How it works
AI that reads centuries of Dutch handwriting
Transkribus applies handwritten text recognition (HTR) models trained on millions of historical pages. The AI has learned the distinct characteristics of Dutch scripts across five centuries – the angular strokes of Gothic cursive, the flowing loops of clerks' running hands, and the compact abbreviations used by notaries and town scribes.
Handles Gothic cursive (gotisch cursief), kanselarijschrift, and old notarial hands
300+ public models covering different periods, regions, and writing styles
Upload phone photos, flatbed scans, or digitised microfilm
What you can transcribe
Whatever Dutch record you have, the AI can read it
Tracing ancestors in a Frisian village? Researching VOC ship logs? Sorting a box of family papers from Flanders? Almost every Dutch record before 1900 is handwritten – Transkribus turns those pages into searchable text.
Church records (DTB) – doop-, trouw- en begraafregisters going back to the late 1500s
Notarial acts (notariele akten) – wills, contracts, inventories, and property transfers
VOC and WIC records – trade correspondence, ship logs, and colonial administration
Tax rolls (kohieren) and census lists (volkstellingen)
Court records (rechtbankverslagen) and guild registers (gildenarchieven)
Personal letters, diaries, and family account books

Beyond the demo
Process entire collections, not just single pages
The demo above works on one page at a time. The full Transkribus platform handles batch uploads of thousands of pages, lets you train a custom AI model tuned to a specific archive's handwriting, and gives you full-text search across everything you have transcribed.
Batch-process entire archive boxes in a single upload
Train custom models on the specific handwriting in your collection
Full-text search across all transcribed documents at once
Export as TXT, DOCX, PDF, TEI-XML, or PAGE XML

Genealogy and research
Common documents in old Dutch handwriting
If your ancestors came from the Netherlands, Flanders, or Dutch-speaking colonies, you will run into handwritten records at every turn. The Dutch kept meticulous records – church authorities, notaries, and municipal clerks all wrote by hand, and much of it survives in provincial and municipal archives across the country.
DTB registers – the primary source for Dutch genealogy before 1811
Burgerlijke stand records from 1811 onward (birth, marriage, death certificates)
Notarial archives – property sales, marriage contracts, estate inventories
Bevolkingsregisters (population registers) from the mid-19th century
Weeskamer records (orphan chamber) – guardianship documents and estate settlements
Ship passenger lists and emigration records for Dutch settlers abroad

Which script is it?
Found a Dutch document you can't read? Here's what you're probably looking at.
The older the document, the stranger the letters. Most Dutch records fall into one of a few script types – and knowing which one you have helps you date the document too:
Gothic cursive (gotisch cursief) – angular, compressed letters with heavy abbreviation. Most Dutch documents from the 1300s to the 1600s are written in it.
Kanselarijschrift (chancery hand) – a formal, upright script in charters and legal documents
Running clerk's hands – the fast, joined-up writing of notaries and town clerks, common in records from the 1500s and 1600s
Rounder "Latin" writing – gradually replaced Gothic script from the 1600s onward; if you can make out most letters, your document is probably from this period or later
1800s copperplate – the neat, slanted school hand of later records. Readable, but full of old spelling and abbreviations.

The technology
How does Dutch handwriting recognition work?
Handwritten text recognition (HTR) uses neural networks to turn an image of a handwritten page into digital text, character by character. Unlike printed-text OCR, HTR has to deal with connected strokes, inconsistent letter sizes, and centuries of evolving script conventions. Transkribus trains its models on large collections of transcribed Dutch documents so the AI learns the specific patterns of each script type.
Deep learning models trained on transcribed Dutch manuscripts and documents
Automatic layout analysis separates text lines, margins, and annotations
Character-level recognition handles connected cursive strokes and ligatures
Built-in language modelling uses Dutch word patterns to improve accuracy

AI Models for Dutch Handwriting
Browse public models trained on historical Dutch documents – Gothic cursive, notarial records, church registers, and more.
Ready to read old Dutch handwriting?
Create a free account to process unlimited documents, train custom models, and unlock the full platform. Or try our free old handwriting reader to test it on a single page.
50 free credits every month – No credit card required
200M+Pages processed
500K+Users worldwide
300+Public AI models