Image to Text

How it works

1. Step 1

   Upload

   Drop in images or PDFs. Every page of a PDF is split out automatically.

2. Step 2

   Set the language

   Pick the language the transcription should come back in.

3. Step 3

   Test the prompt

   One random page is transcribed so you can judge the result before committing.

4. Step 4

   Run and edit

   Approve the prompt, let every page transcribe, then correct anything by hand.

Built for documents OCR gets wrong

Frequently asked questions

What file types can I convert to text?

PNG, JPG, JPEG, WEBP, HEIC and HEIF images, plus multi-page PDFs. A PDF is split into pages automatically and every page is transcribed on its own, so a 200-page scan comes back as 200 separate transcriptions you can read and edit individually.

Does it work on handwriting?

Yes. Handwritten letters, field notebooks, ledgers, minutes and archival manuscripts are the main use case. Because you write the transcription prompt yourself, you can tell the model how to treat abbreviations, unclear words and original spelling instead of accepting a generic OCR result.

How is this different from ordinary OCR?

Traditional OCR matches character shapes and fails on handwriting, faded ink and unusual layouts. Evidano reads the page with a vision language model, so it can follow columns, tables and marginalia, and it can mark uncertain readings rather than silently guessing.

Can I test the prompt before running a whole document?

Yes. Before a project starts, one page is sampled at random and transcribed with your prompt so you can see the result. Edit the prompt and try again as often as you like; only when you are happy does the prompt run across every page.

Can I get the text in another language?

Choose an output language from about a hundred options and the transcription is returned in that language. A German archival letter can come back in English, or an English report in Hindi, without a separate translation step.

Can I correct a transcription afterwards?

Every page can be edited in place. Open the file, compare the image with the text side by side, fix anything the model misread and save. Your correction replaces the stored transcription.

Is my document kept private?

Files are processed for your account only, are never used to train models, and can be deleted at any time — individually or a whole project at once.