A First Step Towards Binaural Beat Classification Using Multiple EEG Devices


This study analyzes the influence of binaural beats, an acoustic phenomenon which rises from two sonic waves of insignificantly different frequencies noticed as a single tone on each ear, on brain activity. Although this topic is not new, few scientific reports have been released. Therefore this study investigates if there is a measurable impact on the EEG activity for different frequency bands using a low budget, a mid-price ranged and a medical EEG device. Probands were exposed to a 12 minute 40 seconds counting task while listening to ambient music. During this task binaural beat samples were played covering different frequency bands, each followed by a pause of 10 seconds where no binaural beat sample was played. The probands were not informed about the presence of binaural beats, the only information given was to count the red shapes moving over the screen. The aim was to classify to which particular frequency the proband was exposed by analyzing their brain activity. The best result was achieved with the medical EEG by the FT classifier with a mean absolute error of 7.14% averaged over 4 datasets. In comparison the best result achieved on the NeuroSky Mindwave headset using the Naive Bayes classifier was a mean absolute error of 26.04%, while introductory experiments with the Emotiv EPOC headset showed a mean absolute error between 2.83% and 27.70%. Further experiments have to be conducted to validate the Emotiv EPOC classification performance. This shows that it is possible to classify the currently played binaural beats frequency rather accurate with appropriate hardware.

In Proceedings of the AIHLS