Wispa 🎤

Fast, local voice-to-text for macOS using faster-whisper. Hold a hotkey to record, release to transcribe and inject text into any input field. The most simple privacy-focused voice-to-text solution for macOS.

Features

• Ultra-fast transcription with faster-whisper small model optimized for Apple Silicon
• Hold-to-record interface (Cmd+Shift+Space by default)
• Local processing - no internet required, completely private
• Automatic text injection into focused input field using AppleScript
• Voice Activity Detection to filter out silence

Requirements

• macOS (Apple Silicon M1/M2/M3)
• Python 3.9+
• Microphone access

Installation

1. Install dependencies:

pip install -r requirements.txt

2. Grant permissions:
   • Microphone Access: You'll be prompted on first run
   • Accessibility Access: System Settings > Privacy & Security > Accessibility
     • Add Terminal (or your Python IDE) to the list

Usage

Run the script:

python main.py

To use:

1. Hold Cmd+Shift+Space to start recording
2. Speak your text
3. Release the keys to stop and transcribe
4. Text will be automatically typed into the focused input field

Exit: Press Ctrl+C in the terminal

Configuration

Edit main.py to customize:

wispa = Wispa(
    model_size="small",  # Options: tiny, base, small, medium, large-v3
    hotkey="<cmd>+<shift>+<space>"  # Change hotkey combination
)

Language: Change line 115 from language="en" to your language code, or None for auto-detection

Performance

On Apple Silicon M1:

• Model loading: ~2-5 seconds (one-time at startup)
• Transcription: ~1-3 seconds for typical voice clips (5-10 seconds of speech)
• Memory usage: ~500MB-1GB

Troubleshooting

"No audio recorded!"

• Check microphone permissions in System Settings

Text not injecting:

• Grant Accessibility permissions to Terminal/IDE
• Try clicking into the input field before recording

Slow transcription:

• Use tiny or base model for faster results
• Reduce cpu_threads if CPU usage is too high

Credits

Built with:

• faster-whisper - Fast Whisper implementation
• pynput - Keyboard listener
• sounddevice - Audio recording