#wekinator

Day 1: This is a blog where I am planning to document the progress of my Sonic Arts project for my masters in Goldsmiths University. In this project, I am exploring Machine Learning and how it can be used as a means for Augmented Embodiment in the contemporary practice of performing the electric guitar. 

During this workshop we discussed how ML is a technique used in order to get closer to AI. Letting the machine learn through showing repetitive examples and telling the machine what it is. This spurred some thoughts of supervised learning contra unsupervised learning which I brought up in with my group colleagues, which resulted in a very interesting discussing. This discussion led to understandings of how supervised learning makes the designer dominant whereas in unsupervised learning (such as a twitter bot) could potentially lead to racism, sexism and hurtful knowledge. This made us more prone to go with supervised learning in our own AI that we’re developing the concept of throughout this week.

I learnt a bit more about ML, but it was difficult to do this workshop over zoom. Our teacher expressed frustrations over not being able to do this as we normally do it, physically present, and I agree with that frustration. Since it was over zoom we couldn’t do any “hands on” projects or demos together, we could only watch our teacher go over some basic demos and talk about what we had created with wekinator.

DUST - - - a physical sound sequencer

DUST yields to the idea of using physical objects as an apparatus for modulating sounds, much like an acoustic instrument. The outcome is a direct response between image and sound, whereas the actual physical objects are the building blocks of expression.

The program uses computer vision and machine learning as tools for playing music with physical objects. A real-time camera feed is computerised, the image pixels are translated into parameters that control the sound. For translating and mapping the camera feed into sounds, a polynomial regression algorithm is used.

The instrument forms a kind of analogy of a digital audio workstation (DAW), but in a physical state. We created DAWs to imitate analogue mixers and synthesisers. This system will imitate immaterial DAWs by going back to the state of 'real', 'physical' and 'analogue', but nevertheless with a computational method.

Pour Reception - A playful radio

Pour Reception is a playful radio that strives to challenge our cultural understanding of what an interface is and can be. By using capacitive sensing and machine learning, two glasses of water are turned into a digital material for the user to explore and appropriate. The design materials that we have available when designing digital artifacts expands along with the technological development, and with the computational machinery it is possible to augment our physical world in ways that challenges our perceptions of the objects we interact with. In this project, we aim to change the users perception of what a glass is - both cultural and technical.