News-Article

Audio transcription and formula recognition: DSC Supports Didactics of Mathematics with an AI-Based Workshop

In the research seminar of the Didactics of Mathematics group, Annika Nolte and Nele Fuchs demonstrated how audio data can be automatically transcribed using the right tools – including a tailored solution for handling mathematical formulas.

Annika Nolte and Nele Fuchs delivered a practice-oriented workshop for the research seminar of the Didactics of Mathematics group at Faculty 03 within the framework of the DataNord project. Held on January 13, the session focused on automated transcription of audio data and its application in qualitative research contexts.

The 90-minute workshop introduced concrete tools for automatically transcribing spoken content. In particular, it showcased the open-weight model Whisperand the open-source software NoScribe, which builds on Whisper while offering a user-friendly interface for transcription workflows.

A specific challenge: transcribing mathematical formulas

One focus addressed a challenge that frequently arises in mathematics education research: the transcription of mathematical formulas. Spoken mathematical expressions are typically rendered as plain text by conventional transcription systems, which can be limiting for further analysis.

To address this issue, Annika Nolte and Nele Fuchs presented an extended workflow in which an additional, specialized language model is applied after the initial transcription. This model translates detected formulas into proper mathematical notation, ensuring that formulas are represented using symbols rather than spelled-out words. As a result, mathematical content can be documented in a more structured and analytically usable way.

Strengthening data skills in subject-specific research

The workshop illustrated how data-related tools can be adapted to disciplinary needs and integrated into existing research processes. In qualitative studies based on audio recordings – such as interviews, classroom observations, or reflective discussions – automated transcription can significantly reduce time demands while improving transparency and traceability.

For the DSC, collaboration with subject-specific research groups like the Didactics of Mathematics team is an important part of providing data-related support across disciplines. Formats such as this workshop ensure that data skills are conveyed not in the abstract, but directly within the concrete contexts of everyday research practice.

Interested in a workshop for your research group or institute?

If you have questions about data, research data management, or data science – or if you would like to explore a similar format for your research group or institute – please contact our coordinator Dr. Lena Steinmann. For further inspiration, feel free to take a look at our existing training portfolio.


If you have any questions, please contact:

Nele Fuchs
DSC Data Scientist
Tel. +49 (421) 218 59853
E-Mail: n.fuchsprotect me ?!uni-bremenprotect me ?!.de

Annika Nolte
DSC Data Scientist
Tel. +49 (421) 218 59856
E-Mail:anolteprotect me ?!uni-bremenprotect me ?!.de

Illustration eines Laptops mit transkribiertem Audiotext und mathematischen Formeln, symbolisch für automatische Transkription in der Forschung
Updated by: News