Got a question?
We've got answers.

A Mac with Apple Silicon (M1 or later), macOS Ventura 13.0 or later, and at least 8 GB of RAM. Intel Macs are not supported — the on-device AI inference relies on the Neural Engine in Apple Silicon chips.

Download the .dmg from the releases page, open it, and drag Tellaflow to your Applications folder. On first launch, macOS may show a security warning — go to System Settings → Privacy & Security and click "Open Anyway". This is normal for apps distributed outside the App Store.

No. Tellaflow has no accounts, no login, no email required. Download, open, done.

Open Tellaflow, grant microphone access when prompted, and download your first model from the Models tab. We recommend starting with Whisper small — it's a good balance of speed and accuracy. Then set your hotkey and test it in a text field.

Whisper small is the best starting point for most users. It downloads quickly (~250 MB), runs fast on any M1+ chip, and has solid accuracy. If you need higher accuracy and have an M2 or later, try Whisper medium or large.

Parakeet (parakeet-tdt-0.6b-v2) is NVIDIA's open-source ASR model. It's smaller and faster than Whisper large but optimized specifically for US English. If you speak US English and want the fastest possible transcription, Parakeet is often the best choice. It does not support translation.

Models are stored in ~/Library/Application Support/Tellaflow/Models/. You can delete them from there if you want to free up disk space, or remove them from within the app via Settings → Models.

Whisper tiny: ~75 MB. Whisper base: ~145 MB. Whisper small: ~250 MB. Whisper medium: ~780 MB. Whisper large: ~1.5 GB. Parakeet-tdt-0.6b-v2: ~600 MB.

Yes. You can download as many models as you want and switch between them in the app. Only one model is active at a time.

In the Tellaflow app, go to Settings → Transcription and toggle "Live Translation" on. This switches Whisper into translation mode. Note: Live Translation requires a Whisper model — Parakeet does not support it.

Any of the 100+ languages supported by Whisper. Spanish, French, German, Japanese, Korean, Mandarin, Hindi, Arabic, Portuguese, Italian, Russian, Turkish, and many more. The output is always English.

Not yet. Whisper's translate mode outputs English specifically. Bi-directional translation is on the roadmap.

Translation is slightly slower than transcription because Whisper is doing more work. If you need faster results, try a smaller Whisper model (small or base). The trade-off is accuracy on accented speech.

Make sure Tellaflow has Accessibility permissions. Go to System Settings → Privacy & Security → Accessibility and ensure Tellaflow is listed and enabled. Without this, it can't inject text into other apps.

Another app may be capturing the same shortcut. Check System Settings → Keyboard → Keyboard Shortcuts for conflicts. You can change Tellaflow's hotkey in Settings → Hotkey.

A few things to try: (1) Try a larger model — tiny and base are fast but less accurate. (2) Reduce background noise. (3) Speak closer to the microphone. (4) If your accent isn't well-represented, Whisper medium or large tends to handle variety better.

Run this command in Terminal: xattr -cr /Applications/Tellaflow.app — this clears the quarantine flag macOS sets on downloaded apps. Then try opening again.