Analyzing the acoustic parameters of bird vocalizations and their perceived pleasantness by humans

1 Introduction

Due to their diversity and proximity to humans, bird vocalizations are studied in a variety of contexts, including intra- and interspecific communication among birds and mammals (Randler, 2006; Magrath et al., 2007; Crouch and Mason‐Gamer, 2019; Kalb et al., 2019; Rose et al., 2019). Bird vocalizations can consist of few, repeating elements or different element types often following a specific sequence (Kershenbaum et al., 2016; Kumar, 2003) and the size of the vocal repertoire varies both between and within species (Kipper et al., 2004). Generally, vocalizations are produced by an air flow from the lungs to the bronchi and trachea which causes the membranes of the syrinx as well as the surrounding air to vibrate and thus produce sounds (Rüppell, 1933). Due to the different arrangements of the membranes in the syrinx and the contraction of various muscles that regulate the tension of the membranes, sounds of a wide variety of timbres, frequencies and amplitudes can be produced (Rüppell, 1933; Kumar, 2003). Consequently, birds produce some of the most complex sounds in nature (Catchpole and Slater, 2003) and form an important part of the acoustic landscape.

Although bird vocalizations did not evolve for human pleasure or interaction with humans (with a few exceptions, parrots and honeyguides (Pepperberg, 2006; Spottiswoode et al., 2016) there is an increasing line of research studying the emotions, especially well-being and (perceived) restoration, that bird songs can elicit in humans.

Psychological restoration is defined according to the Stress Reduction Theory (Ulrich, 2023) and the Attention Restoration Theory (Kaplan and Kaplan, 1989) as the process of regaining depleted mental and emotional resources. The Perceived Restorativeness Scale assesses environmental features that support psychological restoration (Korpela and Hartig 1996; Berto 2005) including being away, fascination and compatibility. More recently, this scale also incorporates the perceived restorativeness of sound, indicating that acoustic characteristics play a crucial role in restorative experiences (Payne, 2013; Fisher et al., 2021; Ratcliffe, 2021).

Natural sounds are known to have a high positive restorative potential (Björk, 1985; Grahn and Stigsdotter, 2010; Payne, 2013), and natural soundscapes are often reported to have a higher psychological and mental restoration than sounds from built environments (Alvarsson et al., 2010; Benfield et al., 2014; Medvedev et al., 2015). Many studies identified bird sounds as an important source of the acoustic landscape associated with higher reported restoration and pleasantness (Ratcliffe et al., 2013; Bakolis et al., 2018; Ratcliffe et al., 2020; Nwankwo et al., 2022; Liu et al., 2024).

In psychoacoustics and soundscape research, pleasantness is a purely subjective measure of human perception that can be affected both by acoustic properties of the sound and personal characteristics of the listener (Uebel et al., 2022; Erfanian et al., 2021). Various studies showed bird sounds to differ in their reported pleasantness and restoration depending personal associations with environments and memories (Björk, 1985; Ratcliffe et al., 2013; Cox and Gaston, 2015; Ratcliffe et al., 2016).

Generally, sounds with low frequencies are perceived as less pleasant (Björk, 1985; Halpern et al., 1986; Kumar et al., 2008). Consequently, we hypothesize for our study that bird sounds with higher low (minimum) frequencies are rated as more pleasant than those with lower low frequencies.

According to Kumar et al. (2008), sound with frequencies between 2.4 and 5.5 kHz (mid-range to higher frequencies) are perceived as unpleasant as the human auditory system is most sensitive to this range and sounds in this range may be perceived as louder and more intense (Fletcher and Munson, 1933). Therefore, we hypothesizes that bird sounds with lower high and maximum frequencies are perceived as more pleasant in our study.

Sounds with higher amplitudes carry more energy and are perceived as louder (Benson, 2006) and participants perceived louder bird sounds as less restorative (Ratcliffe et al., 2020). Consequently, amplitude may affect how pleasant humans perceive bird sounds and we predict bird vocalizations with lower amplitudes to be more pleasant.

Lastly, participants like sounds with higher perceived sound complexity more and rate them to be more restorative than less complex sounds (Ratcliffe et al., 2020; North and Hargreaves, 1995). Here, complexity is usually measured by asking listeners to subjectively rate the complexity of a sound on a scale from low to high (North and Hargreaves, 1995; Marin and Leder, 2013; Sallavanti et al., 2016). In our study, we investigated one dimension of sound complexity by counting the number of different elements within a vocalization. We hypothesize that sounds with a higher number of different elements, i.e. more complex sounds, are perceived as more pleasant.

While most studies focus on subjective criteria to determine the pleasantness of sounds, only few studies shed light on single objective acoustic measures that might affect human perception of sounds. To our best knowledge, our study is the first to examine the relationship between multiple acoustic properties of bird vocalizations (frequency, amplitude, bandwidth and sound complexity) and human perception, rather than focusing on a single acoustic feature (commonly fundamental frequency). This approach allows us to investigate how different parameters might interact and jointly shape human perception of bird sounds, acknowledging that perceived pleasantness is likely affected by a combination of multiple acoustic properties rather than by a single acoustic feature.

2 Methods

2.1 Human perception of bird sounds

This study has been granted ethical permission by the ethics committee of the Faculty of Social Sciences and Economics of the University of Tübingen (AZ.: A2.5.4-402_hb). Participants were recruited through an email via the university mailing list. Additionally, the email invited the university members to share the study information potentially interested family members and friends to broaden the outreach of the study. All participants received a prior information sheet about the study and gave their written consent to participate before starting the survey. Data on vocalization pleasantness was collected from November 2022 to January 2023 (N = 37) and May 2025 (N = 47). Participants were invited into the lab where the sounds were presented in a controlled environment. We used vocalizations of 123 bird species (see Supplementary Material for all species used during this study) which were determined to be wild birds according to Barthel et al. (2018) and are widely distributed in Germany. Audios were taken with written permission from the DVD “Die Stimmen der Vögel Europas” by Bergmann et al. (2020). We selected vocalizations of adult birds. Other instrumental bird sounds such as flapping wings or drumming were not used. 15 seconds audio files were played in the survey. The number of vocalizations within the 15 s sound file varied among species but reflected the natural calling patterns and typical call rates of species.

For each testing session, four to six participants were tested simultaneously. Participants were seated evenly spaced throughout a quite seminar room to prevent communication among participants as well as any interference between headphones. Each participant was provided with a tablet (Samsung Galaxy Tab 6, 10.1 inches) on which the SoSciSurvey questionnaire had been pre-loaded. After a brief standardized introduction and explanation of the equipment, participants were asked to put on the headphones (Philips TAPH805) and start the survey. The volume of all devices had been pre-set to a fixed level before the start of the session. Participants then completed the survey at their own pace. Audio files were presented one at a time in a randomized order without any information on the bird species. Participants were able to play the audio more than once without a time limit. Survey participants were asked to rate sound pleasantness on a scale from one (not very pleasant) to five (very pleasant).

Additionally, participants were asked to provide information on their age and gender (male, female, diverse). Furthermore, we asked participants about their birding specialization, i.e. how many bird species they can identify by sight and acoustically as well as how they self-report their birding skills (Randler et al., 2023) These three items were then summarized to variable “Birding specialization” (Cronbach’s alpha = 0.857). Further, we asked three questions about how much participants like and value birds (perception of birds) on a scale from 1 (do not agree) to 5 (fully agree). These three items were summarized to the variable “Perception of birds” (Cronbach’s alpha = 0.724). Table 1 gives an overview of the constructs and items used in the study.

Table 1. Overview of constructs and items used in this study.

2.2 Acoustic characteristics of bird sounds

We used Raven Pro for all acoustic analysis. Each recording was inspected in the spectrogram view (Hann window, sample size: 512, bandwidth: 124 Hz). To enhance visibility of single elements, we adjusted the brightness and contrast individually, whereby brightness was varied between 50 to 85 and contrast from 60 to 80. Each element within the 15 s recording was manually marked, whereby an element was defined as visible trace without temporal gap in the spectrogram (Kershenbaum et al., 2016). For each element, we used the spectrogram to measure four frequency-domain acoustic properties: low frequency, high frequency, peak frequency and 90% bandwidth. Frequencies are measured in kHz and are defined as the number of cyclical soundwaves per second, whereby a higher frequency causes a higher perceived sound (Rüppell, 1933; Catchpole and Slater, 2003). The low frequency (kHz) describes the frequency which contains the lower 5% of an element’s total energy. The high frequency (kHz) is the frequency below which 95% of an element’s total energy is contained (Charif et al., 2010). Hence, the low and high frequency measurements mark the lower and upper bound of an element’s energy distribution. The 90% bandwidth (kHz) describes the difference between the low and high frequency, representing the frequency range that contains 90% of an element’s energy (Charif et al., 2010). The peak frequency (kHz) describes the frequency at which the element has its maximum spectral power.

Further, we used the waveform in Raven to measure three time-domain acoustic parameters: minimum amplitude, maximum amplitude and peak amplitude. The minimum amplitude (kU) refers to the lowest amplitude in the waveform of an element. The maximum amplitude (kU) is the highest value in the waveform of each element. The peak amplitude (kU) describes the largest absolute value between the maximum and the minimum amplitude (Charif et al., 2010), representing the greatest deviation from zero.

Lastly, we assessed one dimension of complexity of the sounds. In human perception research, complexity is often subjectively rated by participants (Ratcliffe et al., 2020). To have a more objective measure of complexity, we counted the number of element types in the 15 s audio files. A low number of different element types thereby reflects are more monotone sound whereas a higher number is a more complex sound. Elements with similar acoustic structures were categorized as single element types and classified based on visual similarity in the spectrogram (Bloomfield et al., 2005; Zhang et al., 2022). Lastly, we counted the total number of elements, irrespective of element type, within the 15 s sound file to assess the frequency of vocalizations within the sounds.

2.3 Statistics

All analyses were performed using SPSS version 29.

Because all participants rated multiple bird vocalizations, our data has a repeated measured structure at the rating level. As our research question is the relationship between acoustic properties and perceived pleasantness, our main analyses were conducted at the bird species level. We therefore averaged pleasantness across participants for each species to avoid pseudo-replication in the species level analyses. We additionally averaged pleasantness ratings per participant to obtain a single mean pleasantness score per individual to ensure independence of observations at the participant level. This approach represents a simplification of the repeated measure structure in our data.

The mean pleasantness score per participant was used in a univariate general model to analyze the effects of the fixed factors age, gender, birding specialization and perception of birds on pleasantness. Moreover, we included the year of the survey (2023 or 2025) as a fixed factor.

Second, we performed a Principal Component Analyses (PCA) with a varimax rotation to analyze the acoustic features of the sounds used in the study. The PCA included the acoustic parameters minimum amplitude, maximum amplitude and peak amplitude, 90% bandwidth, low frequency, high frequency and peak frequency.

To determine the number of factors to be extracted, we compared the Eigenvalues of the PCA with randomly calculated Eigenvalues during a parallel analysis (Horn, 1965; Patil et al., 2008) and retained those factors with a greater eigenvalue in the PCA than in the random dataset. Afterwards we saved the factor scores derived from the PCA for further analysis. We performed a multiple regression model to evaluate the contribution of each PCA component on pleasantness while accounting for shared variance. Here, we used the mean pleasantness rating per species across all participants, as response variable and the three factor scores derived from the PCA as independent variables. Lastly, we conducted a multiple regression model to determine if the total number of elements within 15 s and sound complexity (i.e. number of different elements within 15 s influence pleasantness. Whereas PCA scores are derived from the acoustic parameters of all individual elements, both complexity and element number are single summary values obtained from counting the elements within the same 15 s sound file. Hence, we analyzed the PCA scores and complexity measures separately to avoid mixing predictors that differ in their level of aggregation and are not directly comparable. However, to assess whether the effect of sound complexity was independent of the acoustic characteristics represented by the PCA components, we conducted an additional multiple regression including the three PCA scores, total number of elements, and sound complexity as simultaneous predictors and sound pleasantness as independent variable. The multicollinearity diagnostics of the model indicated low collinearity among predictors (all VIFs < 1.3).

Finally, we extracted the Perceived Restorative Potential (PRP) ratings reported by Ratcliffe et al. (2016) for the 21 bird species that were included in both studies and correlated these values with the pleasantness ratings obtained in the present study.

3 Results

In total 84 participants (22 male, 61 female, 1 diverse) completed the survey. Age ranged from 19 to 81 (Mean age 33.88 ± 15.22). Mean pleasantness score across all species was 3.13 ± 0.47. The highest rating of 4.52 ± 0.84 obtained the common blackbird (Turdus merula), whereas the barn owl (Tyto alba) received the lowest rating of 1.32 ± 0.95 (see Supplementary Material for all pleasantness scores).

3.1 Demographic aspects

The GLM revealed a significant effect of perception of birds on pleasantness (F = 2.66, df=1,7, p=0.028), while birding specialization (F = 1.46,df=1,10, p=0.201), age (F = 1.14, df=1,29, p=0.362), gender (F = 0,012, df=1,1, p=0.912) and year (F = 0.90, df=1,1, p=0.350) had no significant effect. Pleasantness correlated positively with perception of birds (rs=0.42, n=84, p<0.001). That is, people who value and like birds more also rated bird sounds as more pleasant.

3.2 Acoustic parameters

The PCA and parallel analysis suggested the extraction of three components (Table 2). The variables minimum, maximum, and peak amplitude load onto the first component (Table 3), which is henceforth called “Amplitude Component”. Low frequency and peak frequency load onto the second component (Table 3), hereafter referred to as the “Frequency Component”. Bandwidth and high frequency load onto the third component (Table 3), which we labeled the “Bandwidth/High- Frequency Component”.

Table 2. Comparison of eigenvalues extracted from PCA and parallel analysis.

Factors with a higher eigenvalue in the PCA than the parallel analysis should be retained for further analysis.

Table 3. Results of the PCA depicting the three extracted components.

1: amplitude component, 2: frequency component and 3: bandwidth/high-frequency component.Bold values represent strong positive or negative factor loadings for each principal component.

The multiple regression model revealed all components as significant predictors of pleasantness (Table 4). Hence, sounds with higher low and peak frequencies, but lower amplitudes as well as narrower bandwidth and high frequency were perceived as more pleasant (Figure 1).

Table 4. Results of the multiple linear regression model with pleasantness as dependent and the PCA factor scores 1–3 as independent variables.

Regression Model: N = 123, R²=0.468, R²adjusted=0.455, F_(3,119)=34.912, p<0.001.

Concerning the number of elements and sound complexity, the regression model revealed sound complexity but not element number as significant predictor of pleasantness (Table 5). Vocalizations with a higher number of different element types, i.e. more variable vocalizations were rated more pleasant (Figure 1).

Table 5. Results of the multiple linear regression model with pleasantness as dependent and the element number and sound complexity as independent variables.

Regression Model: N = 123, R²=0.084, R²adjusted=0.069, F_(2,120)=5.507, p=0.005.

The pattern of results remained unchanged in the multiple regression including all PCA components, sound complexity and total number of elements. All predictors except total number of elements were significant (all PCA components: p<0.001, element number: p=0.506, sound complexity: p=0.037), suggesting that the effect of sound complexity is independent of the acoustic properties summarized by the PCA components.

3.3 Sound pleasantness and PRP

The Perceived Restorative Potential by Ratcliffe et al. (2016) correlated positively with song rating (Spearman: rho=0.542, n=21, p=0.011).

4 Discussion

People perceive some sounds as pleasant whereas other sounds are rated as unpleasant experience. This might be explained by associations between a sound and previously experienced events or by the acoustic properties of a sound (Kumar et al., 2008). Our study aimed to investigate if the acoustic properties of bird vocalizations are related to their perceived pleasantness of human listeners.

Participants in our study rated vocalizations as more pleasant when mean low and peak frequencies were higher and bandwidth was narrower. This is in line with our hypotheses and previous studies showing sounds with higher frequencies to be rated more pleasant and lower frequency bands being perceived as unpleasant (Björk, 1985; Halpern et al., 1986; Kumar et al., 2008). Additionally, very low frequencies of 20–200 Hz seem to be more difficult to ignore or habituate to than other noises and are rated as more annoying, unpleasant (Benton, 1997) or even aggressive (Tsai et al., 2010). Kumar et al. (2008) argued that frequencies between 2.4–5.5 kHz are perceived as unpleasant as the absolute threshold of our auditory system is lowest in that range, leading to the greatest perceived loudness of a sound within that range (Fletcher and Munson, 1933). Therefore, sounds may be perceived as unpleasant if their frequency lays in a range to which the auditory system is maximally sensitive. Our results are in line with these assumptions, as the mean low and high frequency as well as the bandwidth of the sounds within the highest pleasantness scores are outside this unpleasant frequency range (2.4-5.5 kHz).

Our results are further in line with findings by Ratcliffe et al. (2020), showing that the perceived restorative potential (PRP) of listeners increases with an increasing fundamental frequency, i.e. with an increase in the lowest frequency of a sound. The pleasantness ratings in our study were positively correlated with the PRP scores reported by Ratcliffe et al. (2016), indicating that perceived pleasantness and perceived restorative potential are possibly associated constructs. Nonetheless, further research assessing sound pleasantness and perceived restorative potential simultaneously is necessary to determine the relationship between these constructs and to investigate whether perceived pleasantness translates into restorative benefits.

Further, the perceived restoration of bird sounds is linked to personal memories and associations with certain environments. Sounds of songbirds such as great tit (Parus major), blue tit (Cyanistes caeruleus), greenfinch (Chloris chloris), and blackbird (Turdus merula) are associated with green, natural environments and receive higher perceived restoration scores (Ratcliffe et al., 2016). Similarly, these species received a higher rating in our study and had rather high low and peak frequencies.

Lastly, as hypothesized, songs with lower amplitudes were rated more pleasant. Songs with the highest scores had higher minimum and lower maximum and peak amplitudes, i.e. sounds of moderate loudness. The analyzed recordings originate from a commercial DVD (Bergmann et al., 2020), which compiles field recordings collected over many years and recordings most likely did not follow a standardized recording protocol. Consequently, amplitude differences between bird vocalizations may reflect not only characteristics of the vocalizations but also variation in recording conditions. Therefore, the amplitude related results of our study should be interpreted with caution. Ratcliffe et al. (2020) also found sound level to be a significant predictor of RRP and participants considered loud bird sounds to be not restorative while listening (Ratcliffe et al., 2013). Moreover, a relationship between sound level and arousal in response to natural sounds has been reported (Björk, 1985; Tsai et al., 2010). Loud animal noises might be associated with aggressive behavior (Tsai et al., 2010) and consequently rated less pleasant and restorative.

Acoustic complexity is generally considered a multidimensional construct involving temporal, spectral, spatial and perceptual characteristics of sounds (Pijanowski et al., 2011). In studies concerning the composition of bird songs complexity usually combines measures of repertoire size, frequency, timing and structure (Garamszegi et al., 2005; Najar and Benedict, 2019). In the context of aesthetics, complexity is often a subjective measure and rated by participants or experts. Ratcliffe et al. (2020) asked participants to subjectively rate song complexity on a scale from 1 (very simple) to 5 (very complex). More complex bird sounds received higher scores of reported recovery potential and were perceived as more pleasant than simple sounds. Delplanque et al. (2019) objectively measured sound complexity and found participants to prefer stimuli with intermediate complexity over high and low complexity ones. We used the number of different element types within a sound as proxy for sound complexity and assumed that more variable sounds would be rated more pleasant. Indeed, vocalizations with a higher number of different element types, i.e. more variable sounds, were rated more pleasant. We further found participants to rate sounds with low to medium bandwidth, i.e. moderate changes in frequency throughout the vocalization, higher than other ones, which might be related to the perceived complexity of a sound. Changes in moderate to high frequency ranges are known to be better perceived by humans (Benson, 2006) and therefore perceived as more variable or complex than sounds with low frequencies which may result in higher perceived pleasantness. Our method of measuring sound complexity by counting the number of different element types is a simplified approach and captures only one aspect of acoustic complexity. Future research may investigate further aspects of sound complexity such as redundancy, predictability and entropy (Delplanque et al., 2019) in bird sounds and its effects on human perception. Additionally, in our study we did not quantify the number of calls or songs contained in each 15 s sound file presented to the participants. Call rates differed among species, but the excerpts reflected natural vocal activity. Therefore, future studies might investigate a possible relationship between call rate and perceived pleasantness.

When looking at the individual level, age and gender as well as birding specialization had no significant effect on pleasantness ratings. In our study we had a modest sample size (n=84) and a surplus (67.6%) on female participants, which might limit the ability to statistically detect subtle variations on an individual level. It is known that there are evolutionary differences in the preferred frequency between males and females (Re et al., 2012), whereby women prefer men with lower voices, men choose women with higher voices (Re et al., 2012). Additionally, there seem to be small differences in musical perception (Bertolo et al., 2023) and musical preference between genders (Millar, 2008; Dobrota et al., 2019). Future studies might aim for a higher sample size and a more balanced gender ratio to further evaluate possible differences in the perception of bird vocalizations on the individual level. Perception of birds had a significant effect on pleasantness ratings as participants that had higher scores also rated songs as more pleasant. This might be connected to findings by Ratcliffe et al. (2016) as the perceived restorative potential of a bird sound is related to personal memories. Participants that have higher scores in perception of birds most likely spend more time with nature related activities such as birding leading to more positive associations and memories with bird vocalizations. Further studies might compare the rating scores between participants with low (novices) and high (experts) birding skills to further investigate how prior knowledge affects the perception of bird vocalizations.

Lastly, it is worth mentioning, that our results represent ratings from a single cultural context, i.e. German adults as participants and bird’s native to Germany as sound stimuli. Consequently, our results regarding sound pleasantness may not be generalized to other cultural or regional backgrounds. Previous studies already showed a relationship between personal associations and human sound perception (McDermott, 2012; Ratcliffe et al., 2013, 2016) as well as regional differences in music preferences (Mellander et al., 2018; Hanquinet and Taylor, 2025). Furthermore, while our analyses focused on average pleasantness ratings across participants, individual differences in sound perception were not explicitly modelled. Future studies could therefore use mixed-effects models that account for the repeated structure of data (multiple ratings per participants) to investigate how cultural, socioecological, and individual factors contribute to variation in sound pleasantness ratings.

5 Conclusion

Our results suggest that both subjective (perception of birds) and objective measurements (acoustic parameters) of sounds are associated with perceived sound pleasantness of bird vocalizations. Besides expanding the understanding of human perception of bird vocalizations, these results may offer practical implications for soundscape management and the design of restorative environments. It is increasingly recognized that acoustic environments play an important role for human well-being and our results suggest that not all bird vocalizations contribute equally to perceived pleasantness, but that specific acoustic characteristics such as higher low frequencies, moderate amplitudes, narrow bandwidth and greater vocal complexity, are associated with more positive perceptions. Consequently, assessment of soundscape quality may benefit from considering the acoustic composition of bird communities rather than solely reporting species richness or abundance. Considering that our pleasantness scores positively correlate with the perceived restorative potential of Ratcliffe et al. (2016), management practices supporting the presence of acoustically pleasant bird species could potentially contribute to a more enhanced restorative acoustic environments in urban green spaces and recreational areas. However, further studies are needed to determine whether the presence of species with higher pleasantness scores translates into measurable restorative or well-being benefits in humans. Additionally, our results indicate that acoustic characteristics may represent an additional dimension of charisma that affects human perception of wildlife. While conservation programs primarily focus on ecological needs, visually charismatic species are often used as flagship species to motivate public support (Clucas et al., 2008; Home et al., 2009; Colléony et al., 2017). Our findings suggest that considering acoustic pleasantness alongside visual attractiveness when selecting flagship species may improve conservation communication and strengthen human-nature connectedness.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbirs.2026.1794594/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Tina Phillips, Cornell University, United States

Reviewed by: Canwei Xia, Beijing Normal University, China

Rocio Bahia, National University of Comahue, Argentina