iiinova — where your digital self keeps evolving

Reposted by Transkribus

Christian Lendl @dchr.is · 23/09/2026

Really enjoyed the @transkribus.bsky.social conference in Passau. Lots of inspiring researchers and interesting projects. Really happy with the great feedback I received after my presentation. #phdlife

Group photo of all conference participants

053

Reposted by Transkribus

Anna Michalcová @burnedlegate.bsky.social · 23/09/2026

Nice to see the Winter School of HTR mentioned as one of the highlights of TUC26 and the Transkribus community in general! Super proud to be part of it since the very beginning! @digilab-imafo.bsky.social @transkribus.bsky.social (also if someone has a better photo, do let me know 💙)

023

Reposted by Transkribus

Melissa Terras @melissaterras.bsky.social · 23/09/2026

Finishing up the @transkribus.bsky.social user conference. I’ve been working on this for over 13 years now and I am so, so proud of the infrastructure, community, and team delivering our ATR and information extraction tools, cooperatively. A strong sense of community here! 🦾

We Have Built Unique Al in a Sea of AI
Hundreds of Features
Thousands
of Projects
Tens of
thousands of Models
Hundreds of thousands
of Users

Slide presentation

3172

Transkribus @transkribus.bsky.social · 23/09/2026

Last day of #TUC26 is here! Starting with scholarship recipients presenting their research, then more lightning talks, roundtables, and our final "Road Ahead" session.

010

Reposted by Transkribus

Ellen Forget @ellenforget.bsky.social · 22/09/2026

A big thanks to everyone who came out to my talk about my braille character recognition model at the @transkribus.bsky.social conference in Passau today! (And thanks to @melissaterras.bsky.social for the photo!)

Ellen presenting in a lecture hall. The slide above them reads Training Braille Character Recognition Using Transkribus.

0213

Transkribus @transkribus.bsky.social · 22/09/2026

The #TUC26 is already in session! There are lots of talks and discussions, and of course there's plenty of opportunity to network with the community! Here are some highlights from the first day 📸

042

Reposted by Transkribus

Anna Michalcová @burnedlegate.bsky.social · 18/09/2026

Next week we'll speak about our Winter School of HTR in Passau 💫

021

Reposted by Transkribus

DigiLab, Institute for Medieval Research, ÖAW @digilab-imafo.bsky.social · 19/09/2026

Join us next week at #TUC26! @catominor.bsky.social and @burnedlegate.bsky.social will share their experiences from 4 years of training researchers at the Vienna Winter School for HTR. Catch their lightning talk in the Teaching Machines, Teaching Humans panel!

042

Reposted by Transkribus

Melissa Terras @melissaterras.bsky.social · 21/09/2026

And so we kick off the @transkribus.bsky.social user conference in Passau, with @caromein.bsky.social, chair of READ-COOP board, introducing and welcoming us to discuss all things advanced text recognition, and cooperative, ethical, AI approaches.

Annemieke kicks off the event with not all AI is created equal behind her on the screen

042

Transkribus @transkribus.bsky.social · 18/09/2026

It’s almost time! At #TUC26, we will explore how AI can support historical research with three days of keynote speeches, workshops and round-table discussions. Join us in person or online - see you in Passau on Monday! 👋 tuc.transkribus.org/tickets

010

Transkribus @transkribus.bsky.social · 17/09/2026

HTR is usually framed as a research tool. But once it works, people start teaching it, adapting it, and finding new uses for it. 💡 Next week, six talks explore what it actually takes to teach HTR literacy, and what happens once people start using it creatively. Catch the full session at #TUC26!

tuc.transkribus.org

Panel 7: Teaching Machines, Teaching Humans

HTR as a pedagogical tool — in university classrooms, secondary schools, and professional training settings.

031

Transkribus @transkribus.bsky.social · 16/09/2026

How do you train a model to read a script that almost no one is building tools for? Six speakers share how they trained models for Braille, early modern Spanish calligraphy, Classical Nahuatl, Cyrillic-script Romanian registers, and endangered languages more broadly. Join the session at #TUC26 🌏

tuc.transkribus.org

Panel 1: Scripts of the World

All six talks show what happens when language or script sits outside the European mainstream — methodologically, linguistically, technically.

041

Transkribus @transkribus.bsky.social · 15/09/2026

"Human-in-the-Loop" has become a popular buzzword. But who's actually in the loop? For us, experts are the engine, not the backup - they bring context AI can't. At #TUC26, this panel explores collaborative systems in HTR and research 👇 tuc.transkribus.org/programme/pa...

tuc.transkribus.org

Panel 6: The Human Element

HTR as a communal practice — institutional, citizen-driven, and international.

000

Transkribus @transkribus.bsky.social · 11/09/2026

Is the biggest model always the best? Well, that depends. Text Titan II, one of the largest models, remains a strong starting point for Latin-script documents. For smaller scripts, less widely represented languages or hands, a specialised model can often do better. Full benchmarks and details 👇

transkribus.org

Introducing Text Titan II | Transkribus Blog

Transkribus introduces Text Titan II: a powerful, multilingual transcription model for Latin-script documents. Boost accuracy for print and handwriting today.

020

Transkribus @transkribus.bsky.social · 10/09/2026

What do an archivist, a genealogist, and a data scientist want from a historical document? Different things, but the same problem: turning unstructured documents into structured data at scale. Our round table explores exactly that at #TUC26 👇 tuc.transkribus.org/programme/ro...

tuc.transkribus.org

From Documents to Data

Structuring the Historical Record at Scale

040

Transkribus @transkribus.bsky.social · 09/09/2026

Can’t make it to Passau? You can still be part of #TUC26. Join keynotes & selected sessions live online, take part in Q&As, and catch up with recordings afterwards. 📅 21–23 Sept 2026 🎟️ In-Person and Virtual tickets: tuc.transkribus.org/tickets

000

Transkribus @transkribus.bsky.social · 08/09/2026

Bring collections, datasets, models, Sites and collaborators together with Projects in Transkribus — and give everyone involved a shared space to work from. See how Projects can support your workflow: help.transkribus.org/managing-pro...

help.transkribus.org

Managing Projects

Projects in Transkribus are a collaborative workspace that group your collections, AI models, datasets, and published sites together. They provide structure, collaboration, and visibility to your tran...

100

Transkribus @transkribus.bsky.social · 07/09/2026

We are just two weeks away from #TUC26! As document analysis technologies evolve, the question isn't just whether a machine can read a historical source — but what researchers can do with what's inside it. Join us in Passau or online on 21–23 September. 👇 tuc.transkribus.org/programme

020

Transkribus @transkribus.bsky.social · 07/09/2026

Challenging layouts. A complex schema you need as structured output. Until now, getting there meant building and maintaining a long, multi-step model pipeline. We've found a way to do it in one step! Watch the video to see how Smart Extract Models work: youtu.be/zeqxhD8ARe8?...

youtu.be

Smart Extract explained: from page image to structured data in one step (Atlas demo)

YouTube video by Transkribus

120

Transkribus @transkribus.bsky.social · 03/09/2026

History goes far beyond the Latin alphabet. 📜 Explore five Transkribus models for Ethiopic, Hebrew, Devanagari, Balinese and Ottoman Turkish, and the historical collections behind them: blog.transkribus.org/en/ethiopic-...

blog.transkribus.org

Ethiopic, Hebrew, Devanagari, Balinese and Ottoman Turkish: 5 public AI models for transcribing non-Latin scripts - READ-COOP

Read the Post Ethiopic, Hebrew, Devanagari, Balinese and Ottoman Turkish: 5 public AI models for transcribing non-Latin scripts in our Blog.

172

Transkribus @transkribus.bsky.social · 02/09/2026

A messy, multistep pipeline or an unpredictable LLM. Is there really no third option? 📢 The new Smart Extract Models go from image to structured output in one step! Watch the video to see how Smart Extract Models work: youtu.be/zeqxhD8ARe8?...

youtu.be

Smart Extract explained: from page image to structured data in one step (Atlas demo)

YouTube video by Transkribus

120

Transkribus @transkribus.bsky.social · 28/08/2026

“Which data did we train this model on?” Six months into a project, that simple question isn't always easy to answer. Transkribus Datasets help you keep Ground Truth traceable as it evolves. Read our guide to Datasets: help.transkribus.org/managing-dat...

help.transkribus.org

Managing Datasets

Datasets in Transkribus are curated sets of pages used to train, validate, and test custom AI models.

062

Transkribus @transkribus.bsky.social · 27/08/2026

What can 2,600 love letters tell us about life during wartime? Roland and Hilde Nordhoff wrote to each other almost daily during WWII, offering a rare glimpse into everyday life. Read how volunteers &Transkribus turned the collection of letters into a searchable digital archive 👇 buff.ly/Rp3An2a

transkribus.org

Trug und Schein: unlocking the letters of wartime lovers | Transkribus Blog

Unlock the letters of wartime lovers Roland and Hilde Nordhoff, and explore how their story bridges generations through a unique citizen science project.

071

Reposted by Transkribus

Martin Roček @martin.rocek.dev · 21/08/2026

If you want to learn more about HTR and @transkribus.bsky.social, definitely check out our Winter School! 😊

073

Reposted by Transkribus

C. Annemieke Romein @caromein.bsky.social · 20/08/2026

If you haven’t registered yet… do check out the wonderful program of the @transkribus.bsky.social User Conference in Passau 21-23rd of September. Be there in person or join us online! There is much to be learned and… some new things coming up! It’s going to be cool!!! #TUC2026

054

Transkribus @transkribus.bsky.social · 21/08/2026

Historical manuscripts rarely follow neat rules. Join us at #TUC26 to hear how researchers are making sense of some of the more challenging documents in their collections, and what we can learn from their approaches. 👉 Panel 4: Mastering Complexity tuc.transkribus.org/programme/pa...

tuc.transkribus.org

Panel 4: Mastering Complexity

All six talks wrestle with material that actively resists standard HTR — physically, structurally, or linguistically.

010

Transkribus @transkribus.bsky.social · 19/08/2026

How do you transcribe thousands of handwritten historical documents with a small team? The Hanse.Quelle.Lesen! project used Transkribus to build an efficient transcription workflow for 14th–17th c. Hanseatic records, bringing the collection online. Learn how👇 www.transkribus.org/blog/hanse.q...

transkribus.org

How the Hanse.Quelle.Lesen! project made Hanseatic records accessible through Citizen Science | Transkribus Blog

Discover how the Hanse.Quelle.Lesen! project utilizes Citizen Science and AI to digitise Hanseatic records, making historical documents accessible to all.

041

Transkribus @transkribus.bsky.social · 13/08/2026

Working with Garshuni-Arabic manuscripts? A public model for Garshuni-Arabic (Syriac Serto Script) is now available in Transkribus, offering a starting point for researchers. Thanks to Ephrem Aboud Ishac and all collaborators who contributed! www.transkribus.org/models/garsh...

transkribus.org

Garshuni-Arabic (Syriac Serto Script) | Arabic, Classical Syriac Text Recognition Model | Transkribus

Model trained on several Garshuni-Arabic manuscripts, especially on colophons and the Syro-Coptic Synaxarion manuscript of Aleppo. This first public model of...

030

Transkribus @transkribus.bsky.social · 06/08/2026

Great ideas in digital humanities grow through collaboration. Join researchers, archivists, developers, and heritage professionals at the Transkribus User Conference 2026 in Passau to explore real-world projects, exchange ideas Secure your ticket now:

tuc.transkribus.org

Transkribus User Conference 2026

Join us at the Transkribus User Conference 2026 - Connect, Learn, and Explore the Future of Unlocking Written Heritage with AI

031

Transkribus @transkribus.bsky.social · 04/08/2026

How do you get 250,000 insect specimen labels into a searchable database? The Museum für Naturkunde Berlin worked with Transkribus to transcribe and structure handwritten and printed labels into accessible research resources. Learn how:

transkribus.org

Creating a model to transcribe 250,000 specimen labels at the Museum für Naturkunde Berlin | Transkribus Blog

Transkribus collaborates with the Museum für Naturkunde Berlin to digitise and transcribe 250,000 insect specimen labels, enhancing digital access for researchers and educators.

052

Transkribus @transkribus.bsky.social · 30/07/2026

For historical document research, transparency and reproducibility matters. With Datasets in Transkribus, researchers can freeze dataset versions during model training, creating a traceable link between training data and custom models. Create your own Dataset:

help.transkribus.org

Managing Datasets

Datasets in Transkribus are curated sets of pages used to train, validate, and test custom AI models.

030

Reposted by Transkribus

Melissa Terras @melissaterras.bsky.social · 29/07/2026

Well, it doesn’t have to be all or nothing. You can choose to make a lot open while ringfencing the assets your business model depends on. As our #DH2026 poster for @transkribus.bsky.social shows!

As open as possible, as closed as necessary - Transkribus poster for DH2026

"As Open as Possible, as Closed as Necessary":
Balancing Community Needs, Sustainability, and Access in
Transkribus
M. Terras12, A. Stauder?, C.A. Romein245, F. Stauder?, M. Prien?
READ
*University of Edinburgh, +READ-COOP SCE, University of Twense, "University of Bern
The Crisis of Openness in the Age of Predatory Al Mass-scraping by generative Al providers has created an unprecedented dilemma: open licenses now inadvertently surrender "our collections wholesale to Al as part of mass training" (Cohen 2025), and cultural heritage platforms face crippling exploitation. Digital Humanities scholars must recalibrate expectations regarding openness.
The DH-Dillema
How can we protect digital infrastructure and training data from predatory appropriation by large technology providers-ensuring long-term sustainability-without abandoning
our
commitment to accessibility and
community service?
Case Study: READ-COOP and Transkribus
Transkribus (an Al platform for Automated Text Recognition from historical documents) navigates these opposing forces. Maintained by READ-COOP, a European Cooperative Society established in 2019, the platform embodies a critical tension: balancing openness with the protection of proprietary algorithms and 400TB of training datasets developed through significant public investment. Without such protection, larger technology providers could re-appropriate these resources, thereby threatening the cooperative's viability and its capacity to serve its community and under-resourced languages.
Mechanisms of Accessibility
READ-COOP demonstrates that strategic closure need not preclude meaningful access:
• Users retain full data ownership and export rights (open formats).
. 400 public models: 367 ATR models, 21 layout models, 17 field models, 1 table model published for community use.
. Free monthly credits for basic features & models
. Scholarship Programme:
507
students
from
91 coun…

1308

Transkribus @transkribus.bsky.social · 28/07/2026

Starting with Transkribus? Our Beginners Webinar is now available on YouTube. Learn the essentials, from uploading documents and adjusting layouts to running text recognition models and building an efficient workflow. 👉Watch the recording: youtu.be/bLXMXvhuz6k?...

youtu.be

Transkribus Webinar for Beginners (English)

YouTube video by Transkribus

043

Transkribus @transkribus.bsky.social · 27/07/2026

How do you turn thousands of volunteer transcriptions into meaningful research data? Join our #TUC2026 roundtable “Crowdsourcing the Past” with experts from leading cultural heritage institutions discussing citizen science, transcription workflows, and lessons learned. 👉https://buff.ly/KiZ3BPq

tuc.transkribus.org

Crowdsourcing the Past

Citizen Science Strategies for Transcription at any Scale

031

Transkribus @transkribus.bsky.social · 22/07/2026

Millions of church record pages are available on Archion, but reading Kurrent script can be a challenge. With the integration of Transkribus, users can now generate transcriptions directly within the Archion viewer. Learn more about the integration: www.transkribus.org/blog/strateg...

transkribus.org

Strategic AI integration: How Archion implemented the Transkribus API | Transkribus Blog

Discover how Archion streamlined family history research by integrating AI text recognition through the Transkribus API, making historical documents more accessible.

020

Transkribus @transkribus.bsky.social · 21/07/2026

Too many collections, models, and pages of Ground Truth to keep track of? Our new Projects and Datasets help bring structure to your Transkribus workflows, making training data traceable and projects easier to manage. 🎥 Watch the webinar recording: youtu.be/-WRf80CDRPg?...

youtu.be

Datasets & Projects Webinar (English) | Organising Transkribus Workflows & AI Training Data

YouTube video by Transkribus

031

Reposted by Transkribus

RCI - Reichsstadtregesten @rci-projekt.bsky.social · 10/07/2026

Archivreise heißt oft auch: Warten auf verspätete Züge...dafür aber immer in bester Begleitung vom unserem Scantent von @transkribus.bsky.social 😉🤭

072

Transkribus @transkribus.bsky.social · 15/07/2026

Historical documents often mix languages on a single page or collection. Read our latest blog post to see how researchers are training Transkribus models to transcribe multilingual collections in one go! 👉 www.transkribus.org/blog/multili...

transkribus.org

Breaking the language barrier: 3 multilingual Transkribus research projects | Transkribus Blog

Discover how Transkribus is breaking language barriers in historical research with innovative multilingual text recognition projects. Unlock the potential of your archives.

020

Transkribus @transkribus.bsky.social · 09/07/2026

Were society magazines the social media of the late 19th century? At #TUC2026, Christian Lendl will show how Transkribus helped transform the Wiener Salonblatt—with 300k aristocratic notices and 20k portraits—into structured data for exploring historical social networks. buff.ly/b61Qt5J

tuc.transkribus.org

030

Transkribus @transkribus.bsky.social · 06/07/2026

Historical records aren't always neat blocks of text. Many are handwritten tables, ledgers, or registers. 📜 In this blog article, we show how to turn those tables into digital spreadsheets with Transkribus:

transkribus.org

From paper to pixels: How to convert handwritten tables into digital spreadsheets | Transkribus Blog

Transkribus makes it possible to convert handwritten tables into digital spreadsheets. Find out how in this post.

050

Transkribus @transkribus.bsky.social · 02/07/2026

📢 We've just released Text Titan II, our most accurate general-purpose model yet. It reduces transcription errors by ~47% compared to Text Titan I on our internal benchmark, with the biggest gains on handwritten documents! Find out more and try Text Titan II: www.transkribus.org/blog/introdu...

transkribus.org

Introducing Text Titan II | Transkribus Blog

Transkribus introduces Text Titan II: a powerful, multilingual transcription model for Latin-script documents. Boost accuracy for print and handwriting today.

082

Transkribus @transkribus.bsky.social · 01/07/2026

Reminder: Join our free webinar for beginners at 4 pm today!

000

Transkribus @transkribus.bsky.social · 24/06/2026

Wondering what the new Projects and Datasets workspaces in Transkribus are all about? Before everyone heads off for the summer break, we are hosting two webinars to get you up to speed 📆 For beginners: www.transkribus.org/event/for-be... Projects & Datasets: www.transkribus.org/event/projec...

transkribus.org

Transkribus for beginners | Transkribus Events

Get started with Transkribus (4 PM CET/9 AM CST)

000

Transkribus @transkribus.bsky.social · 23/06/2026

Notarial archives are notoriously difficult to process, especially when filled with crossed-out text, marginalia, and additions. At #TUC26, Gena Lemmens (Regionaal Archief Tilburg) shares a multi-year journey of teaching Transkribus to read "chaotic" layouts. Full programme: buff.ly/1DuGJDF

tuc.transkribus.org

Panel 4: Mastering Complexity

All six talks wrestle with material that actively resists standard HTR — physically, structurally, or linguistically.

030

Transkribus @transkribus.bsky.social · 12/06/2026

Can a handwriting recognition platform read text that was never written in ink? At the TUC 2026, Ellen Forget from the @ualberta.bsky.social will share how she trained a Braille character recognition model in Transkribus. 👉 tuc.transkribus.org/programme/ta...

tuc.transkribus.org

Training Braille Character Recognition Using Transkribus

021

Transkribus @transkribus.bsky.social · 11/06/2026

Deciphering old Cyrillic shouldn't be the most time-consuming part of your research. We’ve highlighted five models for historical Russian texts, so you can spend less time deciphering and more time actually working with the sources. 👉

blog.transkribus.org

5 AI Models For Transcribing Old Russian Handwriting And Printed Russian Texts - READ-COOP

Read the Post 5 AI Models For Transcribing Old Russian Handwriting And Printed Russian Texts in our Blog.

120

Transkribus @transkribus.bsky.social · 05/06/2026

It was wonderful connecting with the digital humanities community at #DHBenelux! Our Chair of the Board, @caromein.bsky.social, presented a poster exploring how the Transkribus cooperative balances community needs, long-term sustainability, and open access.

071