Single-trial EEG-based classification reveals instrument-specific timbre perception via traditional machine learning classifiers

Many users of hearing aids report challenges when listening to music. In the future, it may be possible to develop hearing aids that monitor brain activity in real-time and adapt their output to the volitions of the user. In music, this could mean selectively amplifying the sound of the instrument the listener wants to hear....

Continue reading at PubMed →

No actionable change — this is early-stage brain-computer interface research with no clinical translation yet; audiologists should not change practice based on these findings.

Demonstrating that instrument-specific sound qualities (timbre) can be decoded from single-trial EEG opens a long-term research pathway toward neurofeedback-driven, music-optimised hearing aid signal processing.

1. 01Single-trial EEG signals were classified by machine learning to identify which musical instrument a listener heard.
2. 02The study focuses on timbre — the tonal quality that distinguishes instruments playing the same note.
3. 03Traditional (non-deep-learning) ML classifiers were used, making the approach relatively interpretable.
4. 04Authors propose future hearing aids could adapt in real-time to a user's neural responses during music.
5. 05Published in NeuroImage, a peer-reviewed neuroscience journal.

PMID

42162676

DOI

10.1016/j.neuroimage.2026.122010.

Journal

NeuroImage

Publication type

research_article

Evidence level

4

Population

Human participants undergoing EEG recording during music/instrument listening tasks (specific demographics not stated in abstract)

Intervention

Single-trial EEG decoding of instrument-specific timbre perception using traditional machine learning classifiers