WhisperDrop

Speak. Drop. Done.

WhisperDrop is a tiny, fast speech-to-text tool for Linux and Windows. It sits as a dark floating widget on your screen -- press F8 anywhere, speak, and your words drop right at the cursor. No copy-paste, no switching windows, no hassle. macOS is not supported.

How It Works

1. Press F8 (or click REC)
2. Speak naturally
3. Press F8 again (or click STOP)
4. Your words drop at the cursor -- in any app, instantly

Features

• Instant text drop -- words appear at your cursor via xdotool (Linux) or clipboard paste (Windows)
• GPU accelerated -- CUDA with automatic CPU fallback
• Global hotkey -- F8 works from any application, no sudo needed
• Compact dark UI -- frameless floating widget, draggable, always on top
• Live waveform -- visual audio feedback while you speak
• Whisper powered -- uses OpenAI Whisper (small, 464MB) via faster-whisper for fast and accurate transcription

Requirements

• Python 3.12+
• Linux (X11) or Windows — macOS is not supported
• Linux: xdotool installed
• Windows: No extra system deps
• CUDA GPU (optional, falls back to CPU)

Install

git clone https://github.com/AnandBhandari1/WhisperDrop.git
cd WhisperDrop

# Install uv if you don't have it
# Linux: curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows: https://astral.sh/uv

# Linux only: install xdotool
# sudo apt install xdotool

# Install python dependencies
uv sync

Run

Linux:

./run.sh

Windows (CPU only):

run_cpu.bat

Windows (CUDA GPU):

run_cuda.bat

Project Structure

WhisperDrop/
├── app.py             # Linux application (xdotool)
├── app_windows.py     # Windows application (pyautogui)
├── run.sh             # Linux run script
├── run_cpu.bat        # Windows CPU run script
├── run_cuda.bat       # Windows CUDA run script
├── pyproject.toml     # Dependencies
├── uv.lock            # Lock file
├── LICENSE            # MIT License
└── README.md          # This file

Troubleshooting

Mic not working -- Check microphone permissions and default input device in system settings.

No GPU -- App automatically falls back to CPU. Install NVIDIA drivers + CUDA toolkit for GPU acceleration.

F8 not responding -- Make sure no other app is capturing F8. The hotkey uses pynput and works without sudo.

Text not inserting (Linux) -- Ensure xdotool is installed (sudo apt install xdotool). Only works on X11, not Wayland.

Text not inserting (Windows) -- Ensure the target window is focused before transcription finishes. The app uses Ctrl+V to paste.

Running on macOS -- WhisperDrop does not support macOS. Use Linux or Windows.

License

MIT