Audio data augmentation techniques for frame drum stroke recognition

Vasileiou L, Vryzas N, Pagonis A, Dimoulas C. Audio data augmentation techniques for frame drum stroke recognition. In: Express Paper. Audio Engineering Society; 2026. Paper 421. Available from: https://aes.org/publications/elibrary-page/?id=23255

@inproceedings{Vasileiou2026_23255,
  author = {Vasileiou, Labros and Vryzas, Nikolaos and Pagonis, Antonis and Dimoulas, Charalampos},
  title = {{Audio data augmentation techniques for frame drum stroke recognition}},
  note = {Paper 421},
  year = {2026},
  month = may,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=23255}
}

TY  - CPAPER
TI  - Audio data augmentation techniques for frame drum stroke recognition
AU  - Vasileiou, Labros
AU  - Vryzas, Nikolaos
AU  - Pagonis, Antonis
AU  - Dimoulas, Charalampos
M1  - Paper 421
PY  - 2026
DA  - 2026/05/28
UR  - https://aes.org/publications/elibrary-page/?id=23255
PB  - Audio Engineering Society
LA  - en
AB  - This work addresses the problem of the traditional frame drum (bendir) stroke technique recognition in simulated real-world conditions. The playing technique includes three discrete strokes that are used to create rhythmic patterns, dum, tek and slap. In the presented work, audio data augmentation is investigated on a dataset containing recordings of instruments of various construction attributes. The used techniques are selected in the direction of generalizing classification in real-world conditions. Moreover, the mixing of the frame drum samples with accompanying guitar chords is introduced, simulating the more complicated problem of hit technique recognition when playing in a duo. The application of the aforementioned data augmentation leads to the formation of different available datasets for training and testing. Two convolutional neural network architectures (one- and two-dimensional) are taken into consideration, trained on waveforms and melscale spectrograms of the different subsets accordingly.
ER  -