60 Minutes of DH: eScriptorium for Handwritten Text Recognition in Humanities Research

16 March 2021

In January 2021, the DH Lab of the Leibniz Institute of European History in Mainz launched its series of (online) events “60 Minutes of DH” with a webinar on the automatic transcription tool eScriptorium.

60 Minutes of DH

The monthly events, planned as one-hour long afternoon sessions, are mainly intended for academic staff at the IEG and focus on joint discussions of tools, methods and literature from the field of Digital Humanities, as well as insights into the international project landscape. The goal is to encourage and support researchers when it comes to digital solutions supporting their history- and religion-related research. For this kick-off, however, the invitation was extended to a wider audience and was met with overwhelming interest by researchers from all over Europe.

eScriptorium Demonstrated

For the webinar we invited RESILIENCE partner École Pratique des Hautes Études (EPHE) in Paris, where a Digital Humanities team has been developing eScriptorium, a cutting-edge software for hand-written text recognition (HTR). Peter Stokes, research professor at EPHE, gave a demonstration of the tool, which was truly impressive.

While eScriptorium can be used for all kinds of scripts, it has been developed for hand-written texts of different writing systems and can handle complex page-layouts, which makes it especially interesting for religion-related research. Peter showed examples of Arabic, Hebrew and Syriac texts and demonstrated how the deep-learning software allows users to train the system for the material they are working with. The interface seems very user-friendly so you don’t need to be a programmer to work with it. What is particularly interesting is that it’s completely open source and training models can be shared for the benefit of the research community.

How to Use eScriptorium?

For researchers, the most important question is if and how they can use the software for their own projects. Peter pointed out that eScriptorium is still work in progress but it is possible to download and use it and researchers are more than welcome to do so. However, most of them probably won’t have access to a computer which is powerful enough to actually work with it. As a large number of high-resolution images are being processed, high-performing computers with GPUs are necessary to run it efficiently, which are very expensive and therefore hard to find in the humanities.

This is where RESILIENCE comes in: besides co-financing the development of eScriptorium as part of a Horizon 2020 project, the planned Research Snfrastructure will provide access to the HTR software on suitably powerful computers, along with a wide range of other tools and services for religion-related research. However, it will take a few more years until RESILIENCE is running (we wrote about the current state of development of the Research Snfrastructure here).

Peter’s demonstration of eScriptorium – the recording of which you can see below – was followed by an enriching discussion. Peter and his colleagues from EPHE, Daniel Stoekl Ben Ezra and lead developer Robin Tissot, answered to the partcipants’ questions, the transcript of which can be found here.

How Humanities Research Can Benefit from Digital Innovations

All in all, the kick-off of “60 Minutes of DH” was a success, the evaluation we conducted afterwards confirmed our impression. The webinar impressively showed how humanities research can benefit greatly from digital innovations such as HTR software and that there is great interest from the international research community. We were thrilled about the engaging participants and the positive feedback and thankful to have had the DH team of EPHE on board. We were reassured that the format of one-hour long sessions with a concise presentation – “short and sweet” as one participant has put it – and a great part of the time for discussion was well received.

Reason enough to look forward to the next editions of “60 Minutes of DH”!

– Sofie Sonnenstatter, Leibniz Institute of European History


This article was first published on the DH Blog of the IEG Mainz.