60 auditory Scenes For Investigating CocktailParty Deafness (2021)

For Orchestra With Electronics

Commissioned by 64th Interantional Festival of Contemporary Music “Warsaw Autumn” and Wien Modern

line-up: 3 flutes (1.piccolo), 3 oboes, 3 clarinets (3rd also bass clarinet), 4 horns, 2 trumpets, 2 trombones, 1 tuba, 2 percussionists, 2 keyboards (88keys, using both 2 Sample Banks each), 8 – 6 – 4 – 2 – 2
duration: 20 Minutes
date of premiere: 25.09.2021, Filharmonia Narodowa Warsaw

Natural auditory environments involve many different concurrent sounds. They all sum together to one mixed signal that enters the ear. How does a human brain extract information from this permanent overwhelming chaos of surrounding sound waves? This problem is commonly known as The Cocktail Party Phenomenon.

The challenge of an auditory system in this situation is to find out which parts of the sound should be grouped together and treated as parts of the same source or object. Grouping them incorrectly can cause the listener to hear non-existent sounds built from the wrong combinations of the original components.

The cocktail party problem is a significant challenge for computational audition and machine listening systems. The remarkable ability of a human brain to direct attention to the sound of interest while ignoring the others, as well as switching attention between the sources in a multilayered sound environment is still, to this day, unmatched by machines.

The following examples are taken from a Standard Computational Auditory Scene Recognition Test (SCASRT).  In this test, humans, on average, were able to recognize 57 different scenes with an accuracy of 78%. The best obtained recognition rate of a machine listening system was 56 % for 33 different scenes by the Cithaeron Audition Model  3.1.