Perceived vibrato and the singing power ratio explain overall evaluations in opera singing

In opera singing competitions, judges use an overall score to evaluate the singers' voices and determine their rankings. This score not only guides the singers' technique and expressiveness but also serves as a crucial indicator that can significantly influence their careers. However, the specific elements captured by this overall score remain unclear. To address this gap, the present study analyzed opera singing recordings to identify the factors that explain the overall score. Ten trained female Japanese singers performed "<i>Caro mio ben</i>" under standardized recording conditions. Four experts evaluated the recorded performances by assigning an overall score of 100 points and rating six vocal attributes: vibrato, resonance, timbre, diction, intonation, and expressiveness. The recordings were then analyzed to calculate specific acoustic and audio features, including the singing power ratio (SPR), harmonic-to-noise ratio (HNR), and loudness units full scale (LUFS). We developed two linear mixed models: the first regressed the overall score on the subjective vocal attributes, whereas the second predicted the overall score from the acoustic features. Evaluator identity was included as a random effect in both models. The results showed that vibrato was a significant predictor of the overall score in the first model. In the second model, only SPR emerged as a significant predictor. These findings suggest that vibrato, which reflects emotional expressiveness and vocal control, and SPR, which indicates the relative power in the high-frequency band (2-4 kHz) and assists a voice clearer than the accompaniment, are key factors in explaining the overall score in opera singing.