Dimitrios Bralios
Audio ML Researcher · PhD Candidate at UIUC
Audio ML Researcher · PhD Candidate at UIUC
I am a PhD student at the University of Illinois Urbana-Champaign (UIUC), advised by Paris Smaragdis, working on AI for audio.
I develop efficient and structured representations for neural audio models. My work explores structured latent spaces, content-adaptive frame rates, and latent-domain audio processing, often by adapting pretrained audio autoencoders rather than rebuilding them from scratch.
Recent work includes latent restructuring through Re-Bottleneck, dynamic latent frame rates through Elastic Time, and latent-domain bandwidth extension and upmixing. Earlier, I worked on source separation and generative modeling, including investigating training-data replication in audio diffusion models.
My broader goal is to make generative audio systems more efficient, controllable, and reusable through better representations.
I am always happy to connect and discuss. If any of this sounds interesting, feel free to reach out.
Interspeech, 2026.
IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2025.
IEEE International Workshop on Machine Learning for Signal Processing (MLSP), 2025.
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024.
IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2023.
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023.
