VoxFlow Support

VoxFlow uses WhisperKit, an on-device speech recognition engine powered by OpenAI's Whisper model optimized for Apple Silicon. All transcription happens locally on your device's GPU and Neural Engine — no internet connection required, no audio sent to any server.

Yes, 100%. VoxFlow has no servers, no backend, and no cloud infrastructure. Your audio is processed entirely on your device. We have zero access to your recordings or transcriptions. See our Privacy Policy for full details.

VoxFlow supports 99 languages via WhisperKit, including English, Spanish, French, German, Chinese, Japanese, Arabic, Hindi, Slovak, Czech, and many more. You can select your preferred language in Settings, or use Auto-detect mode.

VoxFlow uses Apple's CloudKit to sync recordings across your iPhone, Mac, and Apple Watch. Sync happens directly between your devices via your personal iCloud account. GoSpace Labs has no access to your synced data. You can disable sync in Settings.

VoxFlow offers three Push-to-Talk triggers on Mac, each independently toggleable in Settings:

• Globe/Fn key — Hold for 0.7 seconds, speak, release. Text is pasted at your cursor.
• Caps Lock — Toggle ON to start recording, toggle OFF to transcribe and paste.
• Right Shift — Hold for 0.3 seconds, speak, release. Text is pasted at your cursor.

Push-to-Talk requires Accessibility permission (System Settings → Privacy & Security → Accessibility). VoxFlow listens ONLY for these specific keys — it does not log keystrokes.

Follow these steps:

1. Open System Settings → Privacy & Security → Accessibility
2. Find VoxFlow in the list and ensure the toggle is ON
3. If VoxFlow is not in the list, click "+" and add VoxFlow.app
4. Restart VoxFlow after enabling the permission
5. If still not working, try removing VoxFlow from the list and adding it again

After recording, tap the AI provider button (ChatGPT, Claude, Gemini, Grok, or Perplexity). VoxFlow sends your transcription via Apple's Siri Shortcuts. On Apple Watch, this works with just one tap — no API tokens needed, you use your existing AI subscription.

VoxFlow offers five model sizes:

• tiny (~40 MB) — Fastest, good for quick notes
• base (~75 MB) — Fast, balanced quality
• small (~150 MB) — Recommended for most users
• medium (~400 MB) — Accurate, good for professional use
• large-v3 (~1.5 GB) — Most accurate, best for complex content

We recommend starting with small. Larger models give better accuracy but take longer to transcribe.

Yes! Go to Settings → Export and choose a folder (Synology, Dropbox, Google Drive, or any local folder). You can enable auto-export for new recordings or export all existing recordings at once.

Yes. VoxFlow is completely free with no subscriptions, no in-app purchases, no ads, and no data collection. Professional-grade transcription without the price tag.