Yasaman Ahmadiadli

Ph.D. Candidate · Electrical & Computer Engineering · Toronto Metropolitan University

Researcher in audio deepfake detection, adversarial robustness, and multimodal AI systems. My work combines signal processing, deep learning, and affective computing to build detection systems that generalize reliably across real-world conditions. Four years of hands-on teaching experience at the undergraduate and graduate level.

Email · LinkedIn · Google Scholar · ResearchGate · GitHub · ORCID

Research

My research is in audio deepfake detection and multimodal AI, with a focus on building systems that remain robust when deployed on audio signals outside the training distribution. I work at the intersection of deep learning, signal processing, and representation learning, developing frameworks that draw on acoustic, linguistic, and affective features to detect synthetic speech reliably across diverse conditions.

A newer thread in my dissertation explores the role of affective and emotional signals in audio as a complementary detection modality, alongside ongoing work on stress detection using physiological signals.

Research threads

Audio deepfake detection and adversarial robustness Developing deep learning systems for synthetic speech detection, with emphasis on cross-dataset generalization and robustness to unseen audio conditions. Combines bi-level optimization, contrastive representation learning, and acoustic feature engineering. Evaluated on ASVspoof2019, ADD2022, FoR, and In-the-Wild benchmarks.

Affective signal analysis for synthetic speech Exploring the role of affective and emotional signals in audio as a complementary modality for deepfake detection. Ongoing dissertation work.

Physiological signal analysis and stress detection Applied research on stress detection using physiological signals. Industry collaboration (WellWave).

Survey and taxonomy of speech deepfake detection Co-authored a comprehensive survey synthesizing over 100 detection methods from IEEE and ACM venues into a unified taxonomy covering acoustic features, neural architectures, training paradigms, and evaluation protocols. Published in ACM Computing Surveys (2025).

Publications

Journal articles

A survey on speech deepfake detection Menglu Li, Yasaman Ahmadiadli, Xiao-Ping Zhang ACM Computing Surveys, 2025

Low-distortion and adaptive image steganography by enhancing DBSCAN, Sobel operator, and XOR coding A. Rezaei, Yasaman Ahmadiadli, L. Farzinvash, M. Asadpour Journal of Information Security and Applications, vol. 70, 2022

Conference papers

Probing and mitigating identity leakage in audio deepfake detection using artifact-centric learning (under review) Yasaman Ahmadiadli, Naimul Khan, Xiao-Ping Zhang CCECE 2026

Robust deepfake audio detection via bi-level optimization Menglu Li, Yasaman Ahmadiadli, Xiao-Ping Zhang IEEE International Workshop on Multimedia Signal Processing (MMSP), 2023

A comparative study on physical and perceptual features for deepfake audio detection Menglu Li, Yasaman Ahmadiadli, Xiao-Ping Zhang 1st International Workshop on Deepfake Detection for Audio Multimedia, ACM MM, 2022

Preprints

Identity independent audio deepfake detection Yasaman Ahmadiadli, Naimul Khan, Xiao-Ping Zhang arXiv:2505.06766, 2025

Teaching

Instructor of record - Toronto Metropolitan University

Course Title
COE848 Fundamentals of Data Engineering - Designed and delivered all lectures, labs, and assessments. Topics: relational databases, SQL, NoSQL, distributed systems, Big Data architectures. 100+ students per semester.
COE428 Engineering Algorithms and Data Structures - Sole instructor for 40+ students. Algorithm design, complexity analysis, dynamic programming, graph algorithms.
COE691 Software Requirements and Specifications - Requirements elicitation, UML modelling, microservices architecture. 20+ students.
COE70A/B Capstone Design Project - Faculty supervisor for undergraduate engineering design teams through full project lifecycle: scoping, system design, implementation, and final presentation.

Graduate teaching assistant - ECE/CS courses

Course Title
ELE632 Signals and Systems II - DTFS, DTFT, DFT/FFT, Z-transform, LTI systems. 60+ students. Supervisor: Dr. Dimitri Androutsos.
COE328 Digital Systems - Digital logic, sequential circuits, hardware design. 100+ students. Supervisor: Dr. Nagi Mekhiel.
COE608 Computer Design and Architecture - Pipelining, parallelism, processor architecture.
COE628 Operating Systems - OS principles, scheduling, memory management.
COE818 Advanced Computer Architecture - Advanced processor design and parallelism.
COE692 Software Design - UML, microservices. Supervisor: Dr. Faezeh Ensan.
CPS310 Computer Organization II - Computer organization and architecture. Supervisor: Dr. Alireza Sadeghian.

Additional roles

   
AI4Good AI4Good Lab Teaching Assistant - Mentored student teams on ML applications including multimodal AI and LLM-powered project development.
CEN100 First-Year Engineering Advisor - Academic advising and technical guidance for 500+ first-year engineering students. Supervisor: Dr. Lamya Amleh.

Teaching assistant - University of Tabriz

Experience

WellWave - AI/ML Intern

2025 - present

Applied ML work on stress detection using physiological signals.

Ph.D. Researcher - Toronto Metropolitan University

2021 - present

Signal Processing and Communications Lab. Supervisors: Dr. Naimul Khan & Dr. Xiao-Ping Zhang.

Designing and evaluating deep learning systems for audio deepfake detection, with a focus on adversarial robustness and cross-dataset generalization. Published in ACM Computing Surveys and IEEE venues.

Awards & scholarships

Queen Elizabeth II Graduate Scholarship in Science and Technology (QEII-GSST) 2025-2026. Competitive provincial award from the Ontario government recognizing excellence in graduate research in science and technology, administered through Toronto Metropolitan University.

Student Access Guarantee (SAG) Bursary Winter 2026. Awarded by Toronto Metropolitan University in recognition of continued academic commitment and enrolment.

Education

Toronto Metropolitan University, Canada · 2021 - present Ph.D., Electrical and Computer Engineering Dissertation: Robust Audio Deepfake Detection via Artifact-Centric Learning and Emotional Signal Analysis Supervisors: Dr. Naimul Khan & Dr. Xiao-Ping (Steven) Zhang Award: Queen Elizabeth II Graduate Scholarship in Science and Technology (QEII-GSST), 2025-2026

University of Tabriz, Iran · 2017 - 2020 M.Sc., Computer Engineering - Artificial Intelligence and Robotics Thesis: Intelligent target hitting in path with obstacles under real environment circumstances using physical animation and evolutionary algorithms.

University of Tabriz, Iran · 2012 - 2017 B.Sc., Information Technology Engineering Final project: Pedestrian detection using thermal cameras for driving in dangerous road conditions.

Skills

ML / Deep learning - PyTorch, TensorFlow, Keras, Scikit-learn, CNNs, RNNs, Transformers, GANs, GNNs, Foundation models

Optimization & training - Bi-level optimization, adversarial training, contrastive learning, domain adaptation, transfer learning, hyperparameter tuning

Signal & audio processing - Librosa, SoundFile, DSP, Mel-spectrograms, CQCCs, FFT, Z-transform, time-frequency analysis, Wav2Vec2, Emotion2Vec

Physiological signals - HRV, ECG, gaze tracking, EEG, Movesense sensor integration, biometric pipeline engineering

Languages & tools - Python, Java, C, MATLAB, SQL, Git, CUDA, Jupyter, Weights & Biases, TensorBoard, Google Colab

Data engineering - Relational databases, NoSQL, distributed systems, Big Data pipelines, Streamlit

Software engineering - OOP, UML, requirements analysis, system design, microservices, testing

Hardware - VHDL, FPGA (Quartus, Logisim), digital logic design, ROS

Service

Peer review

Invited reviewer for journals across IEEE, Elsevier, and Springer, covering machine learning, cybersecurity, and signal processing.