The Problem with Voice Memos

Apple Voice Memos is great at one thing: recording. The app is simple, it syncs across devices via iCloud, and it works reliably on Apple Watch, iPhone, iPad, and Mac.

But once you've recorded something, good luck getting the text out. Apple's built-in transcription (via Apple Intelligence) only works in about a dozen languages, requires specific hardware, and doesn't let you export the text in any useful way.

The result: people have dozens or hundreds of voice memos sitting in the app, unlistened, unsearchable, and effectively lost.

The Old Workflow (6 Steps)

Before My Transcriber, transcribing a voice memo meant something like this:

1. Open Apple Voice Memos
2. Find the recording you want
3. Export/share the M4A file
4. Open a transcription tool or website
5. Upload or drag the file in
6. Copy the resulting text into your notes or document

Six steps. For every single recording. And that's assuming you remember to do it at all.

Most people do this for one or two important memos, then give up on the rest. The friction is too high.

The New Workflow (2 Steps)

With My Transcriber, the workflow is:

1. Record a voice memo (Apple Watch, iPhone, or Mac)
2. The transcription appears in your folder as a markdown file

That's it. No exporting. No uploading. No switching between apps. You record, and the text shows up.

The app handles everything in between: detecting the new recording, converting the audio format, running the transcription, and writing the file.

How It Works Under the Hood

Here's what happens when you record a voice memo:

1. iCloud syncs the recording to your Mac. Apple Voice Memos stores recordings in a specific folder on your Mac. When you record on Apple Watch or iPhone, the file syncs via iCloud to this folder automatically.

2. My Transcriber detects the new file. The app uses macOS Full Disk Access to watch the Voice Memos capture folder. When a new .m4a file appears or an existing one finishes syncing, the app picks it up.

3. Audio format conversion. Voice Memos records in M4A format. The app converts it to WAV using a bundled copy of ffmpeg. This happens in seconds.

4. Whisper AI transcription. The WAV file is processed by OpenAI's Whisper model running locally on your Mac with Metal GPU acceleration. The language is detected automatically.

5. Markdown file output. The transcription is written as a .md file in your chosen output folder. The file includes YAML frontmatter with the capture date, duration, and detected language.

Full Disk Access: Why It's Needed

The one permission My Transcriber requires is Full Disk Access. This is because Apple stores Voice Memos files in a protected system directory that normal apps can't read.

Without Full Disk Access, the app simply can't see your voice memos. There's no other way to access them programmatically on macOS.

The app prompts you to grant this during setup. It takes about 10 seconds in System Settings. You only need to do it once.

If this feels like a lot of trust to place in an app, I understand. The app processes everything locally. No audio data leaves your Mac. You can verify this with any network monitoring tool.

Live Watching: Transcribe While You Record

If you're recording a voice memo on your Mac, My Transcriber can transcribe it in real time as you speak. It reads the audio in chunks and processes each chunk as it becomes available.

You'll see the text appearing in the app while the recording is still active. When you stop recording, the final transcription is assembled from all the chunks and saved to your folder.

This is useful for long recordings — meetings, brainstorming sessions, lectures — where you want to see the text building up in real time rather than waiting for the whole thing to finish.

Background Catch-Up

Your Mac isn't always on. You might record several memos on your Apple Watch over the weekend while your laptop is closed. Or your Mac might have been asleep when a recording finished syncing from your iPhone.

That's fine. My Transcriber handles this gracefully.

Every time the app starts up, it scans for voice memos that haven't been transcribed yet. It processes them one by one in the background. Nothing gets skipped. Nothing requires your attention.

You can leave the app running and forget about it. When you come back, all your voice memos from the past hours or days will be sitting in your folder as text files.

What the Output Looks Like

Each voice memo becomes a markdown file named by date and time. For example:

2026-02-22-143025.md

Inside, you'll find YAML frontmatter and the transcription text:

---
captured_at: "2026-02-22T14:30:25+01:00"
duration: "2m 15s"
language: "en"
source: voice_memo
---

I just realized we need to push the launch date back
by a week. The design review isn't finished and I
don't want to ship without it. Let me ping the team
about this after lunch.

Plain text. Searchable. Portable. Opens in anything.

Why Apple's Built-In Transcription Falls Short

Apple added transcription to Voice Memos through Apple Intelligence. It's a step in the right direction, but it has real limitations:

• English-only (mostly). Apple Intelligence transcription supports about a dozen languages. If you speak anything outside that list, you're out of luck.
• No export. The transcription stays inside the Voice Memos app. You can't easily copy it into a document or notes app.
• Hardware requirements. Apple Intelligence requires specific chips. Older Macs can't use it.
• No automation. You still need to open each memo individually to see the transcription.

Apple's solution works if you have one English recording and want to glance at it inside the app. It doesn't work if you have a dozen recordings in three languages that need to end up as searchable text files.

Setup Guide

Getting My Transcriber running takes about two minutes. Here's the process:

Step 1: Download and install. Download My Transcriber from the website. Open the DMG and drag the app to Applications.

Step 2: Launch and grant Full Disk Access. The app will ask for Full Disk Access on first launch. Click the button to open System Settings, toggle the permission for My Transcriber, and restart the app.

Step 3: Choose your output folder. Pick where you want your transcription files to go. This can be any folder on your Mac — a Dropbox folder, an Obsidian vault, a Google Drive directory, whatever works for you.

Step 4: Record a test memo. Open Voice Memos on your Mac or Apple Watch and record a short test. Within a minute or two, the transcription should appear in your folder.

That's the complete setup. The app downloads the Whisper model (~1.5GB for the default large-v3-turbo) automatically on first use. After that, everything runs offline.

Tips for Best Results

Whisper is good, but it's not magic. A few things help:

• Speak clearly. Obvious, but it matters. Whisper handles accents and natural speech well, but mumbling into your wrist while jogging will produce worse results than speaking at a normal pace.
• Minimize background noise. Wind, traffic, and crowded rooms make transcription harder. The larger Whisper models handle noise better if you need it.
• Try a smaller model for faster processing. The default large-v3-turbo gives the best accuracy. For quick drafts or clear audio, switch to base or tiny in settings for faster speed.
• Let iCloud sync finish. If you record on your Apple Watch and immediately open your laptop, the file might still be syncing. Give it a minute. My Transcriber will process it as soon as the sync completes.

Common Questions

Does it work with recordings I've already made? Yes. On first launch, it can process your existing voice memos that haven't been transcribed yet.

What happens if my Mac is asleep when I record? The recording syncs via iCloud. When your Mac wakes up, My Transcriber processes it. Nothing is lost.

Can I delete the original audio after transcription? Yes. The app has a built-in accept-and-delete workflow. Review the text, accept it, and the source audio gets removed.

How long does transcription take? A 10-minute memo typically takes under 30 seconds on Apple Silicon with GPU acceleration. Shorter memos are near-instant.

Does it use the internet? No. Everything runs locally on your Mac. No audio is uploaded anywhere. The only network use is downloading the Whisper model on first launch.

What You End Up With

After a week of using My Transcriber, you'll have a folder full of dated markdown files. Each one is a voice memo, transcribed.

You can search through them with Spotlight. You can open them in Obsidian and link them to other notes. You can drop the whole folder into a Claude Project for context. You can grep for a keyword and find that thing you said three weeks ago.

Your voice memos stop being an audio graveyard and become a searchable text archive.

That's the whole point.

Try It

Record on your Apple Watch. The transcription shows up in your folder. No steps in between.

macOS 15+ required. Apple Silicon only. Setup takes about two minutes.