Sounding AI

04th Feb 2024 04th Feb 2024 12:00 13:30

Matthias Jung, Ada Ada Ada, Aline Zara. Host: Anita Jóri

00:00

radialsystem

Free entry

2

  • Anna Von Hausswolff

    SE

    Pushing the boundaries of sound through the use of pipe organs, the Swedish composer evokes a fluidic plain that swells with the harmonies of alien celestials.

  • Concepción Huerta

    MX

    With a sharp instinct for compelling audiovisual collaborations and perpetual activity within noisy fields, the Mexican photographer, composer, and video artist conjures continuously intriguing sound narratives.

    • The Research Networking Day is an exchange platform for graduate or postgraduate students – as well as independent artists conducting self-guided research – traversing the fields of music, sound, arts, media, design, and related theoretical disciplines. RND 2024 will assemble nine students, scholars, and artists/researchers from a variety of fields of study and approaches who will present research touching on CTM's Sustain festival theme. The selected candidates will give short presentations (10 min.) within different thematic modules, with discussions after each presentation and at the end of each session.

    Photo: Eunice Maurice

    Welcome

    In a short welcome word, Anita Jóri of CTM Festival and Berlin University of the Arts (UdK) introduces CTM 2024's »Sustain« theme and the day's three modules that address the topic from a variety of perspectives. She will then take over hosting the first module (due to unforeseen circumstances Christoph Jacke (Paderborn University) will not be able to attend). This RND edition will take place in collaboration with the Berlin University of the Arts (UdK), the Institute Art Gender Nature at the Basel Academy of Art and Design FHNW, the C:POP. Transdisciplinary Research Center for Popular Music Cultures and Creative Economies of the Paderborn University, and the Berlin-based network and project space Trust.

    Symbiosis as a Sustainable Paradigm for Computational Creativity

    Matthias Jung (PhD candidate, University of Agder, Kristiansand, NO)

    Many search-based AI methods for music generation are based on evolutionary coding and genetic algorithms today. These computational approaches are inspired by evolutionary biology and often tend towards the Darwinian idea of the survival of the fittest: sustaining the most adaptable individuals of a population to create the next generation. This evolutionary principle, however, is not the only that has been suggested to explain evolutionary life. Lynn Margulis, for example, has suggested in the 1960s that symbioses play a much bigger role during evolution than previously assumed. Individuals and groups of the same and of different species have been forming collaborative entities and in fact even the cells in the human body have evolved as a merger of before separate organisms.

    This brings up the question of what we might learn from these symbiotic associations for the field of music coding and how we can apply this principle for creating sustainable creative scenarios. Can biological theory even inspire the way we interact with creative technology in the first place, i.e. can we see human-computer interaction as a form of symbiosis and how can we use that perspective for understanding computational creativity? Finally, on a more concrete level, what are ways to integrate symbiotically intelligent organisms into creative practice and how can those contribute to sustain planetary life after all?

    Matthias Jung holds a Bachelor from Paderborn University in Popular Music and Media and a Master in Electronic Media from Stuttgart Media University. Before starting his Ph.D. in Norway, he worked at Berlin’s state academy as audio production specialist and headed the programme team of Most Wanted: Music conference. His recent research focuses on Intelligent Music Performance Systems, bio-inspired coding and creative symbioses, some of which he carried out at Georgia Tech University as a visiting scholar.

    Cultures of the Paralinguistic in AI Voice Cloning Tools

    Ada Ada Ada (Research Assistant, IT-University of Copenhagen, DK)

    What happens when voice design becomes a matter of statistical analysis and averaging? Artificial intelligence tools are rapidly invading the world of sound. The field of voice cloning has particularly gained a lot of attention for both its ethical and legal ramifications. AI tools depart significantly from previous synthetic voice tools, due to their inherent reliance on statistical analysis. From a research angle, we dissect what this departure means for the tools' paralinguistic capabilities. In short, we want to know more about the politics and aesthetics of AI voices, and how they are able to perform.

    In this presentation, we discuss how various voice cloning tools create cultures of the paralinguistic in different ways. We look at e.g. laughter, stuttering, pacing in voice-to-voice and text-to-speech tools like Microsoft Azure, ElevenLabs, and FakeYou. We speculate on how the limitations and possibilities inherent in AI tools can shape the future of voice, music, and sound design.

    Ada Ada Ada is an algorithmic artist and research assistant. She works with gender and bodies as perceived by computers through algorithms, software and »artificial intelligence.« Her own trans experience often comes into play in her works. As part of Voice as a Matter of Design at the IT-University of Copenhagen, she discovers novel ways of using »AI« voice cloning software. She investigates how the design space of synthetic voices can be broadened to include more vocal identities.

    Sustaining Voice: AI-generated Voices and the Human

    Aline Zara (PhD student, University of Toronto, CA)

    We are increasingly surrounded by AI-generated voices: the voice of the telephone answering machine, the voice of the train announcer, or the voice of the TikTok narrator. Some we can identify as AI-generated by ear, but as technology progresses these synthetic voices lose their robotic drawl to become nearly human, with realistic paces, pitches, and intonations alongside localised and racialised accents. We are adeptly and recognisably cloning the human voice – deconstructing it into units and recombining it into mimetic tonalities, words, and phrases via neural networks – to exist in presumed perpetuity. Does this cloning succeed in its overarching project to sustain?

    Following the work of Dominic Pettman (2017), these machinic and cybernetic voices forge new social relationships as disembodied and extrahuman. However convincing, AI-generated voices may signify less as clones of their original human voices and more as simulated, hyperreal aurality – an expanded potential of what the human voice and experience can sound like. Grounded in the study of voice, technology, and critical AI, I reposition and relisten to AI-generated voices within human and non-human sustainability.

    Aline Zara is a PhD student at the University of Toronto’s Faculty of Information in Media, Technology, and Culture, and is part of the Book History and Print Culture collaborative programme. She works as a researcher in the Canadian book industry with BookNet Canada and is the founder and publisher of nAIrrative Press, a publishing house devoted to AI-written, AI-narrated audiobooks. Her doctoral research focuses on AI-generated voices, orality, and storytelling in audiobook narration.

    Event Access

    Please consult our Accessibility section for information on volume / lighting, mobility, age restrictions and other access information.

    Supported by SHAPE+ which is co-funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them.