Muscle-Guided Guitar Pedalboard: Exploring Interaction Strategies Through Surface Electromyography and Deep Learning

Davide Lionetti, Luca Turchet, Massimiliano Zanoni, and Paolo Belluco

Proceedings of the International Conference on New Interfaces for Musical Expression

Abstract:

This paper explores a method to innovate the conventional interaction with a guitar pedalboard. By analyzing muscular contractions tracked via surface Electromyography (sEMG) wearable sensors, we aimed to investigate how to dynamically track guitarists’ sonic intentions to automatically control the guitar sound. Two Recurrent Neural Networks based on Bidirectional Long-Short Term Memory were developed to analyze sEMG signals in real-time. The system was designed as a digital musical instrument that calibrates itself to each user during an initial training process. During training musicians provide their gestural vocabulary, associating each gesture to a corresponding pedalboard preset. The selection of the most effective features, in synergy with the best set of muscles, was conducted to optimize the learning rate of the system. The system was assessed with a user study encompassing seven expert guitar players. Results showed that, on average, participants appreciated the concept underlying the system and deemed it to be able to foster their creativity.

Citation:

Davide Lionetti, Luca Turchet, Massimiliano Zanoni, and Paolo Belluco. 2024. Muscle-Guided Guitar Pedalboard: Exploring Interaction Strategies Through Surface Electromyography and Deep Learning. Proceedings of the International Conference on New Interfaces for Musical Expression. DOI: 10.5281/zenodo.13904842

BibTeX Entry:

  @article{nime2024_37,
 abstract = {This paper explores a method to innovate the conventional interaction with a guitar pedalboard. By analyzing muscular contractions tracked via surface Electromyography (sEMG) wearable sensors, we aimed to investigate how to dynamically track guitarists’ sonic intentions to automatically control the guitar sound. Two Recurrent Neural Networks based on Bidirectional Long-Short Term Memory were developed to analyze sEMG signals in real-time. The system was designed as a digital musical instrument that calibrates itself to each user during an initial training process. During training musicians provide their gestural vocabulary, associating each gesture to a corresponding pedalboard preset. The selection of the most effective features, in synergy with the best set of muscles, was conducted to optimize the learning rate of the system. The system was assessed with a user study encompassing seven expert guitar players. Results showed that, on average, participants appreciated the concept underlying the system and deemed it to be able to foster their creativity.},
 address = {Utrecht, Netherlands},
 articleno = {37},
 author = {Davide Lionetti and Luca Turchet and Massimiliano Zanoni and Paolo Belluco},
 booktitle = {Proceedings of the International Conference on New Interfaces for Musical Expression},
 doi = {10.5281/zenodo.13904842},
 editor = {S M Astrid Bin and Courtney N. Reed},
 issn = {2220-4806},
 month = {September},
 numpages = {11},
 pages = {241--251},
 presentation-video = {},
 title = {Muscle-Guided Guitar Pedalboard: Exploring Interaction Strategies Through Surface Electromyography and Deep Learning},
 track = {Papers},
 url = {http://nime.org/proceedings/2024/nime2024_37.pdf},
 year = {2024}
}