Md Shajalal Receives Best Paper Award
At the 28th International Conference on Human-Computer Interaction (HCII) in Montreal, an award was presented to researchers from the University of Siegen. The award-winning paper, titled “Human-Centered Explanations for Audio Deepfakes: Making Machine Reasoning Human-Perceptible Through Voice Traits” was authored by Md Shajalal, Mahedi Hasan Riday, Sima Amirkhani, and Gunnar Stevens. The paper received a Best Paper Award in the “AI in HCI” category.
The award-winning paper was produced as part of the “AntiScam” research project funded by the Federal Ministry of Education and Research (BMBF). Among other things, the project investigates explainable AI-based methods for detecting fraud attempts using synthetic voices and aims to better protect consumers from AI-enabled communication fraud.
The research addresses a growing challenge arising from the rapid development of generative artificial intelligence: increasingly realistic audio deepfakes are difficult for humans to distinguish from authentic recordings. While AI-based detection systems can help identify manipulated audio recordings, their decisions are often difficult for users to understand.
The researchers are therefore investigating how the reasoning behind deepfake detection systems can be translated into voice characteristics that are perceptible to humans. Instead of simply presenting users with a prediction about whether an audio recording is real or fake, the approach aims to provide understandable clues as to why the system classifies a recording as manipulated. This human-centered perspective can help users better interpret AI-generated explanations.
