6. Voice Input
Convert your speech into keystrokes sent to the target computer — hands-free typing and accessibility.
How It Works
- Tap the microphone button
- Speak what you want to type
- Your speech is converted to text
- The text is sent as keystrokes to the target computer
STT Engines
| Engine | How it works | Setup |
|---|---|---|
| System Recognizer (Android) / Apple Speech (iOS) | Uses built-in speech recognition | Requires Google Voice Typing (Android) or network connectivity (iOS) |
| Whisper (both) | On-device or cloud AI transcription | Android: set API key in Settings. iOS: download on-device model files |
Supported Languages (iOS)
15 languages including English (US/UK), Chinese (Simplified/Traditional), Cantonese, Japanese, Korean, French, German, Spanish, Italian, Portuguese, Russian, Arabic, and Hindi.
Silence Detection & Auto-Pause
Both engines feature automatic silence detection that pauses recording when you stop speaking (2.0 second silence timeout by default). Toggle Auto-Pause on/off in the Voice Input view.
Troubleshooting Silence Detection
| Symptom | Solution |
|---|---|
| Recording continues when not speaking | Check the Auto-Pause on Silence toggle. Reduce background noise. Speak clearly and close to the microphone. |
| Recording stops immediately | Speak more loudly or reduce the silence detection timeout. |
Mini Toolbar (Android)
| Button | What it does |
|---|---|
| Copy | Copy transcribed text to clipboard |
| Auto Send | Automatically send text after transcription |
| Auto Line Return | Append an Enter keystroke after sending |
| AI Refine | Send transcribed text to AI for improvement |
Whisper Model Management (iOS)
On-device Whisper models trade accuracy for file size:
| Model | Size | Accuracy | Speed |
|---|---|---|---|
| tiny | ~75 MB | Basic | Fastest |
| base | ~142 MB | Good | Fast |
| small | ~466 MB | Better | Moderate |
| medium | ~1.5 GB | Best | Slowest |
Start with base — it's a good balance of accuracy and speed.
If model downloads fail or stall, check network connectivity and available storage. Try a smaller model first (tiny or base).
Permissions (iOS)
Voice Input requires Microphone and Speech Recognition permissions. If denied, enable them in Settings → Privacy & Security.
Voice Text Not Sending (Android)
Check the connection status. The "Send" button is disabled when not connected.
Next Steps
- ← Macros — Automated key sequences
- AI Integration → — AI-assisted text refinement and command assistant