PDF OCR.

Draw, move or resize the recognition rectangle in the page preview to exclude unrelated columns or borders, then choose Recognize selection. To process another page, return to the workspace, select that page and run recognition again; this is not a whole-document OCR batch.

Set the recognition language to the actual text. Clear, upright scans are easier to recognize than blurry or skewed images. Recognition uses a fixed 200 DPI; it cannot restore detail absent from the original.

Classic is a good choice for most printed documents and offers single-column or multiple-column layouts. Its Searchable PDF retains the selected image area and adds a text layer aligned with the recognized words for searching and selection.

Future is the default for supported languages. Its advanced AI recognition is optimized for difficult scans, some handwritten documents and complex multilingual text. Handwriting results depend on legibility and scan quality, so compare the recognized text with the source image. Its Simple PDF contains selectable text in a new layout instead of preserving the source image. Languages not supported by Future use Classic.

Both engines offer TXT and a genuine Word document (.docx), with clean, editable text and right-to-left formatting where appropriate. Choose the recognition language that matches your document.

The main Download button saves the format chosen in Settings. PDF, TXT and Word links below it offer the other formats from the same recognition run. Choose TXT or Word (DOCX) in Settings to open the source-image and editable-text panels, including Smart edit. Choosing PDF opens PDF preview and download.

The source panel shows the full page. When you recognize a selected area, the rest of the page is faded and a blue outline marks the selection. Zoom in to inspect small characters and use Fit to see the full page again. Compare names, numbers, punctuation and reading order with the image.

Corrections in the editable result update the main TXT or DOCX download. Copy beside the text copies the current result with a brief fading effect. Smart edit opens a larger window for AI actions and formatting. Text changes do not alter the source image or repair a previously generated searchable PDF layer.

Summarize text condenses a passage, Paraphrase text rewrites it, and Fix grammar requests language corrections. These actions replace the selected passage or, with no selection, the entire editor content. Suggest a title adds a heading above the document; a selected passage can guide the title without removing the body.

Each AI request accepts up to 600 words and 12,000 characters, with a 64-KB UTF-8 request limit. Select a shorter passage if a long result exceeds these limits. AI actions send that text to our servers only when you select Apply and request output in the same language.

Review proposed changes before keeping them. AI can omit details, alter meaning or introduce mistakes, especially when the input contains recognition errors. Undo returns to an earlier version. The main result is updated only after Apply Changes; running an AI action alone does not replace it.

Advanced editor provides paragraphs and three heading levels, sans-serif, serif and monospace fonts, font size, bold, italic and underline. Set text color or highlighting, use bulleted or numbered lists, adjust alignment and choose automatic, right-to-left or left-to-right direction. Clear formatting removes styling for a simpler result.

Formatting happens in your browser. Switching to Simple editor does not itself discard the advanced document; editing its plain text removes formatting. AI replacements insert unformatted text in the affected area, so apply final styling after reviewing the wording.

Use Formatted copy to keep this styling in an HTML download. Copy and Apply Changes transfer plain text. Formatting the result does not change the original PDF’s fonts, page layout or text layer.

Remove accents & vowel marks removes Latin accent marks and common Arabic vowel marks from the chosen passage or the whole editor. Normalize numbers standardizes Western, Arabic or Persian digit shapes according to the text’s script; it does not change their numeric values. These cleanup actions run locally in your browser.

Undo and Redo let you review editing steps, including AI changes. Restore OCR text or Restore translation returns the editor to the original result from processing. You can undo that restoration too. Cancel closes the window without applying its current changes to the main result.

Smart edit is a working editor, not a saved project. Download any text or formatted copy you need before reloading the page or starting over.

TXT contains the current plain text. Word document (.docx) is a genuine editable Word file, created with clean text and right-to-left paragraph and text formatting for RTL languages. After text corrections or Smart edit changes, Download sends the current text to our server to regenerate the Word file without repeating OCR. The Word file does not recreate the scan layout or keep Smart edit’s rich styling.

For headings, colors, lists and other Smart edit formatting, choose Formatted copy inside the editor. It downloads HTML that can be opened in a browser and printed with the browser’s print controls. This is an additional text document, not a searchable PDF or an edited source scan.

Smart edit accepts up to 100,000 characters in the editor; individual AI actions have the smaller limits described above. If you chose PDF output in OCR settings, use that separate PDF preview and download. Choose a TXT or DOCX recognition result when you want text corrections with Smart edit.