Related Experiment Video
Updated: May 22, 2026
05:48
Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis
Published on: August 9, 2024
Single-trial EEG-based classification reveals instrument-specific timbre perception via traditional machine learning
P Satkunarajah1, S D Power2, B R Zendel1
1Population Health and Applied Health Sciences, Faculty of Medicine, Memorial University of Newfoundland, Canada.
Neuroimage
|May 20, 2026
Summary
This study shows traditional machine learning can identify musical instruments from brainwaves (EEG). This research could lead to hearing aids that adapt to what music you want to hear.
Area of Science:
- Neuroscience
- Machine Learning
- Bioacoustics
Background:
- Hearing aid users often struggle with music perception.
- Future hearing aids may adapt to user intent by monitoring brain activity.
- Selective amplification of desired musical instruments is a potential application.
Purpose of the Study:
- To assess if traditional machine learning can identify musical instruments from single-trial electroencephalography (EEG).
- To explore the efficacy of various classifiers and feature sets for EEG-based musical instrument identification.
Main Methods:
- Recorded EEG from 73 electrodes while participants listened to tones (Trombone, Clarinet, Cello, Piano, Pure Tone).
- Investigated Linear Discriminant Analysis (LDA), Gradient Boosting (GB), Support Vector Machine (SVM), and k-NN classifiers.
- Utilized raw EEG, event-related potential (ERP)-based, harmonics-based, and regularity-based features.
- Analyzed N1 and P2 ERP components for instrument-specific differences.
Main Results:
- All four classifiers performed significantly above chance (20%) using raw EEG features (LDA: 34%, GB: 35%, SVM: 33%, k-NN: 26%).
- Precision, Recall, and F1-scores correlated with overall accuracy.
- Cello produced the largest P2 amplitude and earliest N1 latency; Pure Tone had the smallest P2 amplitude; Clarinet had the latest N1 latency.
Conclusions:
- Traditional machine learning methods show potential for identifying musical instruments from EEG data.
- Further improvements may be achieved with advanced algorithms or feature transformations.
- Distinct ERP characteristics (N1 latency, P2 amplitude) vary across different musical instruments and pure tones.