Complete guide to My Transcriber for Mac — automatic voice memo transcription, entirely on your device.
My Transcriber watches your Apple Voice Memos, transcribes them with Whisper AI, and saves the text as markdown files in a folder you choose.
Record on your iPhone, iPad, Apple Watch, or Mac. The transcription appears in your folder automatically.
Whisper AI runs entirely on your Mac with Metal GPU acceleration. No audio is sent to any server.
Plain text files with YAML frontmatter. Searchable in Spotlight, compatible with any text editor or AI tool.
Get up and running in a few minutes
Download My Transcriber for Mac
Download DMGOpen the downloaded DMG file and drag My Transcriber to your Applications folder
Grant Full Disk Access
My Transcriber needs Full Disk Access to read Voice Memos synced via iCloud. Go to System Settings → Privacy & Security → Full Disk Access and enable My Transcriber.
First launch — model download
On the first launch, My Transcriber downloads the Whisper large-v3-turbo model (~1.5 GB) from Hugging Face. This is a one-time download. Transcription starts automatically once the model is ready.
Record a voice memo on any Apple device. My Transcriber picks it up and transcribes it automatically.
My Transcriber watches the Voice Memos folder on your Mac. When a new recording appears (either recorded locally or synced via iCloud from your iPhone, iPad, or Apple Watch), the app:
When you launch the app, it checks for any voice memos that were recorded while the app was closed and transcribes them. You don't need to keep the app running at all times — it catches up when it starts.
When the Voice Memos app is actively recording, My Transcriber can transcribe the audio in real time using chunked processing. You see partial text appear as you speak.
Record and transcribe without leaving My Transcriber
The Transcribe tab provides a dedicated recording interface. You can record directly in the app, review the transcription, and choose to accept or discard it.
After recording, you can review the transcription and either accept it (saves to your output folder) or delete it.
Each transcription is saved as a .md
file with structured metadata in the frontmatter:
---
captured_at: "2026-01-15T09:30:45+01:00"
source: voice_memo
duration_seconds: 47
model: whisper-base
language: en
---
Just had a great idea for the project. We should
reorganize the API endpoints to group them by
resource type instead of by action...Choose accuracy vs. speed
My Transcriber uses OpenAI's Whisper models for transcription. All models support 99+ languages with automatic language detection.
| Model | Size | Best for |
|---|---|---|
| Tiny | ~75 MB | Fastest speed, lower accuracy |
| Base | ~150 MB | Good balance, fast |
| Small | ~500 MB | Better accuracy |
| Medium | ~1.5 GB | High accuracy |
| Large v3 Turbo | ~1.5 GB | Best accuracy, fast (default) |
The default model is Large v3 Turbo — best accuracy at good speed. Smaller models (Tiny, Base) are available if you prefer faster processing. Models are downloaded automatically from Hugging Face when first selected.
By default, My Transcriber saves transcriptions to
~/My Transcriber/. You can change this
in Settings to any folder on your Mac.
Point the output folder to a synced folder to access transcriptions from any device:
Multiple people can point My Transcriber at the same shared folder (e.g., a Dropbox or SharePoint folder). Each person's transcriptions land in their own account subfolder, so nothing gets mixed up.
Shared Folder/
├── Registry/
│ ├── 7a571023/ ← Person A's transcriptions
│ │ └── voice/2026/
│ └── b3e8f190/ ← Person B's transcriptions
│ └── voice/2026/Everyone on the team can search and read each other's transcriptions. No extra software or accounts needed — just a shared folder.
After a voice memo is transcribed, you can review the text and choose to accept it. Accepting deletes the source audio recording and keeps only the markdown transcript. This prevents recordings from piling up on your device.
If you prefer, enable auto-delete recording after transcription in Settings. With this enabled, the source audio is deleted automatically after successful transcription — no manual review needed.
Cause: Full Disk Access not granted
Fix: System Settings → Privacy & Security → Full Disk Access → Enable for My Transcriber. You may need to restart the app after granting access.
Cause: iCloud sync for Voice Memos is disabled or lagging
Fix: On your iPhone, go to Settings → [your name] → iCloud → Voice Memos and make sure sync is enabled. On your Mac, check System Settings → Apple Account → iCloud → Voice Memos. Sync can take a few minutes.
Cause: Large model on limited hardware, or app running in debug mode
Fix: Try a smaller Whisper model in Settings (e.g., Tiny or Base). Make sure you're running the release build of My Transcriber, which enables Metal GPU acceleration for 100x+ faster transcription compared to CPU-only mode.
Cause: Microphone permission not granted
Fix: System Settings → Privacy & Security → Microphone → Enable for My Transcriber. The app will request this permission the first time you try to record.
Cause: Network issue during Hugging Face download
Fix: Restart My Transcriber. The download resumes automatically. If the problem persists, check your internet connection and try again. Model files are cached, so a partial download won't start from scratch.
If you're having trouble getting started or have questions about My Transcriber, we're here to help.