Researchers statistically investigate key factors that contribute to an opera singer’s overall score in singing competitions.

Singers appearing in opera singing competitions are typically ranked based on an overall score assigned by the judges. However, it has remained unclear exactly what elements in the singer’s voice contribute to these scores. In a new study, researchers analyzed opera singing recordings to identify the impact of various vocal characteristics and acoustic features on the scores assigned by the judges. Their findings could provide a scientific basis for future vocal training and education.

Opera singers are admired for their powerful and expressive voices, achieved through years of rigorous training. In opera singing competitions, judges evaluate the performances of aspiring singers by assigning an overall score to the singer’s voice that decides their ranking. These scores, however, determine not just the outcome of a single event but the overall trajectory of a singer’s career. Yet, the criteria underlying these scores have remained a mystery. More specifically, it is unclear exactly what specific aspects of a singer’s voice influence their score.

Previous studies have suggested that certain qualities—such as vibrato, resonance, timbre, and diction (the style of enunciation)—play a role. In addition, acoustic features play an important role in such evaluations, especially when based on audio recordings. One key acoustic feature in this regard is the singer’s formant, or the ability of their voice to fill a large hall without amplification. This can be quantified using the singing power ratio (SPR), which measures the energy balance of a voice in a specific range of frequencies. Other indicators include the harmonic-to-noise ratio (HNR), which represents the clarity of a voice, and the integrated loudness units full scale (LUFS), a standardized metric used for normalizing audio. But, which of these factors matter most when judges assign scores?

A new study has now found a clear answer to this question. Led by Ms. Haruka Kondo, a PhD student at the Graduate School of Media and Governance at Keio University, Japan, researchers conducted a statistical analysis of opera performance recordings to identify key factors contributing to the singers’ scores in singing competitions. “For singers, a competition score can define their career, but what it is the judges are listening for has remained subjective and unclear,” explains Ms. Kondo, herself a soprano opera singer. “In our study, we scientifically decoded how different vocal and acoustic factors influence judgment scores in opera singing competitions.”

The research team was headed by Associate Professor Shinya Fujii and included Dr. Sotaro Kondoh, a JSPS Research Fellow, from the Faculty of Environment and Information Studies at Keio University. The study was published in Volume 16 of the journal Frontiers in Psychology on August 08, 2025.

Ten trained Japanese female opera singers participated in the study, with each performing the classical Italian song “Caro mio ben” under standard recording conditions. These recordings were then rated by four vocal instructors, who were also professional singers. They assigned both an overall score out of 100 as well as scores on six vocal attributes: vibrato, resonance, timbre, diction, intonation, and expressiveness. Each singer was evaluated by all the judges in random order. Additionally, the researchers analyzed the acoustic features of the recordings using three parameters, namely SPR, HNR, and LUFS.

To understand the contribution made by the vocal attributes and the acoustic features on the overall score, the team then developed two models. One of them examined the impact of the six vocal attributes, while the other investigated the contribution from the acoustic features.

Remarkably, the results suggested two main predictors of scores: vibrato and SPR. Vibrato, known to have a strong correlation with overall scores from previous studies, stood out as the most significant vocal attribute, with the impact of the other attributes being statistically insignificant. The significance of SPR among the acoustic features was consistent with its known role as an indicator of the singer’s formant. In contrast, HNR and LUFS did not significantly impact the score. The low impact of HNR was explained by the fact that trained singers already exhibit high clarity and low irregularity in their voice. Similarly, loudness alone being a weak contributor toward subjective evaluations explained the low impact of LUFS.

“These findings provide a clearer picture of how judges evaluate performances,” notes Dr. Fujii. “Our research offers objective insights and powerful new tools for vocal training and education. Beyond that, it also addresses foundational questions of how humans perceive sound and art.”

“Imagine a voice lesson where instead of relying solely on the teacher’s perception, students can reliably track specific characteristics of their voice on a screen,” remarks Dr. Kondoh.

Overall, the study could provide a firm scientific foundation for understanding what makes an opera singer sound better and possibly bridge the gap between the art of opera singing and the science of acoustics.