Retrospective longitudinal analysis of spectral features reveals divergent vocal development patterns for treble and non-treble singers

This study investigates the longitudinal spectral development of male and female classical singers throughout conservatory training. While classical singing techniques share commonalities across voice types, physiological differences have led to gender-specific pedagogy. Previous acoustic research has explored differences in resonance strategies between genders and voice types; however, little is known about how these spectral characteristics develop during vocal training. In this retrospective longitudinal study, recordings from 117 classical voice students at the Hochschule für Musik Carl Maria von Weber Dresden were analyzed. Recordings spanned 2008-2018 during the students' 4-year bachelor studies. Countertenors were analyzed with sopranos, mezzo-sopranos, and altos as "treble" voices and tenor, baritone, and bass voices were analyzed as "non-treble" voices. Spectral measures were assessed from three different vocal exercises using long-term average spectrum. Statistical analysis utilized linear mixed-effect models to explore the effect of years of study, voice group (treble or non-treble), and their interaction. Findings reveal that the treble singers increasingly concentrated relative acoustic energy in the f0 range of the sung exercise, while the non-treble singers increasingly concentrated relative acoustic energy above 1000 Hz. Additionally, female singers exhibited increased vocal periodicity over time across all tasks, suggesting a reduction in breathiness.