Amity University Noida has done it again, what it does best. Another accolade added to the list. Two of its PhD researchers secured the first position in the interdisciplinary research category at ANVESHAN 2025 Student Research Convention. Their team or project, “Kryoniq”, addressed a critical and current threat of the artificial intelligence age: detecting fake voices that are generated using AI. This particular threat is affecting billions across the globe as impostors are using it to loot a huge sum of money from people.

Reshaping the DeepFake as a Real Issue

Kryoniq is an advanced audio deepfake detection system that fundamentally rethinks detection as a computer vision problem. Instead of traditional signal processing approaches, we analyse spectrograms the way forensic phonetics experts do, visually identifying anomalies in vowel utterances that betray synthetic speech. Our multi-modal framework combines YOLOv5-based language-agnostic vowel detection (82.6% faster than manual analysis, validated across 10 languages), neuroscience-inspired Cross-Frequency Coupling with Genetic Algorithm optimisation (3.44% EER, 52.6% improvement over SOTA) for audio deepfake detection, and our published challenging dataset ArcticEcho (24,752 speaker-controlled samples challenging enough that humans achieve only 55% accuracy). The system provides real-time, explainable, forensic-grade detection with courtroom-ready evidence reports, addressing the critical gap between current black-box detectors and the needs of legal, financial, and security sectors facing the $12.3 billion annual threat of deepfake fraud.

Significance of Research

Audio deepfakes erode public trust, enable massive financial fraud, and compromise the integrity of legal proceedings. Current detection tools are black boxes that cannot explain their decisions, work only for Western English, and lack forensic validation for courtroom use. Kryoniq's significance lies in its paradigm shift: we systematically apply computer vision to forensic speech analysis, "seeing" what synthesis algorithms get wrong rather than just "hearing" it. This approach is inherently language-agnostic, protecting India's linguistic diversity while our explainable AI generates reports that are human-readable.  By publishing the ArcticEcho dataset open-source and achieving state-of-the-art results through neuroscience-inspired methods, we are establishing a new research direction that makes deepfake detection transparent, trustworthy, and accessible to legal and investigative professionals who need reliable tools to combat AI-driven deception.

Conclusion

The first position secured by the Kryoniq team at ANVESHAN 2025 is the result of Amity University Noida’s commitment to onboarding the cutting-edge interdisciplinary research. Global university ranking platforms, such as THE Ranking, have ranked Amity University Noida in the top 100 globally for multidisciplinary research. Groundbreaking research like deepfake detection is a testament to this direction.