Computers can now predict a song you’re listening to via brain scan
In a new report from the D’Or Institute for Research and Education, scientists claim that they can predict a song someone is listening to based strictly off a scan of their brain within 85% accuracy.
Using magnetic resonance images (MRI), the scientists scanned the brains of six volunteers as they listened to selections of rock, pop, jazz, classical, and folk music to establish a baseline. The criteria were further narrowed to include tonality, dynamics, rhythm, and timbre to give the computers an even more nuanced reference points.
The study hinges on the idea of “brain decoding,” which claims that the brain reacts in specific fashions to certain stimuli. After the initial scans were taken, scientists performed the experiment in reverse and had the computers predict what song was being listened to.
When presented with a choice of two options, the computer was accurate up to 85%. When offered ten options, the computers were still able to predict the correct choice to 74%.
While impressive in and of itself, scientists hope this research can be used to improve brain-computer interfaces for individuals who incapable of hearing or speech.