Vocal Tract Disparity and Potential Implications for Speaker Recognition
Perceptual experiments show that listeners recognize female speakers with lower accuracy than male speakers. Automatic speaker recognition systems may also show performance bias against female speakers even when training data sets are gender balanced. The underlying reasons for this discrepancy are unclear. Here, we apply geometric morphometrics to quantify sex-related morphological vocal tract…
Recent perceptual experiments have revealed that listeners demonstrate lower accuracy in recognizing female speakers compared to male speakers. This performance bias is not limited to human perception, as automatic speaker recognition systems may also exhibit a tendency to favor male voices even when training datasets are balanced across genders.
The reasons behind this discrepancy remain uncertain. In a new study, researchers utilized geometric morphometrics to quantify sex-related vocal tract disparities, specifically focusing on the extent of shape variation in both resting and articulatory configurations. Their findings indicate that male speakers display a greater degree of disparity in both configurations when compared to female speakers.
This morphological difference may contribute to the generation of more distinct acoustic signatures, potentially explaining why male voices are perceived with higher accuracy by both human listeners and machine-based recognition systems. The study's insights suggest that innate variations in vocal tract morphology could play a role in performance bias observed in voice technology and human speaker perception.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
