Image to text
Extract text from a photo, scan or screenshot with on-device OCR. English for now. Copy or download the result. The image never leaves your browser.
Runs on your device. Nothing is sent to a server. The first run downloads a 7 MB recogniser once; later runs are instant.
How it works
Drop a clear photo or screenshot and press Extract text. The recogniser runs inside your browser; the first use downloads it once (about 7 MB) and then works offline.
Best results come from straight, well-lit images with dark text on a light background. Crop out unrelated areas before uploading and the accuracy goes up noticeably.
Questions
Which languages are supported?
English today. Arabic, Urdu, Hindi and others are planned; each adds a language file that downloads on demand.
Does it read handwriting?
Printed text only. Neat block capitals sometimes work, cursive does not.
Why is my photo processed slowly on a phone?
Recognition runs on your device's processor. A page of text takes 5 to 20 seconds on a mid-range phone. Nothing is sent to a server, which is the point.