"As an administration open to new technologies, we are observing speech recognition software and the rapidly advancing development"
As part of the STADIEM* programme, the Austrian Parliamentary Administration and the Hamburg-based startup aiconix are cooperating to jointly develop a speech-to-text solution with automated recognition of Austrian dialects.
In a conversation with Lena Ninaus and Dario Summer, we would like to find out why this project is so interesting for the Austrian Parliament..
What forms of accessibility do you think are urgently needed in the media world?
Ninaus: Easy-to-understand language is one of the most important aspects of inclusive communication. That is why we offer the "Current Issues" format in simple language on our website. These are text contributions on the work of the Parliament at language level B1. In this way, we provide low-threshold access to information about parliamentary events.
In addition, audio and video are becoming increasingly important forms of communication. Therefore, providing text alternatives is of great concern to us. Through the cooperation with aiconix, we hope to be able to provide text alternatives even more quickly in the future.
What is the Austrian Parliament already doing today for the digital inclusion of people?
Ninaus: The reduction of barriers is clearly the focus of our websites. For this reason, they are regularly optimised. In particular, the current web relaunch focuses on digital accessibility.
However, it is also important to us to go beyond the fulfilment of legal requirements for digital accessibility: For deaf people, we offer interpretation into Austrian Sign Language (ÖGS) during National Council sessions and for central content on legislation. Since last year, it has also been possible to read the latest news on the work of Parliament in simple language (B1 language level). Core content on political education is also available on our website in plain language (A2 language level). On our social media channels, we communicate in a barrier-free way, for example by providing subtitles for video and audio as well as alternative texts for images.
How did the cooperation with aiconix come about?
Summer: As an administration open to new technologies, we have been observing speech recognition software and the rapidly advancing developments in this field for quite some time. Unfortunately, we have not yet found a software solution that reliably recognises the Austrian German spoken in our parliament. However, this is precisely the prerequisite for the efficient use of automated speech recognition in the Parliamentary Administration.
The request from aiconix to participate in the STADIEM project as a partner was very convenient for us in this respect, as a cooperation would give us the opportunity to become part of the problem solution on the one hand, and on the other hand, we could test the use of the existing services from aiconix in our real working environment.
What exactly does this involve and what do you expect from the output?
Summer: As a project partner, we provide transcripts of plenary sessions of the Federal Council as well as of public events and the corresponding audio recordings, which are accessible via the videos in the parliament's media library. With this data, aiconix can train the language model for Austrian German. The Bundesrat in particular is interesting as a data pool, as speakers in the Länder chamber traditionally speak more dialect.
Our expectation is that this will enable the model to reliably recognise the peculiarities of the language variants spoken in Austria and that we can thus take a further step towards accessibility by quickly providing text alternatives for the audiovisual content on the website.
Do you already use solutions from aiconix and if so, which ones?
Summer: No aiconix solutions have been implemented in the workflow yet, but we are testing the speech recognition models to see if, where and how we can integrate them into our processes.
What solution enhancements / features would you like to see from aiconix in the future?
Summer: As always, the answer to questions like these is: the egg-laying willy-nilly. The optimal solution for us would be a learning solution that, on the one hand, enables the subsequent subtitling and live subtitling of audiovisual content via an interface and, on the other hand, is integrated into our closed editorial work process for the creation of transcripts.
* STADIEM (Startup Driven Innovation in European Media) - This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 951981”.