Guides

How to transcribe an interview on a Mac for free, with the audio kept on your Mac

By Ryleigh Newman · Published October 5, 2026

To transcribe an interview on a Mac for free, record it on your phone or your Mac, drop the audio file on the Transcribe a file card in YapToText, and copy the plain text it gives back, transcribed on your Mac with Whisper. Then add the speaker names yourself, because YapToText does not label who is talking.

I built YapToText because my hands make typing difficult, and transcribing an interview by hand is exactly the kind of typing I cannot afford. It is free, with no account and no minute cap, and it needs an Apple silicon Mac with macOS 14 or later. This page is for journalists, researchers, students and anyone else with a recorded conversation: how to record it so it transcribes well, how to get the text, how to put names on it, and when another tool fits better.

Download for free. No account, nothing locked, macOS 14 or later on Apple silicon. Source on GitHub.

Steps

1. Record the interview with consent, with the phone close to the person speaking.
2. Add the interviewee's name and any jargon to a dictionary on YapToText's Dictionaries page.
3. On the Utility page, pick Raw Transcription in the Transcribe a file card's mode menu.
4. Drop the recording on the card, or click Choose File and pick it.
5. Click Copy and paste the text into your document.
6. Play the recording, break the text at each new speaker, and label each turn.

Record an interview that transcribes well

The transcript can only be as good as the recording, and most of the work happens before you press record. A phone close to the person talking beats a laptop across the table.

• Ask first. Ask before you record, and get the yes on the recording. The rules on recording people differ from place to place, and I am not a lawyer, so check the ones that apply to you, along with anything your newsroom, school or ethics board asks of you.
• Get close. Put the phone between you, a little nearer the person you are interviewing, since their words are the ones you will quote.
• Pick a quiet room. Music, traffic and a café's coffee machine all compete with the voices you want.
• Take turns. When two people talk at once, neither comes out clean, for a listener or for a model. Let each answer finish.
• Test ten seconds. Record a few words and play them back before the real thing starts.
• Get names on tape. Ask each person to say and spell their name at the start. You will need the spelling for the dictionary and the voice for labeling.

If you record in Voice Memos on an iPhone, Apple says that with iCloud the recording appears in Voice Memos on your Mac too, when both share an Apple Account and Voice Memos is on in iCloud settings. Drag it from Voice Memos to a Finder window to save it as a file, and the guide to transcribing Voice Memos on a Mac covers Apple's own transcript as well. For an interview over a video call, record with the call app's own recording, with consent: YapToText does not record calls or system audio.

Teach it the names before you transcribe

Interviews are full of proper nouns: the person's name, their organization, places and the jargon of their field. A speech model has never heard most of them. Add the names before you drop the file, not after.

On the Dictionaries page, each entry pairs what the app hears, in Heard, with the spelling you want, in Replace with. The Replace with side is also handed to Whisper as words to listen for, within a budget of about 200 characters with your dictionary terms first, and your dictionaries apply to file transcription as well as dictation.

Dictionary entry: A name the microphone may hear the way it sounds.

Heard: shavon
Replace with: Siobhan

What happens: Every shavon in the transcript is written Siobhan, and Siobhan is handed to Whisper as a word to listen for.

• One dictionary per project. Press New Dictionary and give it the project's name. Each dictionary has its own switch, so turn it off when the project ends instead of deleting it.
• Match what it heard. If a name still comes out wrong, put the spelling it actually wrote in Heard and run the file again. The Dictionaries guide covers Whole word, Case and order.
• Set the language. The Language picker on the AI Models page starts on your Mac's system language. For an interview in another language, choose it, or Auto-detect language. The language guide lists all 29. I have not tested a recording that switches languages partway through.

Transcribe the interview

The whole job happens in one card on YapToText's Utility page, Transcribe a file, with the same Whisper model that handles live dictation, on your Mac. YapToText is free on the Mac App Store. Pick Raw Transcription, drop the file, and copy the text.

Step 1: Pick Raw Transcription

• You: Open the Utility page. In the Transcribe a file card's mode menu, pick Raw Transcription.
• YapToText: Uses that mode for the file. Raw Transcription is literal speech to text with basic punctuation, and no AI rewrites the words.
• Check: The menu says Raw Transcription.

Step 2: Drop the recording

• You: Drag the audio or video file onto the card, or click Choose File and pick it.
• YapToText: Reads the file by its content, not its extension, and transcribes it on your Mac. A recording over 4 minutes is cut at quiet moments into pieces of about 2 minutes.
• Check: The card says it is transcribing your file. If it shows an error, the ladder below has the fixes.

Step 3: Take the text

• You: Click Copy and paste the text into your document, or click Insert to type it into the app you were last using.
• YapToText: Hands over plain text, with no speaker labels or timestamps.
• Check: The whole interview is in your document, ready for names.

Containers it recognizes include M4A, MP3, WAV, AIFF, CAF, FLAC, AMR, MP4 and MOV, so files from a phone, a handheld recorder or a camera all work. A file has no minute cap, so a two-hour interview just takes longer. If the text comes back wrong or empty, climb this ladder, smallest step first:

1. Play the file. If the card says no speech was found, listen to it. A voice that is faint at the source is something no model can recover.
2. Check the language. An interview in another language needs that language, or Auto-detect, on the AI Models page.
3. Convert the file. If the card shows an error, export the audio to a common format such as M4A or WAV and drop that instead.
4. Use a bigger model. Download one from the Dictation Model Library on the AI Models page and run the file again.

The guide to transcribing an audio file on a Mac has the full list of formats, and the Utility page guide covers the other cards.

Add speaker names by hand

The text comes back as plain text, with no names and no timestamps, so nothing marks where one speaker stops and the next begins. Putting them in is a listening job, and it goes faster than it sounds, because the hard part, the typing, is done. Listen once with the transcript open, and put a name at every change of voice.

Step 1: Break the text at each new speaker

• You: Open the transcript next to the recording and play it. Each time the voice changes, click into the text at that point and press Return.
• Check: Each paragraph holds one person's turn. If you wrote your questions down beforehand, your own turns are easy to spot.

Step 2: Label each turn

• You: Click at the start of each turn and type a short label, such as a name, initials, or Q and A for a question and answer format. If your hands are like mine, dictate the label with YapToText instead.
• YapToText: Types the label at your cursor, in whatever app holds the transcript.
• Check: Every paragraph starts with a name.

Say: With the cursor at the start of a turn, as its own short dictation.

Siobhan insert colon

What happens: The name lands with a colon attached. Colon is a one-word name, so it needs insert in front by default, and Auto mode leaves a dictation of six words or fewer as spoken. Check the spacing after it.

If each person was recorded on their own microphone, as a separate audio file, you can skip most of this: transcribe each file on its own and paste it under that person's name. YapToText adds no timestamps, though, so it cannot weave the files back into one conversation. The spoken punctuation guide lists every mark you can say, and the first dictation guide covers dictating into any document.

Pull quotes and notes without changing the words

An interview gives you two kinds of text: quotes, which must be exactly what the person said, and notes, which only need to be right. YapToText can do both, as long as you keep them apart. Quote from the raw transcript and the audio, never from a rewrite.

• Quotes come from Raw Transcription. It is what the speech model heard, with no AI rewrite. A model can still mishear, so check every quote against the recording before you use it.
• Clean Up is not for quotes. It removes filler words and false starts and fixes grammar. That makes readable notes and a changed quote.
• Keep the raw version. Copy the raw transcript somewhere safe before you rewrite anything, so the version nothing touched is always there.
• Turn it into notes. Send below passes the transcript to Transform text, where Note makes short headings and bullets with nothing added that was not said. The AI action card can run Summarize or Bullet Points instead.
• Read the rewrite. The cleanup model is small and runs on your Mac, so check a summary against the transcript before you rely on it.

The modes guide explains what each mode does and how to rewrite its instructions, and the guide to dictation for writers covers drafting the piece itself by voice.

Keep your sources' audio on your Mac

An interview can hold things a source trusted you with. YapToText's speech model ships inside the App Store build and runs on your Mac, so transcribing a file uploads nothing, and the only network use the app starts is a model download you choose yourself. There is no account and no analytics. The recording, the transcript and your dictionary stay on the Mac they started on.

• The Utility card keeps nothing. A file transcribed on the Transcribe a file card is not saved to History. The text stays in the card until you copy it, insert it or send it below.
• The menu bar route keeps a copy. Transcribe file in the menu bar panel, or on the History page, files the transcript in History under the file's name. Settings, Advanced, History & audio has Save dictation history, Clear history when I quit, and auto-delete after 30 days by default.
• Dictionaries stay local. Names you add live in one file inside the app's own container on your Mac, and nothing syncs it.
• Test it yourself. Turn Wi-Fi off and transcribe a file. The offline dictation guide walks through the test.
• Read the code. The source is public on GitHub under the GPL-3.0.

I can tell you what the app does. Whether that meets the rules of your newsroom, university or ethics board is a judgment for you and them, and this page makes no claim about it.

When MacWhisper fits better

MacWhisper is a file transcription app for the Mac that also runs Whisper locally, and for some interview work it is the better tool. If you need speaker names or timecodes, MacWhisper does what YapToText does not.

Checked October 3, 2026, on macwhisper.com and its help center. Prices change, so check before you buy.

If you transcribe interviews every week and need to know who said what and when, MacWhisper Pro is worth its price. If you need the words, for notes, a draft or a quote you will check against the tape, YapToText does it for free, and it also dictates into any app when you sit down to write. The full MacWhisper comparison goes further.

What it cannot do

• Speaker labels. It does not say who is talking. Names go in by hand, as above.
• Timestamps. It adds no timecodes and no subtitles, and there is no jump from a word to its place in the audio.
• Recording. It does not record calls or system audio, so the recording comes from your phone, your recorder or the call app.
• Batches. It takes one file at a time.
• Translation as it transcribes. It writes the interview in the language it was spoken. To translate afterwards, select the text and ask Quick Edit for another language, or run Translate to English, one of the built-in AI Actions.
• Some Macs and builds. It needs macOS 14 or later on Apple silicon, and nothing is claimed for Intel Macs. The Homebrew build ships without models, so download a speech model on the AI Models page first.

It gives you the words of the interview, not who said them or when.

The short version

• Record well. Ask first, get close, and get names on tape.
• Transcribe. Add the names to a dictionary, pick Raw Transcription, and drop the file on the Transcribe a file card.
• Finish by hand. Break the text at each new voice, label the turns, and check every quote against the audio.
• Need speakers or timecodes. Use MacWhisper instead.

The interview goes in, the words come out, and your source's voice never leaves your Mac.

If an interview will not transcribe, press Copy Diagnostics in Settings, About and open an issue on GitHub with the file type. Every one gets read.

Download for free. No account, nothing locked, macOS 14 or later on Apple silicon. Source on GitHub.

About the author. Ryleigh Newman is a developer whose hands make typing difficult and who dictates every day with YapToText, the app this site describes. Facts about other apps come from their own sites, with the date they were read. More about the project.

Related

• Transcribe audio to text on a Mac, free, without uploading it
• A free MacWhisper alternative built for dictation first
• How to transcribe Voice Memos on a Mac, with or without Apple's transcript
• Dictionaries: teach YapToText your names and jargon
• Utility, the text workbench, and your dictation statistics
• How to transcribe a Zoom recording on a Mac for free, without a cloud recording
• A free Otter.ai alternative on the Mac, for recordings you already have
• Legal dictation on a Mac: letters, memos and briefs by voice, without Dragon
• Dictation for academic writing on a Mac: papers, theses and grants by voice
• How to transcribe a podcast episode on a Mac for free, and turn it into show notes
• The best transcription apps for Mac in October 2026, ranked for audio and video files
• Dictation vs transcription: what is the difference, and which one you need on a Mac

Sources

• https://support.apple.com/guide/voice-memos/see-your-recordings-on-all-your-apple-devices-vma6cc4d0571/mac (checked 2026-10-03)
• https://support.apple.com/guide/voice-memos/share-a-recording-vm05f9fa82d4/mac (checked 2026-10-03)
• https://www.macwhisper.com (checked 2026-10-03)
• https://www.macwhisper.com/legal/privacy-policy (checked 2026-10-03)
• https://docs.macwhisper.com/article/34-transcription-is-slower-than-expected (checked 2026-10-03)
• https://apps.apple.com/us/app/yaptotext/id6786382289 (checked 2026-10-03)

Questions