Target Speaker Detection with Concealed EEG Around the Ear.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Assistive Technology

Background:

• Target speaker identification is crucial for speech enhancement in hearing aids.
• Electroencephalography (EEG) has shown potential for identifying target speakers.
• Integrating compact EEG systems into hearing aids could significantly improve functionality.

Purpose of the Study:

• To compare the performance of a novel wireless around-the-ear EEG system (cEEGrids) with traditional high-density cap EEG for target speaker identification.
• To evaluate the feasibility of using concealed ear-EEG for brain-computer interface applications in hearing aids.

Main Methods:

• Collected concurrent EEG data from 20 normal-hearing listeners using both a laboratory wired cap-EEG system and a wireless mobile ear-EEG system (cEEGrids).
• Applied an envelope tracking algorithm in a competitive speaker paradigm to analyze the EEG data.
• Assessed the accuracy of target speaker identification from both EEG systems.

Main Results:

• The cEEGrid ear-EEG system achieved above chance-level target speaker identification with 69.3% accuracy.
• The traditional cap-EEG system yielded a higher accuracy of 84.8%.
• Ear-EEG signal quality, specifically channel impedance variability, did not significantly affect the envelope tracking performance.

Conclusions:

• Concealed ear-EEG technology using cEEGrids demonstrates potential for brain-computer interface control of hearing aids.
• While cap-EEG provides higher accuracy, ear-EEG offers a more practical, integrated solution for assistive devices.
• Further development of ear-EEG systems could enhance speech processing and user experience in hearing-impaired individuals.