Turn an image or a scan into text

Features

• Images and scanned PDFs

  Add JPG, PNG, WebP, BMP, GIF or AVIF pictures, or PDFs whose pages are scans. Every page of a PDF is read in turn.

• English and Bengali

  Choose English, Bengali or both for documents that mix the two. The recognition models are loaded from this site the first time and are not sent anywhere.

• Straightens and cleans the page

  Slightly tilted scans are straightened, uneven paper colour and shadows are evened out, and photos and ruling lines are hidden so they do not confuse the reader.

• Paragraphs, not just lines

  Lines that belong to one paragraph are joined into flowing text, or you can keep the original line breaks. Page markers separate the pages of a PDF.

• Batch processing

  Recognise up to 20 files in one run and download the TXT files one by one or as a ZIP.

How to use OCR: Image or Scan to Text

1. 1Choose pictures or scanned PDFs, or drop them onto the page.
2. 2Pick the languages that appear in the document.
3. 3Select Recognise text and wait; each page takes a few seconds.
4. 4Read the result, then download the TXT files or a ZIP.

Supported formats

Best results come from sharp, upright pages of printed text at about 300 dpi.

Limits and things to know

• OCR makes mistakes. Blurry, skewed, low-resolution, handwritten or very small print and decorative fonts are read poorly, and Bengali conjunct letters are harder than English. Always proofread.
• Only English and Bengali are included. Text in other languages comes out wrong.
• Pages must be upright. A sideways or upside-down scan is not rotated for you.
• Tables, columns and layout are not kept; you get running text. For a Word file with the layout, use PDF to Word.
• The first run loads about 4 to 6 MB of recognition data, and large scans use a lot of memory. Up to 20 files, 100 MB each and 300 MB in total.

Privacy and processing

The pictures are decoded and read in your browser by an OCR engine that runs as WebAssembly in a worker. The engine and the language models are served by this site itself. Your files are not sent to FilesKit, which has no server that receives files.

Common uses

• Copy text from a photo

  Pull the words out of a photographed page, a sign or a whiteboard to paste into a message or a note.

• Search scanned documents

  Turn old scans into text you can search, index or quote.

• Digitise printed pages

  Retype nothing: get the text of a printed article, form or letter into a file you can edit.

• Read Bengali print

  Extract Bengali text from a scanned book page or notice so it can be searched or translated.

How this differs from other approaches

Online OCR services upload your pictures, which is risky for contracts, IDs and private letters. Phone apps and desktop OCR programs can be more accurate on difficult scans and support many more languages. FilesKit keeps everything on your device and covers English and Bengali.

Frequently asked questions