Research Roundup: Debashis Sinha
Debashis Sinha is an Assistant Professor in Production at the School of Performance. Discover his ongoing research collaborating with neural networks and other technologies as he explores audio storytelling in a virtual space.
I have been working in sound design and composing music for theatrical production for many years, work that intersects with my own research into audio storytelling on various platforms and using various tools. Currently, a large portion of my research is focused on using machine learning and AI processes to generate sound, text, and video content that arise from various story seeds, specifically seeing how I can leverage and re-use those outputs to explore concepts of failure, sharing, scale and refusal.
It’s lovely to see live theatre slowly return [since the onset of the Covid-19 pandemic]. During that time, I’ve created and collaborated with many theatre companies to make theatre-adjacent digital art - audio dramas, short films, and the like. That work of necessity has been separated from others - we’ve collaborated through zoom calls, email and phone conversations. I’ve missed being in a room with people and feeding off that real-time energy, as those conversations give rise to new ideas, momentum and strategies of creation.
Regarding the research that I do on my own, it’s not really been affected [by the pandemic], as much of it (until now) has been me toiling away in my studio on my own. I will say I’ve had way more access to international festivals, events and conferences, so that’s been at least one positive outcome. I’m hopeful that kind of access will continue, as it includes so many more people locally and internationally, and the discussions and knowledge-sharing is so much more robust as a result.
I’ve always been interested in cultural transmission, particularly cultures that don’t get much airtime in our societal consciousness due to demographics, structural inequity, etc. In my case, this takes place through a kind of speculative mythology practice, where I “uncover” or create alternative myths adjacent to the canon of the more traditional Hindu stories I grew up with. I construct these myths and ruminations through the lens of sound-centric and sonic arts practice - fixed media, AV performance, sound installations, video art. It’s been a lot of exploration over the years using various technologies.
[This research is] at its heart about expanding the conversation around storytelling and ways in which we communicate culture. It’s also very directly about questioning what we consider to be valid systems of knowledge, and why and how that validity has been and continues to be assigned, and by whom.
The work I’ve been doing with machine learning and sound questions this quite explicitly - if a model is trained to “know” that “music = jazz”, then when you present it with, say, classical Hindustani music, what are the outcomes? What does those outputs say about the model, the datasets, and the overall assumptions of building that tool in the first place? How do we use those “failures” as commentary in our own expression, and how do our processes of refusal in fact create extensions to strategies of sharing knowledge? We will continue to express ourselves as human beings in the face of changing technological imperatives. The work I am interested in is to discover ways in which we can do it well, with deep and moving power.
Be curious! And let your curiosity drive your explorations!
Check out some of Deb's work
Deb has two new records being released on the Establishment Records label this year. Visit his Bandcamp (external link) and personal website (external link) to see (and hear!) more.
Images generated through Deb's more recent work.