A complete archive of our research findings, technical breakdowns, and tutorials.
DatasetASVspoofEvaluation
Audio Deepfake Dataset Review: ASVspoof 2021
A comprehensive review of the ASVspoof 2021 challenge dataset, focusing on logical access and physical access scenarios.
2024-02-289 min
WatermarkingProvenanceDefense
Watermarking Audio: A Defense Strategy
Implementing imperceptible watermarks in audio to prove authenticity and trace the origin of synthetic content.
2024-01-124 min
Adversarial MLRobustnessSecurity
Adversarial Attacks on Deepfake Detectors
Investigating how robust current deepfake detection models are against adversarial perturbations. Can we fool the detectors?
2023-12-057 min
Voice ConversionSecurityZero-Shot
The Rise of Zero-Shot Voice Conversion
Analyzing the latest advancements in zero-shot voice conversion models that require only a few seconds of reference audio. Implications for security and privacy.
2023-11-206 min
DetectionForensicsSpectral Analysis
Detecting AI Voice Clones: Spectral Analysis Techniques
Exploring the subtle spectral artifacts left behind by modern voice cloning tools like ElevenLabs and Tortoise TTS. Can we spot the synthetic signature?
A deep dive into WaveGrad, a conditional model for waveform generation through estimating gradients of the data density. How does it compare to WaveNet?