how_it_works.md

How it works

1. The application starts when the user launches the app, which loads the GUI defined in gui.py. The GUI is built with Tkinter and displays the main interface.
2. The GUI sets up a hotkey listener, configured in src/hotkey_combination.py. The hotkey functionality is implemented using the pynput library.
3. When the user presses the hotkey, the app records audio using code in src/audio_recorder.py. Audio recording is handled by the sounddevice library, with data processing by numpy and scipy.
4. The recorded audio is transcribed into text using the Whisper model via faster-whisper, called from src/transcriber.py. This step uses the faster-whisper Python package (which internally uses the Whisper Model by OpenAI with CTranslate2 for improved performance).
5. The transcribed text is copied to the clipboard by code in src/clipboard_manager.py, which uses the pyperclip library for clipboard operations.
6. The tool automatically pastes the copied transcribed text at the currently focused app window where the cursor is.

Privacy and Data Retention

Open-WhisperScribe is a local-only application, meaning all processing happens on your device. No data is sent to external servers, ensuring complete privacy. Below are details about data retention:

1. Recordings:

   • Recorded audio is saved as the base name defined in config.yaml under the audio_file key with the .wav extension automatically appended (default is output.wav) in the root directory of the application.
   • This file is overwritten each time a new recording is made.
   • If you wish to retain a recording, make a copy of the file before starting a new recording.

2. Logs:

   • Logs are written to the nohup.out file in the root directory.
   • This file contains information about the application's runtime behavior and is useful for debugging.
   • You can delete this file at any time without affecting the application's functionality.