ExtractFoxExtract text from handwriting
Drop a photo or scan of handwritten content — notes, letters, forms, historical documents — and get clean text back. Works on printed handwriting and most cursive in Latin scripts.
What should we pull from this handwriting?
Or pick specific fields
Or describe it yourself
Why this matters
Traditional OCR (Tesseract, ABBYY) was built for printed text and falls apart on handwriting. Modern vision LLMs read handwriting the way a human does — using context to disambiguate ambiguous letters — so accuracy on printed handwriting jumps from ~50% to over 90%. A genealogy project has 400 pages of 1940s cursive letters that Tesseract turns into unreadable soup — family names and dates are lost in the noise. Clinicians dictate patient allergies on paper charts during ward rounds; those notes sit in a filing cabinet until someone manually types them for the EHR migration.
How it works
- Step 1Upload the image or PDF
Photo of a handwritten note, scan of a handwritten form, or a PDF page. PNG, JPG, WEBP, HEIC, or PDF.
- Step 2Get the transcription
Plain text comes back preserving line breaks. Specify a layout (paragraphs, list, form fields) in the description box if needed.
- Step 3Copy or export
Copy the transcription straight from the result, or download it as a .txt file. For form-shaped handwriting, use the form data extractor for label-value pairs.
Common use cases
Sample output
Example: handwritten meeting notes from a whiteboard photo
Meeting notes — 2026-04-22 Agenda: - Q2 OKRs review - Hiring plan - Budget reforecast Decisions: 1. Ship onboarding redesign by May 15 2. Backfill the senior PM role this month 3. Cut Q2 marketing spend by 15% Actions: - Jane: draft revised OKRs by Friday - Mark: open requisition by EOW - Sam: present new budget to board next Tuesday
Frequently asked questions
How does handwriting recognition work?+
Modern handwriting recognition uses a vision-language model that reads the image as a whole — recognizing letter shapes in context. Unlike older character-by-character OCR, it disambiguates similar letters (a vs o, n vs h) using the surrounding word and sentence.
What is handwriting recognition software?+
Software that converts handwritten content (printed or cursive) into machine-readable text. ExtractFox is one example — others include built-in tools like the handwriting recognition in Notability, GoodNotes, and OneNote, which work for handwriting created on the device itself.
How accurate are handwriting recognition tools?+
On clearly-written printed handwriting in English, accuracy is typically over 90%. On clean cursive, 80–90%. On heavy cursive, faded ink, or non-Latin scripts, accuracy drops and benefits from human review.
Which app has the best handwriting recognition?+
It depends on what you're recognizing. For handwriting written directly on an iPad, GoodNotes and Notability are excellent because they capture stroke order. For images of handwriting (photos, scans, archival material), vision models like the one ExtractFox uses lead the field — they don't need stroke data.
Can it read historical handwritten documents?+
Yes — period-style cursive (19th- and 20th-century letters, ledgers, certificates) extracts well. Very old or specialty scripts (Old English, Sütterlin, palaeography) are harder and benefit from a description telling the model what era and language to expect.
Does it support handwriting in non-English languages?+
Yes. Latin-script languages (Spanish, French, German, Portuguese, Italian, Dutch, etc.) work well. Cyrillic, Greek, and CJK handwriting work but with lower accuracy than English.
How is this different from the image-to-text tool?+
Both run the same ExtractFox vision model, so raw accuracy on a photo of handwriting is the same either way. The difference is what you get back. This page is built around handwriting: it defaults to a full transcription with line breaks preserved, gives you sub-extractions written for handwritten content (names and dates, action items, translation), and returns text you can copy or download as .txt. The image-to-text tool is the general-purpose version for any image — screenshots, signage, labels, printed scans.
Can it transcribe handwriting mixed with printed form labels on the same page?+
Yes. Use the Handwriting only sub-extraction to skip printed labels and transcribe only the handwritten values. For label-value pairs, use the form data extractor instead.
How do I handle multi-page handwritten letters — one upload or page by page?+
Upload the full multi-page PDF in one go — text comes back with page breaks preserved. For very long documents, page-by-page extraction can be easier to review before merging.