Free secure AI transcription tool now available for everyone at the UG

Do you want to transcribe research interviews or lectures quickly and securely? The University of Groningen now offers a free AI-powered transcription tool for all researchers, staff, and students. Hosted locally on the UG’s High Performance Computing (HPC) cluster Hábrók, this tool makes it easy to convert audio recordings into text, while being GDPR compliant and without compromising data security.

For many researchers, transcribing interviews can be one of the most time-consuming parts of their work. While the rise of AI has brought numerous transcription tools to the market, most either come with steep license costs or pose serious concerns about data security and GDPR compliance. Free versions often mean losing control over what happens to sensitive data.

Local and secure solution

Since mid-2023, the UG Digital Competence Centre (DCC) has been working on a solution: a locally hosted transcription service that is free to use for all UG researchers, staff, and students. The service runs on the High Performance Computing (HPC) cluster Hábrók, managed by the UG’s Center for Information Technology (CIT), and uses Whisper, an AI model developed by OpenAI. Because the tool is installed and run locally by each user, your data never leaves UG premises unless you choose to move it.

Transcribe while you work

Thanks to Hábrók’s computing power, users can submit their audio files, queue the transcription, and return later to collect the results. This means you can focus on other work while the system handles the heavy lifting, retrieving your transcript after a few hours or the next day.

Easy access for all UG users

Getting started is simple: request an HPC account from their portal website. Once approved, follow the step-by-step Whisper installation guide created by the DCC, and run Whisper on Hábrók using a script. The process runs entirely locally, with no communication to external cloud services. Users are, however, reminded to follow proper data management practices when working on Hábrók.