Perceiving Gender in the Voice: Accuracy, Confidence, and Reasoning in Listener Attributions

This study examines how listeners perceive and interpret gender identity from the voice, focusing on 18 speakers across six gender identity groups, including cisgender, transgender, and non-binary. It aims to identify patterns of perceptual accuracy, confidence, and phonetic cue use, and to explore how these reflect broader sociophonetic processes of gender interpretation. Sixty-one listeners categorized the gender identity of speakers of short read-speech excerpts, rated each speaker's masculinity, femininity, and their own confidence in their judgments, and indicated which phonetic characteristics they used to inform their decisions. Accuracy, confidence, and phonetic cue selection were analyzed alongside speaker gender identity and phonetic feature associations. Listeners most accurately attributed the correct gender identity to cisgender speakers, reflecting reliance on normative gendered vocal expectations. Gender-diverse speakers were recognized with lower accuracy, though lenient scoring revealed that listeners often perceived degrees of gender non-conformity even when they failed to select the speaker's specific gender identity. Confidence ratings followed a graded pattern that did not consistently align with accuracy, suggesting that subjective certainty is shaped more by perceived vocal typicality than by reliable gender identity attribution. Masculinity and femininity ratings largely followed expected patterns, with cisgender speakers rated toward the masculine and feminine poles, while non-binary and transgender speakers tended to occupy the middle of the spectrum, highlighting the perceptual complexity of gender expression among speakers whose voices may be less likely to conform to binary gender expressions. Phonetic feature associations emphasized the central role of pitch and articulation. Listener demographics and social attitudes did not significantly influence performance, indicating that perceptual difficulty arises primarily from the complex interaction between vocal cues and socially constructed gender norms. These findings demonstrate that gender is not simply "heard" in the voice but actively interpreted through cultural expectations, with implications for phonetic theory, forensic practice, speech technology, and gender-affirming voice work.