The Information Systems, Security, and Forensics (ISSF) laboratory is founded on a vision of transferring mathematical and statistical tools employed in signal processing to analyze large-scale data to solve challenging problems in the areas of deepfakes, spoof detection for ID verification systems, automotive cybersecurity, sensor data integrity verification, and multimedia forensics. The ISSF lab provides the resources required to train a workforce of professionals in the areas of deepfakes, multimedia authentication, automotive cybersecurity, cyber-physical system security, sensor security, information security, forensic science, and multimedia processing. This lab is equipped with several desktop computers and laptops, an ETAS LABCAR (HIL System), a deep learning server, a variety of smartphones, sensors, a Ford Focus vehicle, multiple testbeds for in-vehicle network communication, sound recording equipment, a recording studio with controlled reverberations, microphone calibration equipment, digital video cameras, and a repository of software for acquiring in-vehicle network (IVN) traffic, tools to analyze large scale data and manipulating digital media.

Research Areas

  • Deepfakes & Multimedia Authentication
  • Spoof Detection for ID Verification Systems
  • Automotive Cybersecurity
  • Sensor Data Integrity Verification
  • Cyber-Physical System Security
  • Multimedia Forensics

Lab Resources

  • High-performance Deep Learning Server
  • ETAS LABCAR HIL System for automotive testing
  • Ford Focus vehicle with in-vehicle network testbeds
  • Controlled recording studio with microphone calibration
  • Digital video cameras & sound equipment
  • Repository of software  & media analysis

Open-Source Contributions in Audio Forensics & Cybersecurity

SSL Anti-Spoofing

  • Implements a self-supervised learning pipeline for detecting spoofed audio.
  • Frontend Extractors: wavlm_large, mae_ast_frame, npc_960hr
  • Backends: AASIST, SLS, XLSR-Mamba
  • Switch models via config or CLI with simple training/evaluation commands.

Audio Deepfake Detection

  • Generates and analyzes codec-based deepfake audio samples.
  • Codecs: AudioDec, HiFiGAN, Encodec, Descript Audio Codec, SpeechTokenizer, Funcodec.
  • Designed to test robustness of detection systems.

ASD-SUPERB

  • Audio Spoof Detection downstream task built on S3PRL/SUPERB.
  • Supports training with ASVspoof2019 LA data and SSL models like WavLM.
  • Provides full training/evaluation scripts.