David Combei

Junior Machine Learning Researcher

Research interests: deepfake detection, contrastive learning, interpretability in speech classifiers, speech emotion recognition, audio processing

About Me

I am a passionate Junior ML researcher and engineer focused on audio processing and deepfake detection. With a strong foundation in machine learning and a deep interest in audio-related AI tasks, I’m currently pursuing my first year as a master’s student in AI and preparing for a PhD. I contribute to multiple research projects in both academia and industry. My work aims to bridge the gap between robust ML models and real-world audio challenges.

I am working under the supervision of Adriana Stan at the Technical University of Cluj-Napoca.

Research Collaborations & Projects

AI4TRUST
Solutions

Publications

  • David Combei, Adriana Stan, Dan Oneață, Nicolas Müller, Horia Cucu, "Unmasking real-world audio deepfakes: A data centric approach", In Proceedings of Interspeech, 2025. [pdf]
  • Adriana Stan, David Combei, Dan Oneață, Horia Cucu, "TADA: Training-free Attribution and Out-of-Domain Detection of Audio Deepfakes", In Proceedings of Interspeech, 2025. [pdf]
  • David Combei, Adriana Stan, Dan Oneață, Horia Cucu, "WavLM model ensemble for audio deepfake detection", In The Automatic Speaker Verification Spoofing Countermeasures Workshop (ASVspoof 2024), pp. 170–175, 2024. [pdf]

Events & Presentations

  • Eastern European Machine Learning Summer School 2025 - I am grateful to have been selected to attend the very competitive EEML summer school in Sarajevo.
  • Interspeech 2025 — I will be presenting the papers “Unmasking real-world audio deepfakes : A data centric approach” and "TADA: Training-free Attribution and Out-of-Domain Detection of Audio Deepfakes" in Rotterdam.
  • Romanian AI Days 2024 — I was one of the students selected to present my BSc thesis entitled "Text and Speech Emotion Recognition using pre-trained self-supervised models".