aidful-whisper-typer

This is a Python script using openai/whisper to transcribe speech to text. After starting the script, use the configured keyboard shortcut to start/stop recording. The transcribed text will be handled according to your settings (typing, pasting, or copying to clipboard).

Note: This tool has been primarily tested on Linux systems using PulseAudio. While it should theoretically work on other operating systems due to the cross-platform nature of the used libraries, your experience may vary. The sounddevice library which is used to record the audio supports multiple backends:

• PortAudio
• ALSA (on Linux)
• PulseAudio (on Linux)
• Core Audio (on macOS)
• WASAPI/DirectSound (on Windows)

Setup Instructions

Step 1:

Download and install ffmpeg and python3.11 (might work with other versions, but this is not tested)

Step 2:

Clone the repository and change into the directory:

git clone https://github.com/AidfulAI/aidful-whisper-typer.git
cd aidful-whisper-typer

Step 3:

Create and activate a virtual environment and install the requirements:

python -m venv .venv
source .venv/bin/activate  # Linux/macOS
.venv\Scripts\activate     # Windows
pip install -r requirements.txt

Using a virtual environment keeps dependencies isolated from your system Python installation and other projects. This prevents version conflicts and makes the project more portable and reproducible.

Step 4:

Adapt the settings.json file in the project directory.

Step 5:

python3 aidful-whisper-typer.py

Usage Instructions

1. Press Ctrl+Alt+S (default) to start recording
2. Speak clearly into your microphone
3. Press Ctrl+Alt+S again to stop recording and start transcription
4. Based on your settings.json configuration, the text will be:
   • Typed out character by character (type mode)
   • Pasted at cursor position (paste mode)
   • Copied to clipboard (clipboard mode)

Autostart

If the script works as intended for you, you might want to add it to start at boot.

Step 1:

Adapt the run-aidful-whisper-typer.sh shell script which activates the virtual environment and runs the Python script to match your system paths.

Step 2:

Configure your system to run the shell script at startup.

Features

• Configurable keyboard shortcuts for starting/stopping recording (default: Ctrl+Alt+S)
• Multiple output modes:
  • type: Types out the text character by character
  • paste: Pastes the text at once using Ctrl+V
  • clipboard: Only copies the text to clipboard
• Different Whisper model options (tiny, base, small, medium, large, turbo). Best based on available VRAM of GPU:
  • turbo: ~6GB of VRAM and ~8x relative speed at highest quality
  • small: ~2GB of VRAM and ~4x relative speed at good quality
  • base: ~1GB of VRAM and ~7x relative speed at ok quality
• Audio transcription logging (can be turned off)
• Sound notifications for recording start/stop