Our second seminar is in two weeks - Thursday 3/3 at 4pm. Email o.ben-tal@kingston.ac.uk if you wish to join.
Human Learning x Machine Learning
Atau Tanaka
Goldsmiths / Bristol Interaction Group / Maison des Sciences de l'Homme Paris Nord
I describe embodied musical performance in interaction with AI technologies. Performer gesture measured through muscle sensing is mapped to sonic granular synthesis through the mediation of a neural network. The concert performance begins with an empty dataset - a tabula rasa. The composition consists of associating a series of corporeal postures with sonic grains. The signal feature space describing these body states are associated to corresponding audio meta-data and a training set is built up live in performance. Machine learning processes are laid bare in the performance dramaturgy, starting with empty data sets, finishing with trained regression models that create a multidimensional gesture-timbre information space. As much as the machine learns the states of the body it is trained on, the human performer learns the musical space created by the neural network, performing through at times unexpected mappings.
___________________________________________________
Dr. Oded Ben-Tal
Senior Lecturer, Music Technology
Kingston University
PI Datasounds, datasets and datasense research network (2022/23)
http://obental.wixsite.com/main
http://soundcloud.com/odedbental
http://obental.wixsite.com/main
http://soundcloud.com/odedbental
ID: 91834177970, passcode: 377186
Oded & Federico
