✳ ISMIR 2025 · Late-Breaking Demo
From MSc thesis to ISMIR publication
The thesis was condensed into a Late-Breaking Demo presented at the 26th ISMIR conference in Daejeon, South Korea.
01 · Daejeon, South Korea · September 2025
Presented at the 26th ISMIR
The Late-Breaking Demo was presented at the 26th International Society for Music Information Retrieval Conference, on the KAIST campus. The trip was supported by an ISMIR First-time Authors Grant and research grants from Leiden University.
02 · Two distinct works
MSc Thesis
Probing the World for Groove: A Comparative Study of Transfer Learning for Drum Audio Style Classification
ISMIR 2025 Late-Breaking Demo
A Comparative Study of Transfer Learning for Drum Audio Style Classification
These are separate citations. The Late-Breaking Demo is a short extended abstract derived from, but not identical to, the full thesis.
03 · Official abstract · ISMIR 2025 LBD
Research in drum classification often trends in two key directions: (1) audio-based single-instrument classification, or (2) automatic drum transcription, which makes exploration of drum-audio style compelling, especially in comparison to a CNN trained on the GMD (Groove MIDI Dataset) with a pretrained transformer-based model, PaSST (Patchout Audio Spectrogram Transformer), with frozen general-audio embeddings from AudioSet. This comparison reveals the ways in which general audio knowledge can affect drum-audio style classification. Experiments with model depth, augmentation, and padding show that PaSST with these frozen embeddings reduces performance in terms of accuracy but reveals a robust feature representation distinct from the CNN.
04 · Presentation
Watch the demo
Hosted on YouTube · also linked from the official ISMIR 2025 program page.
05 · Cite this work
Citation and BibTeX
Eriksen, T. (2025). Probing the World for Groove: A Comparative Study of Transfer Learning for Drum Audio Style Classification. MSc thesis, Media Technology, Leiden University.
@mastersthesis{eriksen2025probing,
title = {Probing the World for Groove: A Comparative Study of
Transfer Learning for Drum Audio Style Classification},
author = {Eriksen, Trent},
school = {Leiden University},
type = {{MSc} thesis, Media Technology},
address = {Leiden, The Netherlands},
year = {2025},
month = jun
}Eriksen, T., van der Heide, E., & Saunders, R. (2025). A Comparative Study of Transfer Learning for Drum Audio Style Classification. Extended Abstracts for the Late-Breaking/Demo Session of the 26th ISMIR Conference, Daejeon, South Korea.
@inproceedings{eriksen2025comparative,
title = {A Comparative Study of Transfer Learning for
Drum Audio Style Classification},
author = {Eriksen, Trent and van der Heide, Edwin and Saunders, Robert},
booktitle = {Extended Abstracts for the Late-Breaking/Demo Session of the
26th Int. Society for Music Information Retrieval Conf. (ISMIR)},
address = {Daejeon, South Korea},
year = {2025}
}06 · Acknowledgements and support
Grants
- ISMIR First-time Authors Grant, August 2025
- Creative Intelligence and Technology Grant, Leiden University, September 2025
- Leiden University Fund Research Grant, October 2025
With thanks to supervisors Edwin van der Heide and Robert Saunders, and to Leiden University Media Technology.