Objective: Operatic singers adjust their vowel production depending on pitch to maintain consistent timbre across their range. Existing studies have focused predominantly on modeling the resonance profile in singing, yet the change in tongue position itself—an important contributor to the resonance space—has been underexamined in relation to pitch variation. The goal of this study is to model how target pitch influences the articulatory vowel space.

Methods/design: 15 professional opera singers participated in a production study. They produced the vowels [æ], [_], [_], [i], [u] in CVC real word contexts. Each singer sang five randomized groups of vowels at every semitone along an ascending scale based on their pitch range. Midsagittal ultrasound tongue images were collected at 81.5 frames per second using an Articulate Instruments Micro ultrasound system with a headset-stabilized transducer (20mm, 2–4MHz). Articulatory data were tracked using DeepLabCut and modeled with functional data analysis (FDA).

Results: Separate FDA models were run for the x and the y coordinates of the tongue splines respectively to maximize the variability captured. Preliminary results with 8 participants showed a general pattern of neutralization at high pitches relative to the singer: lower vowels tend to be raised as the pitch increases, and higher vowels tend to be lowered.

Conclusion: We conclude that the effect of pitch target on vowel modification is potentially modulated by gender, and the direction of vowel modification is vowel specific. Further analysis is underway to investigate whether vowel modification patterns by voice type specifically. These results hold implications in our understanding of vowel modification in singing and vocal pedagogy.