Introducing Transkribus Super Models – Get Access to ‘The Text Titan I’

With the Transkribus Next Gen well on its way, we are excited to announce that access to our new Transkribus super model ‘The Text Titan  I’ is now available for early access.

In this blog post we will talk about why we are introducing this new technology and about the benefits that come with it. We will also show you how to access ‘The Text Titan I’ and when you can expect it to go live.

Why the Change

One of our next steps towards Transkribus Next Gen includes adding new elements and new technologies for our users. And one of these technological elements are the new Transkribus Super Models for text recognition, which are the most advanced models we have to offer. 

They are great at dealing with natural language, which helps them decipher all sorts of written materials. They are able to outperform existing PyLaia models by a significant margin. A key advantage of these models is that they consist of both an optical part that processes the images and an extensive language model that tries to make sense of and improve the extracted text information. The language part is able to deal with more than a single language, and old and new forms of a language at the same time. This allows the models to output very high-quality text.

Thus, the first Super Model we have trained, The Text Titan I, is aimed at tackling the challenges of recognising both handwritten and printed text with remarkable accuracy and efficiency, in a variety of languages. Text Titan I is already outperforming existing large models such as The Dutchess I or The German Giants. 

On a set of 300 very heterogeneous pages, the transformer-based Text Titan model was able to yield better results in comparison to our strongest PyLaia model. Especially the number of pages with very low character error rates is considerably higher with Text Titan.

Benefits of the Text Titan

When it comes to using Super Models such as the Text Titan I, there are a number of benefits. As explained above, a Super Model is one big, very general model with the ability to recognise both handwritten and printed text simultaneously. This will be particularly useful when working with mixed materials. Some archival holdings or manuscript collections can have different types of writing, printed as well as handwritten documents, preprinted forms filled in by hand, index cards etc. With the Text Titan, you are able to use the model on both types of text, this means that you don’t need different models or have to constantly change settings if you are working with handwritten as well as printed documents.

The Text Titan is remarkably adept at processing a wide variety of materials and writings. Although our Super Models are not currently fine-tunable or trainable by users, they deliver outstanding out-of-the-box performance across numerous heterogeneous types of material which will help you to quickly produce Ground Truth for training your own customised PyLaia model. As we recognise the value of customisation and speed, and we are diligently planning for the future where Super Models like the Text Titan will be made adaptable to better meet your specific needs.

A specialised Pylaia model trained for well-defined material can, however, still yield better results, but the creation of the training data for such a specialised model can be speeded up considerably by first processing part of the material with the Text Titan and correcting it manually.

How can I access The Text Titan?

You can try out our new advanced text recognition model, The Text Titan I, by filling out this request form to join the early access list. Once you’ve received access and tried out the Text Titan I, please  please let us know what you think and give us feedback. You can do this by messaging us on our social media channels or, even better, by using the feedback button in the Transkribus web app.

What are the Next Steps? 

The next steps for Transkribus will be the addition of more languages to the Super Models, and efforts towards higher accuracy through improved layout recognition. The Text Titan will be made available to the general public during the summer months. 

The Super Models mean a significant step forward in text recognition technology and in our ongoing commitment to providing our users with the latest technology. With the ability to process both handwritten and printed text at the same time as well as to use them with multiple languages, these models offer unprecedented convenience and flexibility. 

Don’t miss your chance to explore the capabilities of this powerful model. Sign up for early access to be among the first to try The Text Titan I and help shape its development with your valuable feedback.