What PDF OCR helps you do

Run optical character recognition on a scanned PDF and get the recognized text back as plain text. It renders each page to an image, reads it with a local Tesseract engine, and labels the output by page. It does not write a text layer back into the PDF.

OCR runs in your browser with a local recognition engine. Your document is not uploaded to Haivix.

How it works

A PDF stores positioned drawing instructions rather than flowing paragraphs, which is why editing one is different from editing a document. Page content, fonts, images and annotations are separate objects, and a page only carries a text layer if the producer wrote one. Most tools here rebuild or render the file in your browser with pdf-lib and pdf.js; the formats that need a full office engine, such as Word, Excel, PowerPoint, PDF/A and HTML, use the Haivix service and say so in their privacy note. Every tool writes a new file and leaves the original untouched. Because content is placed by coordinates, extraction often returns text runs in visual order rather than sentences, and columns or tables can arrive interleaved.

How to use PDF OCR

Load the scanned PDF and start the OCR run.
Wait while each page is rendered and recognized.
Read the text blocks, which are separated by page markers.
Copy the text or download the .txt file.

When to use it

Make a scanned document searchable and copyable.
Extract text from a photo or scan that has no text layer.
Digitize a printed document for editing.
Recover text from an older scanned report.
Turn a screenshot-heavy PDF into selectable words.
Read a page whose font or encoding blocks normal extraction.

When not to use it

Do not expect a searchable PDF; the output is plain text.
Do not rely on it for handwriting or decorative type.
Do not use it where layout and tables must be preserved.

Example

A scanned page with no selectable text becomes copyable once OCR has recognized its characters. The output begins with a page marker such as Page 1, followed by the recognized lines. A poor scan with skew or low contrast produces garbled words, and a long document takes noticeably longer because recognition runs on your device.

What you can do with PDF OCR

Recognize text in image-based PDFs.
Process every page in sequence.
Show progress page by page.
Label the output with page markers.
Copy the text or download it as .txt.
Run the recognition locally in the browser.

Example: work on a copy and check the output

Keep the original file. These tools produce a download rather than editing in place, so a failed operation never damages the source.

Check whether the PDF has a real text layer before relying on extraction. A scan is an image, so text tools return little or nothing until OCR has run.

Unlock encrypted files first. A password-protected PDF cannot be reordered, cropped or annotated until it has been opened with the correct password.

Review the output before sending it on. Fonts can substitute, overlays do not always remove what is underneath, and page boxes can shift content near the edges.

Note the page size and orientation before laying anything out. A mixed document can contain portrait and landscape pages, and a tool that assumes one size may crop or offset the others.

Keep a note of the operation and its settings. If a recipient questions the result, reproducing it from the original is faster than guessing which options were used.

What to know before you start

Recognition is not perfect, especially on poor scans, handwriting or unusual fonts.
OCR runs in the browser and can be slow for long documents, and is limited to the loaded language model.
The result is plain text, not a searchable PDF with a text layer.
Layout, columns and tables are flattened into reading order.

Common mistakes

Expecting the original PDF to gain a text layer.
OCR-ing a page that already has a text layer instead of extracting it directly.
Assuming non-English text is supported; the loaded model is English.
Interrupting the run and losing the partial result.

Troubleshooting

If the output is empty, the page may be blank or the image too low-contrast.

If the browser becomes slow, close other tabs; recognition is CPU-heavy.

If characters are garbled, try a cleaner scan or a higher-resolution source.

If a page fails, continue with the rest and retry that page separately.

PDF OCR