5 questions for Dr. Arjan van Hessen

In our new ‘5 questions for’ column, we engage with interesting people from our network and experts to dive into the world of oral history and everything related to it. We kick off with Dr Arjan van Hessen who is closely involved in the technology used for oral history.

Dr. Arjan van Hessen / Informatics – University of Twente / Humanities – University of Utrecht / Telecom – Telecats

1/ Can you briefly introduce yourself?
‘I am actively involved in both the use and development of modern speech technology from the University of Twente (in collaboration with Radboud University) and the Enschede-based company Telecats. I am also a steering committee member of ‘Speaking History’. I am concerned with technology used for the transcription of spoken documents (as in oral history) and with the development of better dialogues in the telecom world. In addition, I have until recently served on the board of Dutch Infrastructure Projects for the Humanities: CLARIN and CLARIAH (on behalf of Utrecht University)’.

2/ What is your expertise?
‘My expertise is the combination of new and existing technology in improving and developing automatic speech recognition (converting spoken word into text). Although I am undoubtedly not “the expert” in every single field, I am well able to adequately speak along in all the different fields’.

3/ What are you currently working on?
‘At the moment I am implementing and using the latest speech recogniser Whisper. It performs much better than the current KALDI recogniser. If we manage to get Whisper running well and available to everyone, another huge step has already been taken’.

4/ For the Talking History node, you have developed a tool called Labelmaker. Can you tell a bit more about it?

‘LabelMaker is a simple tool that allows text fragments from subtitle files to be labelled. Those labels are gradually put together by oral history project teams. This makes it very easy to quickly label a large number of audiovisual files. Once the labels are added to the file, they can be read in and you get an overview of all text fragments and their labels.

At the moment, the software has to be installed on your computer, which is the only drawback. This is because not all computers are proprietary and you then have to ask (and get) permission to install it.’

5/ The ambition is to make it a web-based version that can be used by oral historians. How is this progressing and do you have any idea when this will be finished?
‘I am now working to see if a web version (you can use the programme directly through a web browser) of LabelMaker can also be made that does allow the same things to be done as with the stand-alone programmes. It looks like it will work and I hope it will be ready by the end of February’.

Follow Arjan via LinkIn