Algoritmo selector de música basado en el estado de ánimo del usuario utilizando una interfaz cerebro-computador.

Abstract

In this research project two strategies are proposed for the elaboration of a music selection algorithm based on the user’s mood by means of (EEG) electro encephalographic signals. These signals are acquired with the help of a Brain Computer Interface (BCI) called EMOTIV Insigh, which after their collection are processed and quantified, presenting as an average of signals weighted through six mood metrics. As a result of conducting preliminary experiments, detailed in section 5.3., The engagement metric was determined, which presented the best response to different musical genres, based on this performed two types of music selection algorithms that aimed to increase and/or stability of this parameter; the first one based on a heuristic method from which a list of songs or playlist was obtained which increased the metric commitment to a significant level (51/100 with a standard deviation of 4) that was constantly maintained at throughout its duration. The second algorithm consisted of a Vector Support Machine (SVM), which offered the prediction of a specific song based on the current state and previous state of the weighted averages of the “engagement” metric, in response a signal was obtained with an incremental tendency in a general way with the exception of two genders that obtained lower responses to those expected by the system.