DeepMark at EUSIPCO2026

Last week, Petar Nedić and I were in Bruges for EUSIPCO 2026, where we presented AWARE - Audio Watermarking via Adversarial Resistance to Edits, our novel approach to audio watermarking. With AWARE, we introduce a new way of approaching audio watermarking, and demonstrate strong performance against state of the art methods, including Microsoft’s WavMark and Meta’s AudioSeal, achieving greater robustness while remaining highly imperceptible. If you’re interested in learning more, you can check out the paper here: https://arxiv.org/abs/2510.17512
Audio watermarking has become a hot topic with the rise of generative audio. Embedding an inaudible, robust watermark is one of the most practical ways to mark content and trace its origin, but the watermark has to survive compression, resampling, and all the other processing audio goes through in the real world. That's the problem AWARE tackles, and it's where it beats the existing baselines.
EUSIPCO itself is a great conference for this kind of work it brings together a lot of engineers and researchers working on real signal processing problems, not just theory. We had plenty of good conversations after our oral presentation: robustness against neural codecs, and where the field is heading with AI generated audio.
Of course, we couldn’t visit Belgium without exploring a little. In our free time, we took the opportunity to explore Bruges and experience a little of Belgian culture, from wandering around the city to enjoying a couple of the country’s famous beers. We had a great time and are sharing a few photos from the experience.
Mark what you generate. Prove what you ship.
Compliance you can demonstrate today, and a path to verified AI voice your customers actually trust, both built into the media you already produce.
