Text Extractor
Text Extractor reads text from anywhere on your screen: a Wi-Fi password on a conference slide, a command in a video tutorial, an error message in a screenshot. You drag a box, and the text lands on your clipboard.

Use it#
- Press Super+Shift+T from any app, or open Text Extractor in Voltkit and click Capture Region.
- The desktop shows its screenshot tool. Drag a box around the text.
- Voltkit recognizes the text on your computer and copies it to the clipboard. A notification confirms it.
- Paste it anywhere with Ctrl+V.
If you cancel the screenshot, Voltkit shows No screenshot was taken and nothing is copied.
The page#
| Control | What it does |
|---|---|
| Capture Region | Starts a capture, exactly like the keyboard shortcut. |
| Extracted text | Shows the last text captured, so you can read or select part of it. Text from the keyboard shortcut also appears here when the window is open. |
| Copy Text | Copies the extracted text again. It is disabled until something has been captured. |
| Character count | No text captured yet, then N characters once you have a result. |
| Description line | Shows the recognition language and the current shortcut. |
Text is copied to the clipboard automatically after every capture. Copy Text is for copying it again after you have copied something else.
Languages#
Recognition uses Tesseract with English by default. To read another language:
- Install its language pack, for example
sudo apt install tesseract-ocr-deufor German ortesseract-ocr-spafor Spanish. More than 100 languages exist. - Set the language in the configuration file:
[ocr]
lang = "eng+deu" # one or several languages joined with +
From the terminal you can choose per run: voltkit ocr --lang deu.
From the terminal#
voltkit ocr # select a region, text goes to the clipboard and is printed
voltkit ocr --lang eng+deu # several languages at once
voltkit trigger ocr # the same as the shortcut, through the background service
How region capture works#
Voltkit asks the desktop's Screenshot portal for an interactive region capture, which works on Wayland without extra permissions. If the portal is not available it tries, in this order, grim with slurp, gnome-screenshot, spectacle and maim. If none exist, it tells you which to install.
Tips#
- Zoom the text in the source (a video, a PDF) before capturing: bigger text reads better.
- Capture one paragraph at a time. Tesseract is most accurate on plain printed text with good contrast.
- Everything happens locally. The image is not uploaded anywhere.
Troubleshooting#
- Nothing is copied. Run
voltkit doctor. Tesseract or a language pack may be missing. - Garbled text. Install the language pack for the text you are reading and set
langaccordingly. - The shortcut does nothing. The background service may be off. See Global shortcuts and the background service.