./drum.code
Timo Hoogland · Zenodo (CERN European Organization for Nuclear Research) · 2023
https://aimc2023.pubpub.org/pub/micjpv5j For this performance I combine live acoustic drums with live coding of electronic music. During the performance the computer acts as a co-performer/semi-autonomous agent making changes in the code while listening and reacting to patterns that I play on the drums. A multi layer perceptron is trained to classify various patterns that I composed beforehand and it will change the code based on the perceived rhythm. The patterns are recognized via contact-microphones attached to the snaredrum and kickdrum. With a Max-patch the transients of the drums are detect and translate into a binary code of the rhythm, for example I connected patterns to specific decisions such as remove code, add code and replace code. In the cases where the pattern I play doesn't match an exact pre-trained rhythm the classifier will match it with something close based on the training data. This results in a jam together with the computer where I know certain specific patterns result in a specific action while others might surprise me, allowing me to improvise with that. The decisions made that lead to surprises are made based on a markov chain driven by the time of the performance. Where in the beginning of a piece the computer might decide to add more instruments or swap sounds, it may decide to remove sounds near the end. Furthermore the code generated by the computer is based on a combination of short pre-coded snippets of different instruments.