Appreciation of singing and speaking voices is highly idiosyncratic

Tell us about your study

Voice preferences deeply affect human interactions, but very little empirical research has explored what drives these preferences. In this study, we used a validated stimulus set to examine subjects’ enjoyment of contrasting types of vocalizations. To create this stimulus set, we recorded 22 highly trained female singers singing the same material in three singing styles (as a lullaby, as a pop song, and as an opera aria) and speaking the corresponding text out loud in two styles (as if speaking to an adult and as if speaking to an infant). We invited 62 lay listeners to our laboratories in the Max Planck Institute for Empirical Aesthetics, in Frankfurt, and asked them to rate how much they liked each performance on a scale of 1 (not at all) to 9 (a lot). This experimental design allowed us to quantify how much participants agree with each other in their preferences and to test whether some voices are consistently preferred, regardless of the type of vocalization.

How did the idea for the paper come about?

Our study was inspired by two main ideas. First, research in the visual domain has found that people tend to agree more when rating natural images like faces or landscapes compared to artificial images like paintings. This suggests that stimuli from natural categories might be perceived and appreciated more uniformly across individuals. We aimed to explore whether a similar pattern could be observed in the auditory domain, namely with vocalizations. In our case, we considered lullabies to be a more "natural" or universal style of singing than operatic singing, which we understood as a more "artificial" style – it requires a very specific technique and is enjoyed by a niche audience. We thus expected to find higher shared taste (or interrater agreement) for lullabies than for operatic singing, with pop in an intermediate position.

We also explored the idea that voice attractiveness might be influenced by sexual selection. In this view, an attractive voice might signal a person’s biological/reproductive fitness. Supporting this idea, one study found that people rated the attractiveness of a voice similarly whether it was singing or speaking, meaning that certain voices were consistently preferred whatever the type of performance. We found this idea intriguing but were skeptical that sexual selection would largely explain voice preferences. So, we examined whether the same voices would be consistently preferred across the five different styles of vocalization in our study.

What do the results suggest? 

Our results did not parallel the findings reported in the visual domain. We found that, across all styles, participants’ agreement in their preferences (or shared taste) was small. And even though there were differences in the amount of shared taste depending on the style of vocalization, these differences were not necessarily in the direction we expected: we found a higher level of shared taste for lullabies than for pop singing (as expected), but were surprised to find an equivalent amount of shared taste for lullaby and operatic singing. This may have been influenced by a high prevalence of opera fans in our sample. To clarify this pattern, future research should explore different participant groups and include a broader range of singing styles. 

Regarding the consistency of voice preferences across styles, we found only limited consistency in average singer preferences across the singing and speaking performances, suggesting that, if sexual selection does play a role in voice preferences, that role is rather limited.

Our results advance the understanding of the human voice as a particularly complex and multifaceted signal, and suggest that voice preferences are shaped by the function of the vocalization.

What are the future directions for your work?

One interesting direction concerns the development of voice preferences. Research shows that infants prefer listening to infant-directed rather than adult-directed speech. In our study, we observed that adults generally had a "shared dislike" for infant-directed performances. This raises the question of when (and how) this preference shift occurs as people age. In addition to observing listeners from an early age to older adults, examining diverse cultural groups would help us to understand the dynamic changes and cultural variations in voice preferences, to hopefully understand the roots of our voice preferences.

How was your experience publishing with Royal Society Open Science?

Very good. Our study was a Registered Report, which means that our research plan went through peer-review and revisions before we collected or analysed the data. We were thus really happy to learn that Royal Society Open Science supports this format and offers in-principle acceptance to reports approved by the Peer Community in Registered Reports platform (which was the case for our report). This ensured our work was published in a reputable journal, reaching a wide audience with high standards for open science.

About the authors

Camila Bruder is a doctoral researcher at the Max Planck Institute for Empirical Aesthetics in Frankfurt, Germany. With an interdisciplinary background in biology and music, her research interests include music cognition and aesthetic preferences, with a special interest in the perception and performance of singing.

Pauline Larrouy-Maestri is a senior researcher at the Max Planck Institute for Empirical Aesthetics in Frankfurt, Germany. Together with her team and collaborators, Pauline conducts research to understand how we categorize acoustic information and make sense of sounds, with a focus on vocalizations.

Read the full paper here: Appreciation of singing and speaking voices is highly idiosyncratic

Royal Society Open Science is an open access journal that welcomes the submission of all high-quality science. More information about the submission process can be found on our webpage.

Image: Created by Camila Bruder combining edited AI-generated visuals from https://dream.ai.