mac from the menu bar

  1. Find the file in Finder.
  2. Drag it onto the Petal icon in the menu bar.
  3. The menu shows Drop to Transcribe and "Petal copies the transcript when it's done." Release the file.
  4. Petal shows a capsule with the progress. For long files, Petal can speed up the audio first, if Auto Speed-up is on.
  5. When it finishes, Petal says Copied to clipboard.

You can also click the menu bar icon and choose Transcribe Audio File…, and then select the file.

Supported types: AAC, AIF, AIFF, CAF, FLAC, M4A, MP3, MP4, and WAV, plus other files that macOS marks as audio.

Step 4: Copy or clean up the transcript

The transcript is already on the clipboard. Paste it with Command-V into any app.

To find it later:

  1. Open Settings > History.
  2. Find the recording. Use the search field to filter by text.
  3. Right-click the card and choose Copy Transcript.

A dropped file gives the raw transcript with no cleanup. To clean it up with your cleanup model, choose Transcribe and Clean Up on the card. To try a different model, choose Transcribe Again.

Convert a voice memo to text on Mac

  1. Open Voice Memos on the Mac.
  2. Drag the recording from the list to the desktop. macOS makes an M4A file.
  3. Drop the M4A file on the Petal menu bar icon.

The same works for a recording from Zoom, QuickTime Player, or a phone.

Why a file fails, and the fix

  • "File type not supported." The file is not audio, or it is a video type that Petal does not read. Convert it to M4A or WAV with QuickTime Player (File > Export As > Audio Only).
  • "Drop a file, not a folder." Drop the audio file itself.
  • "Finish setup to transcribe." Petal opens setup. Complete it, then drop the file again.
  • "Already transcribing a file." Wait for the current file to finish.
  • "No speech detected." The file is silent or very quiet. Check the file in QuickTime Player.
  • Wrong language. Choose a multilingual model. For Whisper Large V3 Turbo and Qwen3 ASR, you can also set the Spoken Language in Settings > Recording.

Audio to text on Mac for free: tips for a better transcript

  • Use the best source file. A WAV or a high-bitrate M4A gives better results than a small, compressed MP3.
  • Reduce noise before you record. A quiet room and a microphone near the speaker help more than any model.
  • Pick the right language model. For English, Parakeet works well. For a mixed-language file, use Whisper Large V3 Turbo or Qwen3 ASR 1.7B.
  • Split very long files. For a recording of several hours, split it into parts in an audio editor, and drop each part. You can then retry one part with another model.
  • Keep your Mac awake. A long file on a heavy model takes time. Connect the power adapter and leave the Mac on.

Transcribe recording on Mac offline

A dropped file never leaves the Mac. The speech model runs locally, so you can transcribe a recording on a plane or with Wi-Fi off. Only the first model download needs the internet.

Petal vs MacWhisper for files

MacWhisper is a strong app for files, and it has more file tools than Petal.

Item Petal MacWhisper
Price Free, open source Free version; Pro €64 one time
Where speech runs On the Mac On the Mac; optional cloud AI with your key
File input Drop on the menu bar icon or choose a file Main window with file drop
Output Transcript to clipboard and history Transcript editor, timestamps, and export formats
Dictation Yes, the main feature Only in the website edition

Choose MacWhisper when you need subtitles, timestamps, and exports for long recordings. Choose Petal when you want a quick transcript of a file and you also dictate every day. For more detail, read Petal vs MacWhisper and how to run Whisper on Mac. For offline setup, read offline speech to text for Mac.

Questions

Can I transcribe an audio file on Mac without uploading it?

Yes. Petal and MacWhisper run speech models on your Mac, so the file stays local. Petal is free and open source.

Which file types does Petal accept?

Petal accepts AAC, AIF, AIFF, CAF, FLAC, M4A, MP3, MP4, and WAV files, and other files that macOS reports as audio. Drop one file at a time.

How do I convert a voice memo to text on Mac?

Open Voice Memos, drag the recording to the desktop to get an M4A file, and then drop that file on the Petal menu bar icon. Petal copies the transcript to the clipboard.

Which model is best for audio files?

A slower, more accurate model, such as Whisper Large V3 Turbo or Qwen3 ASR 1.7B. Set it in Settings > Router > Speech Models > Audio Files, so dictation keeps a fast model.

Does Petal clean up file transcripts?

Not by default. A dropped file gives the raw transcript. To clean it up, open Settings > History and choose Transcribe and Clean Up on the recording.