Start here
- Choose native Windows or a WSL environment.
- Install dependencies using the current official instructions.
- Verify the version before processing a small scan.
Choose one environment
OCRmyPDF currently documents both native Windows and WSL installation. Native setup includes Tesseract and Python tooling, with Ghostscript relevant to some output routes. WSL runs the Linux installation. Keep the environment explicit: mixing Windows Python packages with Linux executable paths makes diagnosis harder.
| Route | Check first |
|---|---|
| Native Windows | 64-bit dependencies and executable discovery |
| WSL | Linux package setup and access to the input path |
| Graphical workflow | NAPS2 may avoid command-line setup |
Follow the version’s installation instructions
Use the official installation page rather than an old command copied from a forum. Dependency recommendations can change. Complete that setup before testing the input document. If the version command fails, resolve installation and PATH issues before changing OCR settings.
Run a small test
These examples assume the program is already installed. In WSL, use a Linux-accessible path; change to the appropriate directory before running. Choose a separate output name.
ocrmypdf --version
ocrmypdf input.pdf searchable.pdfWSL invocation
The documented Windows invocation prefixes the command with wsl. Confirm which Linux installation is used and where the files are located. A path inside the Linux environment and a Windows drive path are not interchangeable.
wsl ocrmypdf --version
wsl ocrmypdf input.pdf searchable.pdfValidate content and retain the input
Open the result, search for a known phrase, and copy a line containing numbers. If the document already has a text layer or an option is rejected, consult the current help and cookbook for that case. Do not add overwrite or force options until you understand the reported failure. Keep the untouched input as the recovery copy.
Sources
Updated October 30, 2026. Based on the linked documentation; the tools have not been benchmarked by NarrowTools. Editorial policy
