Introduction
Vibrato, an essential characteristic of vocal expression, has historically been studied within controlled laboratory settings, limiting research to small samples and often excluding elite performers. The advent of AI-powered voice separation now enables large-scale analysis of commercial recordings, providing unprecedented access to vibrato characteristics across a diverse range of renowned singers, both living and deceased. This study leverages these technological advancements to examine vibrato features in professional recordings, bridging the gap between pedagogical insight and real-world vocal practice.

Methods
This study utilized 150 commercial recordings of G. F. Händel’s aria “Ombra mai fu,” sung by world-class, international, and national singers of all voice classes. The orchestral accompaniment was removed using the “music.ai” and “Izotope RX10” vocal separation methods, an approach systematically verified in a recently presented study. Within each recording, six sustained notes were chosen for subsequent analysis. Vibrato rate and extent were computed using both the sine wave fit method (Nestorova et al., 2023) and the marked peaks method applied to the fundamental frequency (f_) contours. All results underwent visual inspection through spectrogram analysis and a goodness-of-fit check to verify accuracy.

Results
Our analysis revealed clear trends in vibrato characteristics across voice types and professional categories. Higher voices exhibited faster vibrato rates, with Altos reaching a median of 6.0 Hz, while Baritones and Basses displayed slower rates around 5.4 Hz. Lower voices tended to have broader vibrato extents, whereas Boy Sopranos demonstrated the narrowest extents. In terms of professional taxonomy, “Superstar” and “International” singers stood out with higher vibrato rates and more controlled, narrower extents, reflecting advanced vocal technique. By contrast, Local Community and Child singers exhibited broader extents and lower vibrato rates, suggesting less refined control.

Significant associations between professional taxonomy and vibrato characteristics was achieved. Singers in the “Superstar” and “International” categories consistently displayed higher vibrato rates and narrower extents, indicative of a sophisticated vocal control developed through advanced training and experience. In comparison, broader extents and lower rates among Local Community and Child singers highlight distinctions in vocal maturity and technical development. These findings emphasize that professional level is a strong predictor of vibrato control and consistency in performance.

Conclusion
This study highlights the potential of AI-powered voice separation for large-scale vibrato analysis in classical singing. By analyzing commercial recordings, we identified distinct vibrato patterns linked to voice type and professional level, with advanced singers exhibiting refined vibrato control. These findings validate the use of commercial recordings in vibrato research and provide a foundation for future studies to expand our understanding of vibrato’s role in vocal artistry across genres.