iPhone & iPad
On iPhone: Live Text reads French
- Open the photo of the French document in Photos.
- Tap the Live Text button in the bottom-right corner.
- Tap “Copy All” and paste where you need it.
Copy French from invoices, letters, certificates and scanned PDFs with accents and ligatures preserved, instead of an English reading that strips them. Nothing is uploaded, and there’s no account or watermark.
Drop your French images or PDFs onto the drop zone, or tap “or tap to choose files”.
French is preselected under “Text language”. Add Arabic for bilingual Maghreb documents, or English if English passages appear alongside the French.
Choose “Join into paragraphs” if you plan to paste the text into a letter or a report, then press “Extract text”.
Read through it in “Show text”, paying attention to names and figures, and use “Copy text” or “Download .txt”.
OCR engines don't just look at shapes; each language model also knows which letters and words are likely. Read with the English model, French loses its accents. In our test, “cœur” came out as “coeur”, “déçu” as “dégu” and “îles” as “iles”. The French model knows é, è, ê, à, ç, ù, î and ô, as well as the œ and æ ligatures, and gets these right on clean print.
That is why this page opens with French already selected, whatever language you browse in. The French model is small, about 0.6 MB, and is downloaded once, then kept by your browser.
Typographic details are a different matter. French puts a space before ; : ! and ?, and uses « guillemets » for quotation marks. The engine may output those as ordinary spaces or straight quotes, which is fine for searching or pasting into an email but worth checking if the text is going to be printed or published.
iPhone & iPad
Mac
Windows
On a French invoice or letter, the English model reads most of the words but quietly drops or swaps accents, and the errors are easy to miss: “a” instead of “à”, “ou” instead of “où”, “eleve” instead of “élève”. With the French model the same page usually needs only a quick check. Choose the language of the document, not the language of the website.
Typical French paperwork recognizes well: invoices (factures), certificates (attestations), administrative letters (courriers) and scanned PDFs sent by public offices. Tables of amounts come out as lines of text, so verify the totals, IBANs, SIRET and reference numbers digit by digit, and check proper nouns, which no dictionary can correct.
Documents from Morocco, Algeria or Tunisia often combine French and Arabic on the same page. Check both languages under “Text language”: each script is then read with its own model. For a page entirely in French, French alone is faster and slightly more accurate.
That is the English model at work. Make sure French is checked under “Text language” and extract again; the French model knows the œ ligature and the accented letters.
Yes. Photograph it flat in good light, or drop the PDF you received. Pages that are already digital are copied exactly, and scanned pages are recognized with the French model.
Not always. « Guillemets » may become straight quotes, and the thin space before ; : ! ? may become a normal space or disappear. Fix them with find and replace if the typography matters.
Check both French and Arabic. The engine can then read each script with the right model. It is slower than one language, so for pages that contain only French, uncheck Arabic.
Printed pages from books and handouts work well if the page is flat and evenly lit. Curved pages near the spine and small footnotes are less reliable, so photograph one page at a time and fill the frame.
No. Recognition happens on your device inside the browser. Invoices, payslips and certificates never leave your computer or phone.
Last updated October 28, 2026
Enlarges small text, turns dark backgrounds light and evens out shadows. Try it off if letters come out broken.
Up to 50 scanned pages are recognized per PDF.
Pages with real text are copied exactly and instantly, without OCR. Tick this if that text comes out garbled.