📝🎙️
4 Ways to Transcribe Meeting Minutes on Mac
Built-in Mac: Voice Memos + Word
On-device: MetaWhisp drag-drop
Auto-join: Otter.ai bot
Inside the call: Zoom/Teams transcripts
TL;DR: Four practical ways to transcribe meeting minutes on Mac in 2026. (1) Record with Voice Memos or QuickTime, then drop the audio file into Word for Microsoft 365 Transcribe (300 MB cap, 5 hours per month). (2) Drop the same audio into MetaWhisp for free on-device Whisper transcription with no caps. (3) Let Otter.ai auto-join your Zoom, Google Meet, or Teams calls for real-time transcription plus AI summaries ($16.99/month). (4) Use the built-in transcript feature in Zoom Pro, Google Meet Workspace, or Teams Premium. The right choice depends on privacy needs, meeting volume, and budget.
Four meeting minutes transcription methods on Mac comparison schematic showing built-in Word on-device MetaWhisp Otter auto-join and Zoom Teams transcripts

Which Method Should You Use for Meeting Minutes?

Three things decide it: how sensitive the meeting is, how many you run a month, and what you're willing to spend. Match yourself to one of these: I'm Andrew Dyuzhov — I build MetaWhisp, the on-device option below. I'm covering all four anyway, because there's no single best way to transcribe a meeting. A confidential board call and a Monday standup don't have the same needs, and pretending one tool wins every time would just waste your time.
There are really two camps. Audio-file transcription happens after the call: record it (Voice Memos, QuickTime, a local Zoom recording), then drop the saved file into a desktop tool — MetaWhisp, MacWhisper, or Word's Transcribe feature. It works offline, you keep the audio, and it doesn't care which platform the meeting was on. Live transcription happens during the call: Otter.ai, Zoom transcripts, Google Meet captions, Teams Live Captions. Live tools usually label speakers better, since they already know who joined — but they need platform hooks and almost always send your audio to the cloud. So it comes down to one trade: can you wait 3-7 minutes after the meeting, or do you need the words on screen as people talk? If the content is sensitive, that question answers itself — only an on-device, audio-file tool keeps everything on your Mac.

Method 1: Record with Voice Memos, Transcribe in Microsoft Word

If you already have a Microsoft 365 subscription, this is the most integrated path. Voice Memos (built into macOS) records the meeting. Word for M365's Transcribe feature converts the audio file to a transcript inside a Word document with speaker labels and timestamps. Steps:
  1. Open Voice Memos on Mac (Applications → Voice Memos)
  2. Click the red record button. Voice Memos uses your default microphone — make sure it's set to capture your speakers' audio too (use BlackHole or Loopback for system audio capture on Mac)
  3. Conduct your meeting. Voice Memos handles recordings up to several hours
  4. Click the red square to stop. The recording saves as an .m4a file
  5. Open Word for Mac (M365 subscription required)
  6. Click Home → Dictate dropdown → Transcribe
  7. Click Upload audio in the right pane
  8. Select your Voice Memos .m4a file
  9. Wait for transcription. Word's cloud-side processing typically takes 1-2× the audio length
  10. Click Add to document when complete. Choose with speakers, with timestamps, or plain text
Hard limits: 300 MB per file, 5 hours of transcription per month per M365 account, per Microsoft's official Transcribe documentation. Standalone Word 2019 and Word 2021 don't have this feature — it's M365-subscription-only.
Pro tip: To capture both your voice and the meeting participants' audio on Mac, install BlackHole (free, open-source virtual audio driver). Set up a Multi-Output Device in Audio MIDI Setup that includes both your built-in mic and BlackHole, then select that device in Voice Memos. This captures both sides of the call without any special meeting platform support.

Method 2: Drop the Audio into MetaWhisp for Free On-Device Transcription

If you don't want to upload meeting audio to a cloud service — or your M365 quota is exhausted — MetaWhisp handles the same .m4a file locally on your Mac. No quota, no cap, no upload. Steps:
  1. Record the meeting with Voice Memos (or any other recorder) following Method 1 steps 1-4
  2. Open MetaWhisp on your Mac
  3. Drag the .m4a file directly into the MetaWhisp window
  4. Wait for transcription. On M2 or M3 MacBook Air, a 1-hour recording transcribes in roughly 3-7 minutes
  5. Click Export in the toolbar
  6. Choose format: .txt, .docx, or .srt
  7. Save to your preferred location
Limits: None on file size. MetaWhisp processes audio in chunks, so a 10-hour board meeting recording transcribes the same way as a 30-minute call — just takes longer. Memory usage stays under 6 GB RAM regardless of audio duration because chunks stream through the inference pipeline. The output transcript runs through Whisper large-v3-turbo on Apple Neural Engine, the same model architecture that powers most professional transcription services. Word error rate is comparable to Amazon Transcribe and Google Cloud Speech-to-Text for clean meeting audio.
The privacy property that matters here: the file you drop never leaves your Mac. After the one-time model download, Whisper lives on your disk, processes the audio in memory, and writes the transcript back to disk — no network call in between. Don't take my word for it. Turn off Wi-Fi before you transcribe, watch the connections with Little Snitch, or read MetaWhisp's privacy policy. For a meeting full of confidential strategy, HR details, attorney-client privileged content, or HIPAA-relevant health talk, that on-device path removes the cloud-vendor BAA you'd otherwise need — the HIPAA speech-to-text guide covers the specifics. And it doesn't change with the meeting: a one-on-one therapy session and a 12-person board call run exactly the same way, with no per-meeting tier or quota.
Meeting minutes transcription privacy flow comparison showing cloud upload versus MetaWhisp on-device path keeping audio on Mac

Method 3: Let Otter.ai Auto-Join Your Meetings

If you want real-time transcripts with AI summaries and action items, Otter.ai is the dominant tool for meeting capture. The auto-join feature connects Otter to your calendar (Google Calendar, Outlook, or Microsoft 365), spots meetings on your schedule, and sends a bot to join Zoom, Google Meet, or Microsoft Teams calls automatically. Setup:
  1. Create an Otter account at otter.ai (free Basic, Pro $16.99/month, Business $30/seat/month)
  2. Connect your calendar via OAuth in Otter's settings
  3. Enable OtterPilot auto-join for the calendars you want Otter to monitor
  4. Otter sends a bot to join each upcoming meeting automatically
  5. The bot appears in the meeting as a participant labeled "Otter.ai"
  6. After the meeting ends, Otter generates a transcript with speaker labels, AI-extracted action items, and a meeting summary
  7. The transcript is searchable in Otter's web interface and shareable via link or PDF export
Strengths: Auto-join requires zero per-meeting action from you. Speaker diarization in 2-4 person calls hits 90-95% accuracy. AI summaries and action items save the manual note-taking step entirely. Weaknesses: The Otter bot is visible to all meeting participants, which may not be acceptable in confidential settings. Free tier caps at 300 minutes per month and 30 minutes per file. Pro at $204/year sits at the top of consumer-tier pricing.
For meetings where you can't have a visible third-party bot — legal depositions, HR conversations, executive sessions — Otter's auto-join isn't an option. Record locally with Voice Memos and transcribe afterward with MetaWhisp or Word.

Method 4: Use Built-in Transcripts in Zoom, Meet, or Teams

Each major meeting platform now ships its own real-time transcript feature on paid tiers. These are convenient if you already pay for the platform but each has tradeoffs.
PlatformRequired tierStorageSpeaker labelsAI summary
Zoom Cloud RecordingPro $15.99/moZoom cloudYesSmart Recording add-on
Google MeetWorkspace Business Standard $14/user/moGoogle DriveYesGemini summary on Enterprise
Microsoft Teams PremiumPremium $10/user/mo on top of M365OneDrive/SharePointYesCopilot recap
Steps vary slightly per platform but the pattern is the same:
  1. Start the meeting on the platform
  2. Click More → Record / Transcribe (Zoom uses "Cloud Recording"; Google Meet uses "Start transcript"; Teams uses "Start recording & transcription")
  3. Conduct the meeting
  4. End the meeting
  5. The transcript appears in your platform's storage (Zoom cloud, Google Drive, OneDrive) within 5-30 minutes depending on the meeting length
  6. Download as .vtt, .txt, or .docx depending on platform
Privacy posture: All three platforms process audio and store transcripts in the vendor's cloud. For business meetings under your IT department's existing data-handling policies, this is usually fine. For meetings under stricter compliance regimes (HIPAA, attorney-client privilege, classified content), check whether your subscription tier includes the appropriate BAA or data-residency option.
Built-in transcripts versus a separate tool is really integration versus flexibility. The built-in kind attaches itself to the meeting record, lands in your existing workspace storage, and follows your org's IT rules — clean, if you live inside one platform. Switch platforms, though, and your transcript history scatters across vendor clouds. That's where Otter and MetaWhisp earn their place: they keep every transcript in one spot, no matter where the meeting happened. If your week is Zoom for sales calls, Teams for standups, and Google Meet for everything else, one consolidated tool beats hunting through three separate vendor silos.

How Do I Capture Both Sides of the Conversation on Mac?

A meeting recording is only useful if you can hear all participants. Voice Memos and most third-party recorders only capture your microphone by default — the audio coming from your speakers (the other participants' voices) is not recorded. Three workarounds: For one-off important meetings, BlackHole takes about 10 minutes to set up and then works permanently. The Multi-Output Device in Audio MIDI Setup (Applications → Utilities → Audio MIDI Setup) is the key configuration step — without it, your recordings will only contain your microphone.
Meeting transcription methods Mac comparison table showing Voice Memos Word MetaWhisp Otter Zoom Teams price privacy real-time AI summary
Mac audio routing diagram for meeting capture using BlackHole virtual driver Multi-Output Device combining microphone and system audio into Voice Memos recording

What's the Cheapest Way to Transcribe Meetings?

Cost comparison for transcribing 4 meetings per week (about 16 hours per month of audio):
MethodCost / monthCost / yearNotes
MetaWhisp on-deviceFree$0No cap, no subscription
Word M365 TranscribeAlready paidAlready paid5-hour monthly cap may cut you off
MacWhisper one-time$32 onceFile transcription only, no live
Otter.ai Pro$16.99$204Auto-join + AI summaries
Zoom Pro$15.99$192Includes Cloud Recording
Google Meet Workspace$14/user$168/userTranscript via Workspace Business
Teams Premium$10/user (+M365)$120/user (+M365)Add-on to existing M365
Rev human transcriptionPer-minute$1.50/min × 60 min × 16 = $1,440For perfect-accuracy mandates only
For most users, MetaWhisp's free on-device path covers 90% of meeting transcription needs. The paid options are worth it specifically when you need:

Will the Other Participants Know I'm Recording?

Two-party consent laws in many US states and most of the EU require that all meeting participants be informed when a meeting is being recorded. The specific rules vary by jurisdiction, per U.S. Department of Justice electronic surveillance guidance. Practical guidance: For internal team meetings where everyone is aware of the recording practice, the announcement step becomes routine. For external meetings (client calls, vendor negotiations, candidate interviews), explicit per-meeting consent is the safer practice.

How Do I Find Action Items in a Meeting Transcript?

Once you have the raw transcript, the next step is usually extracting decisions, action items, and follow-ups. Three options: For transcripts produced by MetaWhisp's on-device workflow, the AI-summarization step happens separately. Drag the transcript into ChatGPT, Claude, or your preferred LLM with a prompt that asks for structured output: action items as a bulleted list, decisions as a separate list, names of people assigned to each action item.
Splitting it in two — transcribe on-device with MetaWhisp, then paste the text into ChatGPT or Claude to summarize — puts the privacy where it counts. The audio, the most sensitive part, never leaves your Mac. Only the transcript text goes to an LLM, and only when you choose to send it. Better still, you can redact names or numbers out of the transcript first — something Otter's auto-summary can't do, since it always chews through the full audio and transcript on its own. Yes, running the summary by hand is a little more work. For a call about attorney-client privilege, board-level strategy, or HR-confidential matters, that small friction buys a clean answer to the only question that matters later: where did this audio go?

What File Formats Should I Use for Meeting Transcripts?

Three common formats with different strengths: MetaWhisp exports all three. Otter exports .docx, .txt, .pdf, and .srt. Word M365 Transcribe exports as .docx only.

Frequently Asked Questions About Meeting Transcription

What is the best way to transcribe a meeting on Mac?

The best method depends on three factors: how sensitive the meeting content is, how many meetings you transcribe per month, and your budget. For confidential meetings, record locally with Voice Memos and transcribe with MetaWhisp on-device for free with no cap. For routine internal meetings with AI summaries, use Otter.ai auto-join at $16.99/month. For occasional meetings with existing Microsoft 365, use Word's built-in Transcribe feature (5-hour monthly cap).

Can I transcribe meeting minutes for free on Mac?

Yes. Voice Memos (built into macOS) plus MetaWhisp (free download) is a fully free workflow. Record the meeting with Voice Memos, drop the .m4a file into MetaWhisp, get a transcript on-device with Whisper large-v3-turbo. No subscription, no quota, no cloud upload. Apple's macOS Dictation is also free but doesn't handle long meeting audio well.

How do I record both sides of a Zoom meeting on Mac?

Install BlackHole (free, open-source virtual audio driver), then create a Multi-Output Device in Audio MIDI Setup that combines your built-in microphone and BlackHole. Set this device as the input in Voice Memos. The recording will capture both your voice and the other participants' voices. Alternatively, use Zoom's built-in Cloud Recording feature on Zoom Pro, which captures all sides natively.

Is Otter.ai HIPAA-compatible for medical meetings?

Only the Otter Business tier ($30/seat/month) offers a Business Associate Agreement on request, making it HIPAA-eligible. The free Basic tier and the paid Pro tier do not offer BAAs and are not HIPAA-compatible for ePHI. For medical meeting transcription, either upgrade to Otter Business with signed BAA or use on-device tools like MetaWhisp that don't transmit audio to any vendor at all. See our HIPAA speech-to-text guide for compliance specifics.

How long does it take to transcribe a 1-hour meeting?

Depends on the method. MetaWhisp on-device transcribes a 1-hour meeting in 3-7 minutes on M2 or M3 MacBook Air. Word M365 Transcribe takes 30-90 minutes including upload and cloud queue. Otter.ai is real-time during the meeting (no post-processing wait). Zoom Cloud Recording transcripts are typically ready 5-15 minutes after the meeting ends.

How accurate is Mac meeting transcription?

Whisper large-v3-turbo (used by MetaWhisp and most desktop Whisper apps) scores 2.76% WER (about 97%) on clean read English in our LibriSpeech test-clean benchmark, comparable to the model figures published by other Whisper-based services. Real meeting audio is harder than a clean benchmark: multi-speaker calls with overlapping voices, heavily accented English, and domain-specific vocabulary (medical, legal) all raise the error rate. We don't separately benchmark those conditions, so treat clean single-speaker audio as the best case and always review high-stakes transcripts before relying on them.

Can I transcribe a meeting recorded in another language?

Yes. Whisper large-v3-turbo supports 99 languages and auto-detects the language from the first 30 seconds of audio. Drag a Spanish, French, Mandarin, Japanese, or Russian meeting recording into MetaWhisp and it produces a transcript in the original language. For translation to English, MetaWhisp's Translate mode (Pro tier or own API key) handles this in one step.

Do I need to ask permission before recording a meeting?

Yes in many jurisdictions. Two-party consent laws in California, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Pennsylvania, and Washington require all parties to consent to recording. Most of the EU follows similar consent requirements under GDPR. Announce the recording at the start of the meeting, get verbal consent, and note it in the agenda. For external meetings, explicit per-meeting consent is the safer practice.

About the Author

Andrew Dyuzhov is the solo founder and CEO of MetaWhisp, a free on-device voice-to-text app for macOS that runs Whisper large-v3-turbo on Apple Neural Engine. He built MetaWhisp to give knowledge workers a way to transcribe meetings, dictation, and audio files without uploading anything to a vendor's cloud. The procedures in this article reflect his hands-on testing of Voice Memos, Word M365 Transcribe, Otter.ai auto-join, and Zoom Cloud Recording on his M3 MacBook Air. Connect on X or GitHub.

Related Reading