The problem
Why old wills and probate records are so hard to read
Wills and probate documents are among the richest genealogical sources — naming heirs, listing property down to individual spoons and livestock, and mapping family relationships that predate civil registration. But they are also among the most impenetrable. Before 1733, English wills were often written in Latin with secretary hand. Scottish testaments used court hand well into the 18th century. German Testamente mix Kurrent with legal Latin formulae. And every era brought its own abbreviations: 'viz.' for videlicet, 'ye' for 'the', 'sd' for 'said', and dozens more.
Secretary hand and court hand — letterforms that look nothing like modern cursive
Latin legal formulae: 'In Dei Nomine Amen', 'probatum fuit', 'bona notabilia'
Archaic abbreviations throughout: 'ye', 'sd', 'viz.', 'do.', superscript contractions
Multi-page documents with estate inventories listing hundreds of individual items
Faded ink, damaged parchment, and tightly packed text on 300–500-year-old documents
The solution
AI that reads old will handwriting across centuries
Transkribus uses handwritten text recognition (HTR) trained on millions of historical document samples — including legal documents, wills, and court records. Unlike standard OCR, which fails on connected cursive and archaic letterforms, Transkribus learns how secretary hand, court hand, and legal scripts work: the looped ascenders, the abbreviated Latin, the idiosyncratic contractions. Upload a scan or photo of your will or probate record, select a model, and get searchable text back in minutes.
300+ public models — including models trained on legal documents and testamentary records
Reads secretary hand, court hand, Kurrent, and 100+ historical scripts found in wills
Handles Latin legal formulae and abbreviations that generic OCR tools miss entirely
Confidence scores on every line — so you know exactly which words to double-check
Export as plain text, searchable PDF, or structured XML for your genealogy database
How to transcribe a will or probate record in 4 steps
Upload your document
Take a photo with your phone or upload a scan from an archive. Transkribus accepts JPG, PNG, PDF, and TIFF files — including multi-page probate packets.
Select an AI model
Choose from 300+ public models. Filter by language, century, and script type to find the best match for your will's handwriting style.
Run text recognition
Click 'Recognise' and let the AI work. A single page takes about 30 seconds. Multi-page probate files can be processed in batch.
Review and export
Check the transcription against the original. Fix any errors in the built-in editor, then export as text, PDF, or XML for your genealogy records.
Models for legal documents
AI models for probate record transcription and estate inventories
Transkribus's public model catalog includes models built for the scripts and document types found in wills and probate records. Estate inventories — detailed lists of a deceased person's possessions — are particularly rich genealogical sources, listing everything from land and livestock to household goods and debts. These documents often use tabular layouts and mix narrative text with itemized lists, which Transkribus's layout detection handles automatically.
English wills: Models for secretary hand (1500s–1700s) and later English cursive
German Testamente: Kurrent and Sütterlin models covering legal documents from the 1600s onward
Latin probate records: Models trained on ecclesiastical and civil court Latin
French testaments: Models for notarial scripts and Ancien Régime legal hands
Estate inventories: Layout detection handles tabular item lists, valuations, and marginal notes
See field extraction in action
Field Models detect and extract specific structural elements from documents — names, dates, references, and more — precisely and at scale.
Extracted Fields
Structured extraction
Extract names, relationships, and bequests from wills with field models
Beyond plain transcription, Transkribus field models can extract structured data directly from wills — pulling out testator names, heir names, relationships, bequests, property descriptions, and dates into organized fields. Instead of reading through pages of archaic legal prose, you get a structured summary: who left what to whom. This is especially powerful when processing multiple wills from the same parish or court, turning unstructured handwriting into a searchable database of family connections.
Extract testator name, executor, witnesses, and beneficiaries automatically
Identify bequests and property descriptions within the document
Map family relationships: spouse, children, siblings, grandchildren named in wills
Process multiple wills in batch to build a database of family connections across generations
Frequently Asked Questions
Accuracy depends on the script, condition, and model used. On well-preserved 18th- and 19th-century wills in English cursive, our best models achieve 95%+ character accuracy. Earlier documents in secretary hand or court hand may require a custom-trained model to reach the same level. Every line comes with a confidence score so you can see which words need checking.
Yes. Many wills — especially pre-1733 English wills and continental European testaments — contain Latin legal formulae such as 'In Dei Nomine Amen', 'probatum fuit', and 'bona notabilia'. Transkribus has public models trained on Latin handwriting that handle these standard legal phrases and abbreviations.
Yes. Probate packets often include the will itself, an estate inventory, accounts, and court filings — sometimes dozens of pages per ancestor. You can upload the entire packet as a multi-page PDF or as individual images, then batch-process all pages with a single click.
Yes. Beyond plain transcription, Transkribus field models can extract structured data — testator names, heir names, relationships, bequests, and property descriptions — directly from wills. This turns unstructured legal prose into organized, searchable fields.
Yes. Secretary hand — the dominant legal script in England from roughly 1500 to 1700 — uses letterforms very different from modern cursive. Transkribus has public models trained specifically on secretary hand documents. For particularly difficult hands, you can train a custom model on your own data starting with as few as 50 transcribed pages.
Transkribus accepts JPG, PNG, PDF, and TIFF files. You can upload photos taken with a smartphone or high-resolution scans from archives. For best results with phone photos, use good lighting and keep the document flat.
Transkribus gives you 50 free credits every month — enough to transcribe about 50 pages. No credit card required. If you need more, paid plans start at affordable rates. See our plans and pricing page for details.
Related guides
Explore more genealogy record types
Wills and probate records are just one piece of the puzzle. Explore our other record-type guides: Church records · Census records · Vital records · Military records · Immigration records · Land deeds
Ready to unlock your family's wills and probate records?
Create a free account and start transcribing wills, estate inventories, and probate documents today. 50 free credits every month — no credit card required.
50 free credits every month — No credit card required
200M+Pages processed
500K+Users worldwide
300+Public AI models