Yi Zhu

About Me

I am currently a 4th-year PhD candidate at Multisensory Signal Analysis and Enhancement Lab (MuSAE), National Institute of Scientific Research (INRS) (Montreal, Canada), supervised by Prof.Tiago H. Falk. I am also working as an Applied Scientist II at Reality Defender .

My research is in general focused on learning representations that are generalizable, robust, and explainable from multi modalities (e.g., sound, speech, biosignals, sensor signals). My PhD projects are centered around sound and speech tasks, before that I worked on physiological signals (e.g., sEMG) and human-machine interface during my Master’s at UMN.

Research Interests

Multimodal Signal Processing: Audio, speech, sensors, and bio-signal analysis and understanding
Deep Learning: Self-supervised learning; Representation learning
Speech: Classification, detection, and generation

Awards

[June. 2023] Rising Star in Signal Processing at ICASSP 2023 for PhD research

News

[Sept, 2024] Paper accepted at NeurIPS 2024 on a novel speech SSL representation for deepfake detection!
[Apr, 2024] Two Oral presentations at ICASSP 2024 on deepfake detection and speech health diagnostics.
[Apr, 2024] Paper on Speech anonymization for health monitoring is published at IEEE-TIFS!
[Jan, 2024] Interning as an Applied Scientist on audio deepfake detection at Reality Defender (Finished in August).

Selected Publications (2024)

[Sept, 2024] SLIM: Style-Linguistics Mismatch Model for Generalized Audio Deepfake Detection, accepted at NeurIPS 2024
[Sept, 2024] WavRx: a Disease-Agnostic, Generalizable, and Privacy-Preserving Speech Health Diagnostic Model, published at IEEE Journal of Biomedical Health Informatics
[Aug. 2024] MSPB:a longitudinal multi-sensor dataset with phenotypic trait measurements from honeybees, published at Nature Scientific Data
[Jan, 2024] On the Impact of Voice Anonymization on Speech-Based Health Diagnostics, published at IEEE Transactions on Information Forensics and Security