Project DetailsPhotoplethysmogram (PPG) signals are a compelling biometric modality due to their physiological uniqueness and ubiquity in wearable technology. This study introduces a novel cross-domain framework for PPG-based biometric authentication by adapting architectures originally developed for automated speaker verification. Recognizing that both speech and PPG signals are quasi-periodic and shaped by individual physiological traits, we demonstrate that speaker verification models are exceptionally well-suited to capturing the subtle, dynamic temporal patterns inherent in cardiac cycles. Our methodology involves a preprocessing pipeline that aligns raw PPG data with the structural characteristics of acoustic signals, enabling the direct application of speaker verification models. Notably, this approach yields a highly generalizable framework that maintains robust performance across different physiological conditions and hardware configurations. Experimental results across multiple datasets demonstrate state-of-the-art performance, outperforming previously reported results in the literature.
