The repository contains the Python code I wrote for the paper:
Erdem, C., Lan, Q., Fuhrer, J., Martin, C. P., Tørresen, J., & Jensenius, A. R. (2020). Towards Playing in the'Air': Modeling Motion-Sound Energy Relationships in Electric Guitar Performance Using Deep Neural Networks. In Proceedings of the SMC Conferences (pp. 177-184). Axea sas/SMC Network.
The code in this repository is all written by Qichao Lan including:
- the code for recording the Myo data from both arms (https://github.com/chaosprint/dual-myo-recorder)
- the code for processing the data such as interpolating some missing Myo samples: interpolation_emg.ipynb
- the code for training the model which can predict RMS from the Myo data: emg_rms_training.ipynb
- the code for predicting RMS from the Myo data (both arms, 16 channels): EMG_to_RMS_both_hands_person_load.ipynb
Cagri Erdem has contributed the Max/MSP patch for the visual hint during recording. Based on the code in this repo, he also further tweaked the model and worked on the statistical analysis. See the non-overlapped part in this repo: https://github.com/cerdemo/air_model