How to extract text from a PDF offline

1. Check that the PDF contains text

Open the PDF and try selecting a sentence. If you can select individual words, the page probably contains a text layer. A scan may show readable words while storing only an image of them.

Omia does not perform OCR. If the pages are images with no text layer, ask for the original Word or digital PDF, or use an OCR tool first. A PDF can also mix text pages and scanned pages, so check more than the first page.

2. Save as plain text or Markdown

1. Install Omia using the install guide, then drag the PDF onto its window.
2. Move the pointer over the window to show its controls. Open Save As.
3. Choose Plain text for a .txt file, or Markdown (extracted text) for a .md file.
4. Choose a filename and location in the save dialog, then save.

The export is a new file. It does not overwrite the PDF. Open the saved .txt or .md in Omia when you want to read or edit the extracted content.

3. Choose the output for your next task

• Plain text: useful for copying words into notes, email or another editor.
• Markdown: useful for a .md workflow. Headings, paragraphs and lists may be inferred from the PDF; check the result before using it.
• Keep the PDF: when page layout, tables, diagrams or a drawing's appearance matter. Text extraction does not preserve those visual relationships.

For a scan that you only need to share as pictures, Save As → PNG pages creates a ZIP. It exports up to the first 50 pages and adds a README for longer documents. Image export is separate from text recognition.

4. Review the text against the original

Compare the beginning, middle and end of the saved text with the PDF. Look for missing words, unexpected reading order, repeated page headers, column mixing and table values separated from their labels. Correct the extracted file before sharing it or passing it to an AI tool.

As a concrete example, our pavilion DWG export contains searchable labels including “GROUND PLAN” and “Lakeside Tea Pavilion”. We checked that existing PDF on October 10, 2026. Those labels can be extracted, but text alone does not reproduce the floor plan, dimension lines or spatial arrangement.

If the saved text is empty or incomplete

• Image-only pages: use OCR first; saving them as Markdown cannot create a text layer.
• Some pages work, others do not: check whether the PDF mixes scanned and digital pages.
• Garbled words: the PDF may have unusual font encoding. Compare against the original and ask the sender for a text export if necessary.
• Tables or columns are rearranged: review the reading order, or use the original document when exact structure is required.