We all know that more great musicians are dead than alive. Using deep learning we can train a network on the dead artists' songs and create new ones, basically being able to resurrect the greatest bands!
In this project, we trained an LSTM network on the entire corpus of the epic (but dead!) rock band "The Ramones” in order to write and release two new punk rock songs by the artificially intelligent THE RAiMONES.
Presenter: Matthias Frey
Matthias Frey holds a Ph.D. in Electronics from ETH Zurich (2006) and worked in various positions on analog chip design for communications, as well as signal processing applications such as laser distance measuring, light-weight face detection and ultra-sonic speakers, as well as big data. He lived and worked in Japan for more than ten years: As a researcher on A/D Converters at the Tokyo Institute of Technology, as Technical Consultant at Nikon-Trimble, and at the mobile contents provider MTI as Chief Technical Architect in the Advanced Technology Division. For five years he was heading the Science & Technology Office Tokyo of the Embassy of Switzerland in Japan. He is currently working as Technology Scout for Sony Europe in the domains of signal processing, AI and mixed reality. He is a fan of (artificially) intelligent punk rock music.
Can machines be creative? At least to a certain extent, someone would say. What if we could let a machine create a perfect piece, which fits the mood expectations you have, the choice of instruments, and which is completely copyright free. Is it? Who owns the rights, if you feed it with the work of 1,000 of other music artists?
18:45 Training an LSTM to revive The RAiMONES
19:10 Moderated Discussion
19:30-21:00 Apero / Networking