Konstantinos Drossos

Generalizing DoA estimation across diverse microphone arrays could revolutionize audio processing in mobile and dynamic environments.

A training-free dynamic clustering method significantly improves long speech separation performance, especially in sparse scenarios with unknown speaker counts.

Achieving oracle-level audio equalization performance with a lightweight model could revolutionize real-time audio processing.

Systems with near-perfect EER can still expose significant privacy vulnerabilities, with worst-case disclosures effectively doubling an attacker's success rate.

Context-aware audio denoising can outperform traditional methods by intelligently removing sounds irrelevant to the inferred acoustic scene, even if those sounds would be considered relevant in a different context.