How to transcribe an audio file on iPhone (MP3, M4A, WAV)

Not every recording starts in the app that will transcribe it. Interviews live in Voice Memos, podcasts arrive as MP3 attachments, a colleague sends an M4A from a meeting you missed, and a WAV file from a recorder ends up in Files. This guide shows how to transcribe an audio file you already have on iPhone, and what you get back.

It uses Memo, a free app for iOS 26 or later. Memo accepts MP3, M4A and WAV files and produces a transcript, a summary with a title, a to-do list and a chat that answers questions about the recording.

Route 1: share the file to Memo

This is the route from any app that has the file.

1. Open the recording where it is: Voice Memos, Files, Mail, Messages, a cloud drive, a podcast app that lets you export.
2. Tap Share.
3. Choose Memo in the share sheet. Memo appears for MP3, M4A and WAV files.
4. Pick the engine, and Memo imports the file and starts writing.

From Apple’s Voice Memos app this takes three taps: open the recording, Share, Memo. Every voice memo you have ever made can become text this way, one at a time.

Route 2: import from inside Memo

If you would rather start in Memo:

1. Tap the plus button.
2. Choose Import audio.
3. Pick the file from Files, which shows iCloud Drive and any other storage app you have connected.

Either route ends in the same place: a new memo with the file’s audio and everything Memo produces from it.

Choosing the engine for a file

Memo asks which engine to use before it processes the file, and tells you what happens to the audio.

• On-device AI keeps the file on your iPhone. The transcript and summary are produced by Apple’s speech model and Apple Intelligence on the phone. Nothing is uploaded, which is the right choice for an interview with a source, a recording with medical or legal content, or anything you would not email. It needs an iPhone with Apple Intelligence and a language Apple supports on the device.
• Cloud AI uploads the file to Memo’s servers and works on every iPhone in more than 99 languages, with maximum accuracy. The free tier accepts recordings up to 30 minutes and a daily allowance; Memo Pro lifts both.

You can use both engines on the same phone and pick per file. Which iPhones and languages support On-device AI has the exact conditions.

What comes back

• A transcript with a timestamp on every line. Tap a line to play that part of the file, which is how you check a quote before you use it.
• A summary with a title and the key points, so a 40-minute interview has a two-paragraph version you can read first.
• To-dos, if the recording contains any. An interview usually does not; a recorded meeting usually does.
• A chat that answers questions from the file: “What did he say about pricing?” or “Give me the three best quotes.”

Files longer than 30 minutes

With On-device AI, length is limited only by your phone’s storage, so a two-hour recording is fine. With Cloud AI on the free tier, a file over 30 minutes needs Memo Pro. If the phone supports On-device AI and the language is supported, use that engine for long files; it is also the private option. See Free limits, Memo Pro, cancelling and restoring.

Getting the text out again

Open the memo menu:

• Copy the transcript or the summary and paste it into a document, an email or a chat.
• Share the text through the iOS share sheet.
• Export as PDF to keep the transcript as a file, or Print it.
• Share a private link so someone else can read the transcript and the summary in a browser.

The original audio stays with the memo, and you can share or download it again from the same menu. Exports are made on the phone from what Memo already has; nothing extra is uploaded. Export a memo lists every option.

Quality: what makes a file transcribe well

• Clear speech beats high bitrate. A 64 kbps MP3 of one person speaking clearly transcribes better than a studio WAV of four people talking over each other.
• Trim silence and music. A podcast intro with music produces a few odd lines at the start. Memo copes, but the summary is cleaner without it.
• One language per file gives the best result. Cloud AI can handle a switch mid-recording, but the summary follows the main language.
• If a file comes back as “No speech was detected”, it was silent or too quiet. That article explains what to check.

The short version

Share the MP3, M4A or WAV to Memo from wherever it lives, or upload it from inside the app, pick On-device AI for privacy and long files or Cloud AI for any iPhone and any language, and read the summary before the transcript.