ParrotNotes LogoParrotNotes

Image to Text Converter

Pull the text out of a photo, screenshot or scan (a whiteboard, a business card, a printed page) in your browser; nothing is uploaded. A picture to text converter that runs on your own device: choose or paste the image, pick the language, and copy or save the text.

Drop a photo, screenshot or scan here, paste it, or

Paste a screenshot with Ctrl+V or Cmd+V anywhere on the page.

Language of the text

PNG, JPG, WebP, GIF or BMP, up to 20 MB. Photos larger than 2,400 px on the long side are reduced to that size before the text is read.

The first run downloads the engine and the language model once (a few megabytes); the browser keeps them for next time.

The picture stays on your device. The recognition engine and the language model are downloaded once from a content delivery network (jsDelivr) and cached by your browser; the image is not uploaded to ParrotNotes or anyone else.

How to convert an image to text

Three steps, no account. The first run downloads the recognition engine once; after that it is cached.

  1. 1

    Choose, drop, paste or photograph the picture

    PNG, JPG, WebP, GIF or BMP, up to 20 MB. On a phone, Take a photo opens the camera. A screenshot on the clipboard can be pasted straight onto the page. The file's first bytes are checked before anything is shown, so a renamed file is refused rather than guessed at.

  2. 2

    Pick the language of the text

    English, German or Spanish. The page starts with its own language; switch and the picture is read again. The language model is downloaded once per language and kept by the browser.

  3. 3

    Edit, copy or download the text

    The recognized text lands in an editable box with a word count and a confidence hint. Fix a name or a number if you need to, then Copy or Download .txt. Another image clears everything.

Got a recording instead of a picture? The Audio to Text Converter transcribes it.

What you get

  • Editable text, copy and .txt

    Plain text you can correct in place, copy to the clipboard or save as a text file named after the picture. Line breaks follow the picture; trailing spaces and runs of blank lines are tidied.

  • English, German and Spanish

    One language model per language, picked with a click. Mixed pictures (a German sign with an English brand name) read best in the language most of the words are in.

  • Paste a screenshot

    Press Ctrl+V or Cmd+V anywhere on the page with a screenshot on the clipboard and it is read like a file. Handy for a slide, an error message or a chat you cannot select text in.

  • Nothing uploaded

    The recognition runs in a Web Worker in your tab. The only downloads are the engine and the language model, from a content delivery network, once; the picture never leaves the device.

Business card scanner

A phone, a card and three steps. One card at a time reads better than a photo of the whole stack.

  1. 1

    Photograph one card, flat, filling the frame

    Lay the card on a plain surface, hold the phone straight above it and let it fill most of the picture. Daylight or a desk lamp from the side beats a flash, which glares on coated cards.

  2. 2

    Read it in the card's language

    Most cards are printed text, which reads well. Check the phone number and the email address against the card: a 1 and an l, a 0 and an O are the usual slips.

  3. 3

    Paste the name, title, phone and email where they belong

    Into the contact, the CRM record or the note from the conversation. Write what you talked about next to it the same day; a card without context is just a card.

Working a show? How to qualify trade show leads covers what to write on the back.

JPG to text, PNG to text, screenshots

The converter reads PNG, JPG (JPEG), WebP, GIF and BMP, the formats the Tesseract engine documents for pictures. A screenshot is usually PNG and reads cleanly because the text is sharp and straight. A photo is JPG; the more it looks like a scan (flat, even light, straight on), the closer the result gets to a screenshot. Very large photos are reduced to 2,400 pixels on the long side before recognition, which keeps the page quick and still leaves small print readable.

iPhone photos are often HEIC, which is not one of the formats the engine reads. Apple's own page gives two ways round it: set Camera > Formats > Most Compatible so new photos are saved as JPEG, or export an existing photo from Photos or Preview as JPEG or PNG. The file is checked by its contents, not its name, so a renamed file is refused with a plain message rather than read wrong.

Photo to text: what the camera needs

Recognition is only as good as the picture. Five things that move the result more than any setting.

  • Light, without glare

    Even daylight or a lamp from the side. A flash on a whiteboard or a glossy card washes out the strokes.

  • Straight on

    Hold the phone parallel to the page. A picture taken at an angle stretches the letters and loses lines at the far edge.

  • Fill the frame with the text

    Closer is better until the edges blur. Crop out the table, the hand and the window before you convert.

  • Contrast

    Dark text on a light background. A faded marker on a grey board reads worse than the same words photographed after a wipe and a rewrite in black.

  • One column at a time

    A two-column page or a board with boxes can come back in the wrong order. Photograph each column or box on its own when the order matters.

The whiteboard is usually the summary of a meeting; the meeting notes template is where the rest goes.

Handwriting to text: what to expect

Printed text works well. Handwriting is hit and miss: neat block capitals on lined paper often come back usable; joined-up writing, a hurried note from a call or a whiteboard written in a rush come back with gaps and guesses. The engine was trained mostly on print, and it has no idea what the word should be, only what the strokes look like.

If the note matters, treat the result as a first draft and read it against the picture, numbers first. For handwritten call notes the faster route is often to read them out loud into the Dictation Tool or the ParrotNotes app and let speech recognition do the typing.

What to do with the text

The photo of the whiteboard, the card from the booth and the printed price list all end the same way: as text in the place where the rest of the conversation lives. Paste it into the CRM record, the follow-up email or the meeting note, and add one line about what was said while you still remember it.

In ParrotNotes the meeting itself is already there as a transcript, a summary and the action items; the text you just pulled out of the picture goes into the same note, so the board and the discussion stay together.

Related guides: Note-taking in meetings, Trade show leads, Trade show checklist, Meeting notes template.

The picture is one page; the meeting is the rest

ParrotNotes records the meeting or the call on your phone and gives you the transcript, a summary and the action items, with unlimited written notes to paste the whiteboard text into, free for 100 minutes of recording a month.

Get ParrotNotes free

Related tools

Image to text converter FAQ

Sources