Thomas Thebaud

Thomas Thebaud

Currently transitioning from the Johns Hopkins University, USA to the AMIAD, FR. I work on speech technologies, AI, and security, with a focus on speech anonymization, deepfake detection, speech-aware large language models, and speaker recognition. My research explores trustworthy and privacy-preserving speech processing, as well as national and international collaborations.

About me

After serving as an Assistant Research Professor in the Department of Electrical and Computer Engineering at Johns Hopkins University, I am joining AMIAD in Palaiseau, France. I received my Ph.D. through a joint program between LIUM at Le Mans University and Orange Labs in Rennes, where I investigated reconstruction and spoofing attacks against voice and handwriting biometric systems.

My research lies at the intersection of speech and audio processing, machine learning, biometrics, and security. I develop methods for trustworthy and privacy-preserving speech technologies, with particular interests in speaker recognition, speech anonymization, deepfake and adversarial-attack detection, and the robustness of speech-processing systems. This work has included the SHADOW, GARD, and RED projects, as well as research on voice-privacy attacks and defenses.

I also study how foundation models and multimodal language models can support the analysis of spoken interactions. Through the EDART and SPEAR projects, my work has addressed speaker characterization, dialogue annotation, speech-aware language models, and the evaluation of conversational speech-to-speech systems. Representative first-author contributions include ProPS: Prompted Profile Synthesis for Natural Language-Conditioned Speaker Embedding Distributions, SPEARBench: A Benchmark for Naturalness Evaluation in Streaming Speech-to-Speech Language Models, and Speaker Verification with Speech-Aware LLMs: Evaluation and Augmentation.

A complementary strand of my research investigates digital biomarkers for neurodegenerative disease detection and severity assessment. Within the NLS and ADPIE projects, I have worked with speech, handwriting, eye movements, and gait, including recent work on interpretable handwriting features for clinical assessment. Further details are available on the Projects and Publications pages.

I welcome national and international research collaborations. Students interested in future research opportunities can consult the Join Us page for current information.

(Updated July 2026)