VoiceRadar: Voice Deepfake Detection using Micro-Frequency and Compositional Analysis
Kavita Kumari
Network and Distributed System Security (NDSS) Symposium 2025 · Day 1 · Audio Security
The proliferation of sophisticated **deepfake audio** poses a significant and escalating threat to personal and societal security. From bypassing **voice authentication** systems to propagating disinformation in political warfare and enabling character assassination through fabricated statements, the malicious applications of synthetic speech are diverse and alarming. This talk, presented by Alexandro Pegoro at the NDSS Symposium, introduces **VoiceRadar**, a novel deepfake detection system designed to overcome the inherent limitations of existing detection methodologies.
AI review
VoiceRadar is legitimate academic research with a defensible core idea — using physics-grounded micro-frequency features and energy-based models to generalize deepfake audio detection across TTS and STS domains. The contribution is real, but the framing is doing a lot of heavy lifting: 'drum vibration representation' and 'Doppler effect' sound more impressive in the abstract than they are in practice, and the adversarial robustness claims rely on the assumption that generative models won't adapt — which is a generous assumption to build a defense around.