CV
This is a description of the page. You can modify it in '_pages/cv.md'. You can also change or remove the top pdf download button.
Contact Information
| Name | Daniyal Kabir Dar |
| Professional Title | PhD Student, Computer Science |
| dardaniy@msu.edu | |
| Location | , East Lansing, Michigan |
Professional Summary
PhD student at Michigan State University’s iProbe Lab working on audio deepfake detection, speaker identity bias, adversarial robustness, and explainability in biometric systems.
Experience
-
2023 - Present East Lansing, MI
Graduate Research Assistant
iProbe Lab, Michigan State University
Research on audio deepfake detection, speaker identity bias in detectors, adversarial robustness of speech systems, and explainability in biometric systems.
- Developed the Identity Sensitivity Score (ISS), a label-free diagnostic for identity reliance in audio deepfake detectors (IJCB 2026)
- Studied phoneme-level effects of adversarial perturbations on speaker identity (ICASSP 2026, oral)
- Building a contrastive, speaker-normalized explanation-generation system grounded in acoustic measurements for deepfake detection (ICASSP 2027, in progress)
Education
-
2023 - 2028 East Lansing, MI
PhD
Michigan State University
Computer Science
- Adversarial Machine Learning, Pattern Recognition
- Advised by Dr. Arun Ross, iProbe Lab
- Research focus: audio deepfake detection, speaker identity sensitivity, adversarial robustness, explainable biometrics
Publications
-
2026 Probing Speaker Identity Sensitivity in Audio Deepfake Detectors
IEEE/IAPR International Joint Conference on Biometrics (IJCB)
Introduces the Identity Sensitivity Score (ISS), a per-utterance, label-free diagnostic that measures how much a detector’s output changes across speaker-identity contexts.
-
2026 Impact of Phonetics on Speaker Identity in Adversarial Voice Attack
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Analyzes adversarial audio perturbations at the phonetic level, showing they exploit systematic phonetic confusions that degrade speaker-identity cues. Presented as an oral talk.
Skills
Machine Learning & Audio (Advanced): Audio Deepfake Detection, Speaker Verification, Adversarial Robustness, Explainable AI, Biometrics
Programming (Advanced): Python, PyTorch
Interests
Research: Audio Forensics, Speech ML, Biometric Security, Voice AI
Beyond the Lab: Startup & agentic-tool ideas, PC gaming