Skip to the image to text tool

Image to Text — and it reads it aloud

100% private

Photograph a letter, a bill, a form or a medicine box. Get the words out as text you can enlarge, hear read aloud, and copy. Nothing is uploaded — the reading happens on your own device.

Take a photo of the text
or choose a picture or screenshot you already have
The text
24px
1 Your photo

Fill the frame with the writing and hold the camera square to the page. That does more for accuracy than anything else.

More options

Leave this on Automatic for a letter, a bill or a form. Switch to Scattered text for a medicine box or a product label, where the writing is in separate patches rather than laid out as a page.

Four steps

How to turn a photo into text

Photograph the text

Press Take a photo and use your phone’s own camera. Get plenty of light on the page, hold the phone square to it rather than at an angle, and let the writing fill the frame. Or pick a screenshot you already have.

Press Get the text

The first time, about 4 MB of reading engine downloads — that is the whole wait, and it only happens once. After that it is stored on your device and every read is quick, even with no connection.

Read it, or hear it

The text comes back large and high-contrast. Make it bigger with A+, or press Read aloud and it is spoken one sentence at a time, with the sentence being read highlighted so it is easy to follow.

Fix it, then copy it

No reader is perfect. Press Fix the text to correct anything that came out wrong — a digit in a dose or an account number is worth checking — then Copy the text to paste it wherever you need it.

Private by design

Why this cannot leak your letter

The photo never leaves

Your picture is read straight off your disk and drawn onto a canvas inside this tab. It is never put into a network request. Every other free OCR site works the opposite way round: your document is uploaded to their server, read there, and kept for however long their policy says.

The engine comes to you

Instead of sending your page to the text engine, the text engine is downloaded to you — about 4 MB, once. It runs on your own processor. That is why the first read is slower and why every one after it works with your wifi switched off.

No account, no catch

There is no sign-up, no email capture, no page limit, no credits and no watermark. Nopturnia pays for itself through affiliate links on its product buying guides, not through these tools.

The honest part — everything that does leave this page

Three things, and this is the complete list.

Google Analytics records that someone opened this tool and that a read finished. It never sees your photo, your text or your file name.

One anonymous counter tick. When a read completes, this page sends the single word image-to-text to our own counter so the “used N times” number below stays honest. No image, no text, no length, no file name.

Reading aloud can be the exception, and it is the one to know about. Some of the voices your browser offers do not run on your device at all — Chrome’s default voices in particular are synthesised on Google’s servers, which means using one sends the text of your letter to Google to be spoken. This tool therefore starts on a voice that runs on your device, and any voice that does not is grouped under Online and labelled at the moment you pick it. If the voice list says On this device, nothing is sent. It is your choice to make, but you should get to make it knowingly.

The details

Tool specifications

Input formats
JPG, PNG, WebP, HEIC (iPhone), AVIF, BMP, GIF
Output
Plain text — on screen, read aloud, or copied to your clipboard
Language
English only
Handwriting
No — printed and typed text only
Engine
Tesseract 5 (LSTM), compiled to WebAssembly, running in your browser
First use
Needs a connection — about 4 MB of engine and English model, downloaded once
After that
Works fully offline, including as an installed app
Maximum photo size
100 MB
Text size control
18px to 44px, remembered on this device
Upload
None. Your photo never leaves your device
Sign-up, watermark, page limit
None
Price
Free

Questions

Reading text from photos, answered

Is my photo uploaded anywhere?
No. The reading engine is downloaded to your browser and your photo is read on your own device. The picture is taken from your disk and never placed in a network request. The simplest proof: do one read, then turn off your wifi and do another — it still works, because the engine is already on your device by then.
Why is the first read slow, and the rest quick?
Because the first run downloads the reading engine and the English text model — about 4 MB in total — and that is the entire wait. It is stored on your device afterwards, so every read after it starts immediately and needs no connection at all. Sites that upload your photo instead pay a cost on every single image, forever; this pays it once.
Can it read handwriting?
No, and it is worth being straight about that. This engine is built for printed and typed text — letters, bills, forms, packaging, books, screenshots and signs. Handwriting needs a fundamentally different kind of model, and the ones that do it well are far too large to download to a browser. If the page is handwritten, this tool will produce nonsense rather than a poor attempt, so it is better not to try.
Does it handle a page with two columns?
Usually, yes. The engine analyses the page layout and reads columns in the right order rather than straight across. What actually breaks that is not the columns but the angle: a page photographed slightly skewed confuses the layout detection, which is why this tool automatically straightens the image before reading it. If a page still comes out jumbled, open More options and try One column. For a medicine box or a product label — writing in separate patches rather than a page — choose Scattered text instead.
Why does it sometimes say the text may be wrong?
The engine reports how confident it is in what it read, and when that average drops below a sensible threshold we tell you rather than presenting a guess as a fact. It usually means the photo was dark, blurred, angled, or the writing was too small in the frame. The three things that fix it, in order: more light, hold the camera square to the page, and move closer so the text fills the frame. A wrong digit in a medicine dose or an account number matters far too much to hide behind a confident-looking result.
Does reading it aloud send my text anywhere?
It depends on the voice, and this is the one thing on this page that can leave your device. Many browsers — Chrome especially — offer voices that are synthesised on the provider's servers rather than on your device, so using one sends the text to be spoken. This tool starts on a voice that runs on your device, and groups any voice that does not under Online, labelled at the point you choose it. As long as the voice list shows On this device, nothing is sent anywhere.
Can I correct a mistake before copying?
Yes. Press Fix the text and the text becomes editable at the same large size. Whatever you change is what gets copied and what gets read aloud. This exists specifically because of dosages, account numbers, reference codes and dates, where a single misread character changes the meaning entirely and the person reading it may have no way to notice.
Does it work in other languages?
Not yet — this version reads English only. Each additional language is a separate few-megabyte model to download, so adding them all would make the first run far slower for everyone. If there is a language you would use this for, tell us with the feedback button and we will look at adding it.
Does it work offline?
After your first read, yes — including as an installed app. The engine and the English model are kept on your device once downloaded. Only the very first read needs a connection, because that is when they arrive.
Do I need to sign up or pay?
No. There is no account, no email capture, no page limit, no credit system, no watermark and no paid tier. Nopturnia funds itself through affiliate links on its product buying guides, not through these tools.

Keep it handy

Save it, share it, or tell us what’s missing

Come back to it

Bookmark the page, or pin it to your saved tools so it sits at the top of the tools list on this device. Worth doing here more than elsewhere — the reading engine stays on your device, so the second visit needs no connection at all.

Did it misread something?

Tell us what came out wrong — a kind of document it struggles with, a language you needed, or a control you wanted and couldn’t find. The form opens with the tool already filled in.

Give feedback on this tool

More free tools

Other things you can do here

Every one of these runs the same way — on your device, with nothing uploaded.

New Character & Word Counter Live word, character and reading-time counts as you type or paste. Text New tool New Reduce Any File Size Squeeze a PDF or image down until it fits that stubborn upload limit. Documents New tool New Passport Size Photo Maker Exactly 630 × 810 and under 250 KB — the size the passport portal actually checks for. Images New tool New PDF to Image Turn any PDF page into a PNG or JPG you can drop into a slide, a doc or a chat. Documents New tool New Background Remover Cut the subject out of a photo — full resolution, no watermark, nothing uploaded. Images New tool New Images to PDF Turn photos, scans or screenshots into one tidy PDF you can actually email. Documents New tool