Handwritten Text Recognition (HTR)
Handwritten text recognition turns images of manuscript pages into searchable, editable text. This hub covers how HTR works, where it succeeds and struggles, and how you can make best use of it in your workflows.
11 articles
Leo Team · October 4, 2026
Fraktur OCR: Reading Blackletter Print by Eye and by Machine
A practical guide to Fraktur and other blackletter type. It covers the letters that trip readers and OCR, where family historians meet Fraktur, how OCR has handled it, which tools to use now, and how to read it yourself with an alphabet table.
Read the article
Leo Team · October 4, 2026
OCR for Old Books: Why Historical Print Breaks Standard OCR, and What to Use Instead
Why standard OCR misreads old books even when a person reads them easily, what newer whole-line models change, and how to choose a model by the kind of book. It also covers scanning and photographing old books, and getting searchable text out.
Leo Team · August 18, 2026
Can AI Read Cursive? Why Modern Handwriting and Historical Hands Are Two Different Problems
Can AI read cursive? Modern joined-up writing, yes; nineteenth-century copperplate, usually; secretary hand, Kurrent and other early-modern hands only with a model built for historical documents, such as Leo's ATR-1.
Leo Team · August 6, 2026
Can AI Read Faded Documents? Faded Ink, Foxing, and Show-Through, Triaged
Can AI read faded documents? Often, if the ink survives in the photograph. A triage guide to faded iron-gall ink, foxing and show-through: what capture, recognition and conservation can each fix
Leo Team · August 4, 2026
How to Test AI Transcription on Your Own Documents: An Afternoon Protocol
How researchers can test AI transcription tools on their own documents in an afternoon via random sampling, reference keying, CER scoring, and manual error reading for a directional local benchmark.
Leo Team · July 29, 2026
Can AI Read Old Handwriting? What Current Models Manage, and Where They Fail
Answers whether AI can read old handwriting: which model suits which job, why a model specialised for historical documents keeps the page's own spelling, six failure modes, and how to test a model on your own document
Leo Team · July 22, 2026
Handwritten Text Recognition: A Practical Guide to Reading Historical Documents with AI
Explains what handwritten text recognition (HTR) is, how it differs from OCR, how to interpret accuracy/CER figures, trained vs. zero-shot models, LLM transcription risks, and how to fold machine transcription into a verified historical-research workflow.
Leo Team · July 20, 2026
HTR for Historians: A Practical Guide to Handwritten Text Recognition
A practical guide to HTR for historians: what handwritten text recognition is, which kind of tool suits which project, how to read accuracy figures, and a step-by-step workflow that keeps your sources intact.
Leo Team · July 14, 2026
Which Languages and Scripts Can HTR Read? A Guide to HTR Supported Languages
Which languages and scripts HTR can read, and why real support depends on script, language, and hand rather than language-count marketing.
Leo Team · July 14, 2026
HTR vs OCR: What the Difference Is, and When It Matters for Historical Documents
The difference between HTR and OCR for historical documents: how each reads a page, why conventional OCR struggles with old print as well as handwriting, and where ATR models such as Leo's ATR-1 fit.
Leo Team · July 14, 2026
How Does HTR Work? A Look Inside the HTR Pipeline
How handwritten text recognition (HTR) works, stage by stage: capture, preprocessing, layout analysis, the HTR model that reads each line, decoding and export, and why accuracy depends on how closely your material matches the model's training.