ToneTwist AFx Dataset: Dry with markers

Dry inputs are a selection of clean guitar and bass recordings from different sources:

• IDMT-SMT-GUITAR - dataset 2 (7:23 min)
• IDMT-SMT-GUITAR - dataset 4 - Career SG (6:08 min)
• IDMT-SMT-GUITAR - dataset 4 - Ibanez 2820 (5:14 min)
• IDMT-SMT-Bass-Single-Track - (5:58 min)
• NAM: Neural Amp Modeler - (3:11 min)
• Private Guitar Data - (5:19 min)
• YouTube Bass Recordings - (10:09 min)

Pre-processing:

• All:
  • synchronization markers (2 impulses) added at start and end of every file
• IDMT-SMT-GUITAR - dataset 2:
  
  • peak normalized to -6dBFS
• NAM:
  
  • no pre-processing
• Others:
  
  • peak normalized to -0.1dBFS
  • signal multiplied by random number every 5 seconds (uniform distribution [0.1, 1.0] = [-20dB, 0dB])

@misc{comunità2025nablafxframeworkdifferentiableblackbox,
      title={NablAFx: A Framework for Differentiable Black-box and Gray-box Modeling of Audio Effects}, 
      author={Marco Comunità and Christian J. Steinmetz and Joshua D. Reiss},
      year={2025},
      eprint={2502.11668},
      archivePrefix={arXiv},
      primaryClass={cs.SD},
      url={https://arxiv.org/abs/2502.11668}, 
}