How it works
AI trained on centuries of English handwriting
Transkribus uses deep learning trained on millions of handwritten pages from the 16th century to the present day. The AI recognises the distinctive letter forms and ligatures of historical English scripts — from the looped minims of secretary hand to the flourished capitals of copperplate — and converts them into modern, searchable text.
Reads secretary hand, court hand, copperplate, round hand, and more
300+ public models for different handwriting styles and time periods
Works with phone photos or high-resolution scans
Family research
Unlock centuries of UK and American records
Parish registers, wills, court rolls, census returns — the written record of the English-speaking world stretches back over five hundred years, but most of it is in handwriting that modern readers cannot decipher. Transkribus reads these documents automatically so you can focus on the history, not the palaeography.
Church of England parish registers (baptisms, marriages, burials)
Wills, probate inventories, and administrations
Manorial court rolls and quarter sessions records
Colonial-era American documents and plantation records
Census returns and civil registration certificates
Personal letters, diaries, and family bibles
Beyond the demo
The full platform for historical documents
The demo above gives you a taste. The full Transkribus platform lets you process thousands of pages, train custom AI models on your specific handwriting, search across all your documents, and export in any format you need.
Train custom models on your family's specific handwriting
Full-text search across all your transcribed documents
Export as TXT, DOCX, PDF, TEI-XML, or PAGE XML
Collaborate with family members and share collections
What you can transcribe
Common documents in old English handwriting
Whether you are tracing your family tree in the UK or researching colonial-era America, you will encounter a wide variety of handwritten documents. Transkribus handles them all — from densely abbreviated Latin-English parish entries to neatly penned Victorian letters.
Parish registers from the 1530s onwards
Wills and testamentary records at The National Archives or local record offices
Court rolls, depositions, and petitions
Early American land grants, deeds, and town records
Ship passenger lists and emigration records
Personal correspondence, diaries, and account books
Background
What are old English scripts?
English handwriting changed dramatically over the centuries. Documents from different periods look entirely different, and reading them requires understanding the script conventions of the time. Here are the main styles you will encounter:
Secretary hand – the dominant English hand from the late 1400s to the mid-1600s, with distinctive letter forms very different from modern writing
Court hand – a formal legal script used in courts and government from medieval times into the 1700s
Italic hand – introduced from Italy in the 1500s, gradually replacing secretary hand
Round hand and copperplate – the elegant, flowing scripts that became standard from the 1700s onward
Mixed hands – many 17th-century documents blend secretary and italic forms
The technology
How does English handwriting recognition work?
Handwriting recognition (HTR – Handwritten Text Recognition) uses deep learning neural networks to convert images of handwritten text into machine-readable characters. Unlike OCR for printed text, HTR must handle the infinite variation in human handwriting — different letter shapes, connected strokes, and personal styles that change across centuries.
Neural networks trained on millions of handwritten samples
Layout analysis detects lines and text regions automatically
Character-level recognition handles connected and cursive writing
Language models improve accuracy by understanding word context
AI Models for English Handwriting
Browse public models trained on historical English documents — secretary hand, copperplate, and more.
Ready to read old English handwriting?
Create a free account to process unlimited documents, train custom models, and unlock the full platform.
50 free credits every month – No credit card required
200M+Pages processed
500K+Users worldwide
300+Public AI models