Skip to content
  • Pricing
For Libraries

Make your library's hidden collections searchable.

Manuscripts, rare prints, historical newspapers, correspondence — millions of pages in your stacks that researchers can't find because they aren't searchable. Transkribus reads handwritten and printed text with AI and opens your collections to the world.

Used by the British Library, BnF, National Library of Norway, and libraries worldwide

Bibliotheca Hertziana
©Bibliotheca Hertziana — Max Planck Institute
200M+Pages processed to date
95%+Accuracy on trained models
100+Languages & scripts supported
EU-hostedGDPR-compliant, data stays in Europe

Co-owned by leading libraries worldwide

AI text recognition for every type of library collection

From medieval manuscripts to 20th-century newspapers — Transkribus handles the unique scripts, layouts, and challenges of library holdings.

Manuscripts & Rare Books

Medieval codices, early modern manuscripts, author estates — AI reads historical handwriting across centuries and makes entire collections searchable.

Historical Newspapers

Complex column layouts, Fraktur typefaces, mixed content — layout analysis and text recognition handle even the most challenging newspaper pages.

Correspondence Collections

Scholarly letters, literary estates, institutional correspondence — make thousands of handwritten letters findable by name, date, and content.

Early Printed Books

Incunabula, antiqua, Fraktur — AI models trained on historical typefaces outperform standard OCR on pre-modern printed text.

Address Books & Directories

Structured tabular data across thousands of pages — table recognition extracts names, addresses, and occupations at scale.

In-House Scanning with ScanTent

Equip reading rooms with portable ScanTent stations. Visitors scan documents with their phones — no expensive book scanners needed.

200,000 pages of Viennese history — searchable

The Vienna City Library used Transkribus to digitize the Lehmann Address Books — 200,000+ pages of Fraktur-printed addresses spanning 1859 to 1942. Standard OCR failed on the historical typefaces. Transkribus delivered searchable text, making decades of Viennese social history accessible to researchers for the first time.
200,000+ pages of Fraktur text recognised
Handwritten manuscripts from the Grillparzer estate transcribed
20,000 historical obituary notices made searchable
Lehmann Address Book — Vienna City Library
©Wienbibliothek im Rathaus

A public AI model for Norwegian handwriting

The National Library of Norway built NorHand — a public AI model trained on 400+ different handwriting samples from the 19th and 20th centuries, including works by famous Norwegian authors. With a character error rate of just 4%, the model is now freely available to any institution working with Norwegian manuscripts.
400+ different handwriting samples in the training data
4% character error rate — near-human accuracy
25% of the library's scanned holdings already processed
NorHand model — National Library of Norway
©National Library of Norway

Libraries using Transkribus

From national libraries to specialist research collections — see how libraries use AI to unlock their holdings.

Vienna City Library

200,000+ pages of address books, manuscripts, and obituary notices digitized with AI.

Case Study

NorHand — National Library of Norway

A public AI model trained on 400+ hands, now processing 25% of the library's scanned holdings.

Case Study

Bibliotheca Hertziana, Rome

Rare 15th–16th century art history books made searchable with custom HTR models.

Case Study

ScanTent at the BnF

40 portable scanning stations deployed across France's national library.

Case Study

Full-service

We handle the entire project for you

Many libraries prefer to outsource large-scale digitization. We manage the full process — from scanning consultation to searchable, published collections online. Your team defines what to digitise, we deliver the result.
Scanning consultation and preparation
Custom AI model training on your collection's scripts
Batch transcription and quality review
Published Transkribus Site with full-text search
Ongoing support and model improvements
Grillparzer manuscript — Vienna City Library
©Wienbibliothek im Rathaus
Self-service

Or try it yourself — right now.

Create a free account, upload a scan, and see the AI read your text in seconds. Choose from 300+ pre-trained models — including models for Fraktur, Kurrent, and dozens of historical scripts.

50 free credits every month · No credit card required

Drag an image here

Select a file...

PNG or JPG up to 10 Mb

Wolpi
AI Assistant

By uploading an image, you accept our terms and privacy policy.

200M+ pages transcribed99.9% uptimeCooperative ownedGDPR compliantEU sovereign infra

Built on trust, hosted in Europe.

Transkribus is built and hosted in Austria by a European cooperative. Your documents stay on our own servers — no US cloud, no Big Tech dependencies.

EU-hosted & GDPR-compliant

All data processed and stored in Austria. Full compliance with European data protection regulations. Your documents never leave Europe.

Cooperative, not corporate

READ-COOP is a European cooperative owned by 180+ member institutions — including national libraries, university libraries, and research institutions. You can join and shape the platform.

Proven at scale

200 million pages processed. Used by the British Library, BnF, National Library of Norway, and hundreds of institutions across Europe.

Let's make your collections searchable.

Managed project or self-service — we are happy to discuss both.

200M+pages processed
95%+accuracy on trained models
EU-hostedGDPR-compliant