The Problem with Voice Memos

You hit record on your Apple Watch during a walk. You dictate a thought into your iPhone while driving. You capture a meeting recap on your Mac before you forget the details.

Then what? The memo sits in Apple Voice Memos. You tell yourself you'll listen back later. You don't.

Apple added transcription in iOS 18, but it only works in about a dozen languages, doesn't always trigger, and gives you no way to export the text as a file. If you want the transcription in markdown, in a folder, ready for your AI tools — you're on your own. Copy, paste, format, save. Per memo. Every time.

My Transcriber removes all of that friction. Record however you already do. The text shows up in your folder.

What It Does

My Transcriber is a free Mac app that watches your Apple Voice Memos folder. When a new recording appears — recorded locally or synced via iCloud from your iPhone, iPad, or Apple Watch — it transcribes the audio using Whisper AI and writes a markdown file with the text and structured metadata.

No cloud service. No account. No subscription. Everything runs locally on your Mac.

Built for AI-Native Workflows

Voice memos are some of the richest context you produce. But audio files are useless to AI tools. Text files aren't.

My Transcriber turns every voice memo into a markdown file that you can drop into Claude Projects, feed to ChatGPT, or index with any RAG pipeline. Point your output folder at a synced directory and your transcriptions are available everywhere — searchable in Spotlight, readable in Obsidian, ready for whatever workflow you build around them.

The output format is designed for this. Each file has YAML frontmatter with structured metadata and the transcription as plain text:

---
status: done
audio_status: done
source_file: "Recording 2026-02-15 at 09.30.00.m4a"
voice_source: apple_voice_memos
captured_at: "2026-02-15T09:30:00+01:00"
transcription_language: "en"
---

Here's what I was thinking about the project timeline.
We should move the deadline to March because...

Why Not Apple's Built-In Transcription?

Apple Intelligence added voice memo transcription in iOS 18 and macOS 15, but it has real limitations:

• ~12 languages — Whisper handles 99+. If you speak Swedish, Japanese, Arabic, or any non-mainstream language, Apple won't transcribe it.
• Unreliable — Transcription doesn't always trigger. Sometimes you open a memo and there's just no transcript.
• No file export — The transcription is trapped inside the Voice Memos app. Getting it into a markdown file means manual copy-paste for every single memo.
• Requires Apple Silicon + Siri language — Not available on all Macs, and tied to your Siri language setting.

My Transcriber transcribes every memo, every time, in any language, and writes the result directly to a file. No manual steps.

How It Works

1. You record a voice memo on any Apple device
2. The recording syncs to your Mac via iCloud
3. My Transcriber detects the new M4A file
4. Converts to WAV using bundled ffmpeg
5. Runs Whisper AI with Metal GPU acceleration
6. Writes a markdown file to your output folder

A 10-minute memo typically transcribes in under 30 seconds on Apple Silicon. The app catches up on any memos recorded while it was closed, so nothing gets missed.

Key Features

• 100% local — Whisper runs on your Mac. No audio leaves your device.
• 99+ languages — Automatic language detection per memo.
• Live transcription — See text appear in real time while Voice Memos is recording.
• Push-to-talk — Record and transcribe directly in the app.
• Accept-and-delete — Review transcription, accept it, delete the source audio. Recordings don't pile up.
• Configurable output folder — Local, Dropbox, Google Drive, iCloud, SharePoint.
• Team folders — Multiple people can share a folder, each with their own subfolder.
• Multiple Whisper models — From tiny (fastest) to large-v3-turbo (most accurate).

Free to Use

My Transcriber is free. No cloud services, no subscriptions, no hidden costs. Everything runs locally on your Mac. You pay nothing.

Download it, grant Full Disk Access, pick your output folder, and record a voice memo to test it. The Whisper model downloads automatically on first launch (~1.5 GB for the default large-v3-turbo). After that, everything runs offline.

Download

If you record voice memos and wish they were searchable text files, ready for your AI tools — this is for you.

macOS 15+ required. Apple Silicon recommended.