Deutsch
English

SOUND SYNTHESIS WITH RECURRENT NEURAL NETWORKS

We are happy to announce that we will be hosting a lecture by Kosmas Giannoutakis on the subject of sound synthesis with recurrent neural networks on Thursday, 26.05.2016. The lecture will start at 10:00 and will last till 11:30, in ELA 7.

The entrance is free!

 

Kosmas Giannoutakis creates dynamic sound artworks by interconnecting human agents, sound bodies, acoustic sites and digital audiovisual systems through the medium of sound. Using feedback mechanisms in order to create complexity and to control non-linearity, he is researching the catalysis and communication of emergent sound phenomena.

 

Kosmas's work have been presented in various festivals and workshops, such as inSonic2015 and next_generation in ZKM Karlsruhe, Soundislands Festival/2nd International Symposium on Sound and Interactivity in Singapore, Gaudeamus Muziekweek in Utrecht, REAL/UNREAL BEAST FEaST 2016 in Birmingham, GENERATE! Festival für elektronische Künste in Tübingen, EUROMicroFest in E-Werk Freiburg, XXIX Summer Sounds Festival in Finland, the Avaton Music Festival in Cyprus and 7th international workshop for young composers in Mazsalaca, Latvia.

 

The Institute of Electronic Music and Acoustics - IEM of the University of Music and Performing Arts Graz, is the current inspiring environment for his interdisciplinary art experiments.

 

 

SOUND SYNTHESIS WITH RECURRENT NEURAL NETWORKS

 

 

Recurrent Neural Networks are state-of-the-art digital signal processing techniques for engineering tasks (signal reconstruction, system identification, echo cancellation). In this lecture audio rate digital neurons interconnected with a recurrent topology will be introduced and their musical potential for sound synthesis will be investigated. Different algorithms, which affect the plasticity mechanisms of the neurons will be demonstrated and further research topics in this very promising field will be suggested.