The papers behind the models

We did not invent this from nothing. These are the papers the drum generator stands on.

CedarGrooveNET's architecture draws on current research in neural drum generation. These are the papers that informed our encoder-decoder LSTM approach.

  1. Lanzendorfer, Grotschla, Galal, Wattenhofer. "MaskBeat: Loopable Drum Beat Generation." ISMIR 2025. arXiv:2507.03395
  2. Brosnan. "DARC: Drum Accompaniment with Fine-Grained Rhythm Control." 2026. arXiv:2601.02357
  3. O'Reilly, Barnett, Flores Garcia, et al. "The Rhythm In Anything (TRIA)." ISMIR 2025. arXiv:2509.15625
  4. Dahale, Talwadker, Rao, Verma. "Generating Coherent Drum Accompaniment With Fills and Improvisations." 2022. arXiv:2209.00291
  5. Hutchings. "Talking Drums: Generating Drum Grooves with Neural Networks." 2017. arXiv:1706.09558
  6. Makris, Zixun, Kaliakatsos-Papakostas, Herremans. "Conditional Drums Generation using Compound Word Representations." 2022. arXiv:2202.04464

Related pages

TechnologyHow the pieces fit together. The drum transcriberThe Transformer that writes out what it hears. The generation modelThe LSTM and its post-processing.