“Four musicians from Prague”

Praga, Carol Bridge (fot. Gosia Sachse)

We know very well that music works on us. It can put you in a good mood and lead to tears, calm you down or act on your nerves. But it not only resonates with our mood. It is also an “amplifier” of the message in films, advertisements, etc.

Researchers from the University of Southern California have decided to check out what makes music work for us like this and not differently. They used machine learning.

A prelude to artificial intelligence

It started with collecting material for testing. Scientists needed songs – sad and cheerful. To do so, they searched online music services and discussion forums to find songs marked “sad” and “cheerful” or their synonyms. They wanted to reduce the likelihood that the participants of the survey knew the songs earlier, so they decided to pick out niche songs with a small number of plays from the web.

Score of emotion

The songs were presented to a group of 100 people. Some of the participants were connected to the apparatus testing brain activity, others had their pulse and skin conduction tested (dermal-galvanic reaction – testing changes in skin electrical resistance under the influence of sweat).

Additionally, the songs were analyzed second by second for 74 properties like dynamics, timbre, harmony and rhythm.

All these data were used to train machine learning models. Their task was to find the relationship between the content of the work and the physiological response of the body, and more precisely to determine which of the several dozen properties should be observed to predict the reaction of the body. This would allow to predict how a given piece of work will affect the person who listens to it.

It turned out, for example, that the tone (i.e. the intensity of medium and high frequencies), volume and clarity of rhythm in sad songs affect brain activity. Hue, complexity, clarity of rhythm and predictability of tone are correlated with heart rate changes.

Solo on the listener

Researchers hope that their work in the future will allow them to create powerful models of machine learning capable of predicting how a piece of music will affect our psyche and what reactions it will cause in our body. Practical applications? Music composed with a specific listener in mind, tailored to his or her liking, music suggestively evoking specific emotions, and finally music as a support for psychotherapy.

(Quelle: https://www.sztucznainteligencja.org.pl/)


Let’s support street virtuosos who often delight us with their work. Thanks to them, among others, the world is not so bad. In memory of Genek Loska (*08.01.1975 – † 09.09.2020).

Genek Loska (Pińczów 2013). Quelle: Wikipedia

A short biography of Genek can be found under the link below: