talks
Talks and presentations I have given, newest first.
2026
AI: Embrace, Adapt, Resist? (upcoming)
Panel discussion, 45 minutes. Danish Sound Day 2026, Danish Sound Cluster, DTU Skylab, Denmark, 11 November 2026
Panel with Miriam Plon Sauer (Global Chief Strategy Officer, AKQA), Cumhur Erkut (Associate Professor and AI Engineering lecturer, Aalborg University) and Lars Tirsbæk (Head of Studies, Sonic College), moderated by Nick Dunkerley (Founder, Hindenburg Systems). AI is changing how we develop products, produce sound and teach the next generation. Should we embrace it, adapt to it, or resist it?
2025
A new era of AI: Dozens of miniature brain models
Podcast interview. Baris & Baris (YouTube podcast), Online, 4 June 2025
In Turkish
Conversation on artificial intelligence and virtual reality, covering my academic path, my time in Finland, the links between brain function and AI, deep learning, transformer architectures, the limits of large language models, data quality and synthetic data, and the role of small models and statistical methods. Also covers Active Inference.
2024
Wellbeing beyond expectations with an Edge Intelligence Infrastructure
Dialogue Workshop: Mission Wellbeing of Children and Young People, Aalborg University Innovation Hub, Aalborg, Denmark, 14 March 2024
In 2024, in many occasions I have pitched for an Edge Intelligence infrastructure that can connect our labs and people’s homes. The last was the Dialogue Workshop on 14 March 2024 at Aalborg. I recorded the essence of my talk at the airport while waiting for a cancelled flight. It will hopefully contribute to an AAU mission project with the tentative title: Novel technologies for the well being of children and the young.
2022
Operationalizing Sound and Music with Foundational AI Models
C4DM - Centre for Digital Music, Queen Mary University of London, UK, Online, 10 December 2022
Impact of MLOps on the recent work at Aalborg University and the Multisensory Experience Lab. I emphasized that there are no foundational models for audio yet (but we may expect them soon), and in January 2023 they started to pour in batches!
Enhancing Sound, Music, and Movement Computing with Differentiable DSP
Sound and Music Computing Conference (SMC’22), Saint-Étienne, France, Online, 9 June 2022
Impact of Differentiable Digital Signal Processing (DDSP) on the recent work at the Multisensory Experience Lab. We can deploy DDSP-based real-time plugins fast because we started the work about 25 years ago, and we are implementing the principles of MLOps. The talk compares this approach with what was used two decades ago with shallow-learning models, self-recorded or synthetic data, and self-written code and frameworks.
2021
Multisensory Processing at MEL
Danish Sound Cluster, Struer, Denmark, Online, 7 December 2021
Recent work at the Multisensory Experience Lab on physics-based audio and movement models beyond deep learning, and what is to come: graphical and physics-based deep learning. The lab investigates combinations of input and output modalities in interactive applications, in areas such as health, rehabilitation, education, art and entertainment. MEL focuses on physics-based deep learning and differentiable programming, aiming for interpretable models that optimise well-understood signal processing algorithms with structural biases rather than a generic black box.
2020
Towards Differentiable Sound and Music Computing
AES Applications of Machine Learning in Audio, New York, NY, and online, 29 September 2020
Starting from differentiable IIR filters, I extrapolate the current state of the art in neural audio synthesis towards a research agenda on Differentiable Sound and Music Computing.
2018
Sonic Interactions in Mixed Reality
Talk. Boğaz’da Yapay Öğrenme İsmail Arı Yaz Okulu 2018 (Summer School on Machine Learning, Speech, Language and Bioinformatics), Turkey, 3 July 2018
In Turkish
Sound interactions for mixed reality, presented at the İsmail Arı Summer School 2018 (2–5 July 2018).