Exploring the Soundscape in a University Campus: Students’ Perceptions and Eco-Acoustic Indices

Abstract

Urban noise pollution significantly degrades people’s health and well-being and, furthermore, traditional noise reduction strategies often overlook individual perception differences. This study proposed to explore the role of eco-acoustic indices in capturing the interplay between biophony, geophony, and anthrophony, and their relationship with classical acoustic metrics and the perceived soundscapes within a University Campus (University of “Mila-no-Bicocca”, Italy). The study area is divided in to eight different sites in “Piazza della Scienza” square. Sound measurements and surveys conducted in June 2023 across four paved sites and adjacent courtyards involved 398 participants (51.7% female, 45.6% male, 2.7% other). The main noise sources included road traffic, technical installations, and human activity, where traffic noise was more prominent at street-level sites (Sites 1–4) and technical installations dominated underground courtyards (6–8). Human activity was most noticeable at Sites 4–8, especially at Site 5, which showed the highest activity levels. A circumplex model revealed that street-level sites were less pleasant and eventful than courtyards. Pairwise comparisons of noise variability showed significant differences among sites, with underground locations offering quieter environments. Eco-acoustic analysis identified two site groups: one linked to noisiness and spectral features, the other to intensity distribution metrics. Technical installations, people, and traffic noises showed distinct correlations with acoustic indices, influencing emotional responses like stimulation and liveliness. These findings emphasize the need to integrate subjective perceptions with objective noise metrics in soundscape descriptions.

1. Introduction

The rise in noise pollution across large cities has become an increasing concern for both communities and policymakers [1]. According to the World Health Organization [1], noise is a major environmental risk factor, significantly impacting human health and well-being [2]. The auditory effects of noise include hearing impairment and tinnitus, while non-auditory effects extend to cardiovascular and metabolic disturbances, adverse birth outcomes, deteriorated quality of life, mental health issues, cognitive impairment, and poor sleep. Among these, sleep disturbance and annoyance are the most prevalent effects in urban areas. In Western Europe, road traffic noise is responsible for the loss of at least 1.6 million healthy life years annually [1], which has spurred the EU Noise Directive (2002/49/CE) to target the reduction of environmental noise [3].

“Noise” is a subjective concept, encompassing sounds that are perceived as unpleasant, unexpected, unwanted, or harmful [4,5]. The paradigm shift to noise perception results in varied interpretations of acceptable noise levels among individuals [6], as well as across different socioeconomic and demographic groups [7] and cultural contexts [8]. It has long been recognized that assessing noise solely based on A-weighted sound pressure levels is insufficient for predicting annoyance or the broader noise impacts, nor does it adequately evaluate acoustic quality and well-being [9,10]. Nevertheless, most urban noise studies and city regulations continue to rely on objective metrics such as decibel levels (dB), which aggregate all ambient sounds without source distinction.

For a comprehensive evaluation of sound environments, it is crucial to apply different methods. One such approach is the concept of soundscape, as defined by [11], which accounts for the spatial and temporal variations in sound, influenced by the built environment and various sources, including biophony, geophony, and anthrophony. A key research area that applies this concept is that of soundscape ecology, which is the study of the acoustic relationships between living organisms, either human or other, and their environment. In this field, soundscapes are defined as the collection of sounds emanating from a landscape, forming acoustic patterns across spatial and temporal scales [12]. Soundscape ecology has grown significantly, utilizing sound recordings and calculation of eco-acoustic indices that provide insights into the health and quality of natural areas [13]. These indices summarize audio information by confronting pitch saturation and amplitude across time steps or frequency bins [14,15], providing a foundation for rapid large-scale analyses. However, little is known about how eco-acoustic indices relate to human perception. This highlights a critical gap in the literature, which represents a promising area for future research [16]. A recent study suggests that these indices enhance our understanding of psychoacoustic perception [17]. Thus, integrating traditional acoustic metrics with eco-acoustic indices offers a more comprehensive approach, bridging the gap between objective measurements and subjective individual experiences.

In recent years, urban design and planning have increasingly recognized the interplay between the physical environment and social perceptions in shaping public spaces [18,19,20]. University campuses, as hubs of knowledge creation and interaction, represent a unique context for studying soundscapes. These environments balance functional requirements, such as socialization and education, with the need for restorative spaces that foster well-being and creativity [21]. In this context, the Multilayered Urban Sustainability Action (MUSA) project in Milan, Italy, aims to promote sustainable measures that enhance both environmental quality (e.g., biodiversity) and inhabitants’ quality of life, starting with the urban regeneration of the University district of “Milano-Bicocca”.

This topic is particularly important as university campuses serve as hubs for knowledge creation, interaction, and community building [22,23,24,25]. In fact, recent studies confirm that well-designed university environments can foster creative encounters in public spaces [26]. Furthermore, “quiet areas” are widely recognized for their role in rest and recovery, making them especially important in university settings where they facilitate socialization and education [21].

Despite the growing interest in soundscape research, few studies have integrated both subjective sound perceptions and eco-acoustic indices in urban contexts. This study aims to bridge this gap by exploring the interplay between objective acoustic metrics, eco-acoustic indices, and subjective perceptions of soundscapes within a university campus environment. The study described the sound perceptions of students in “Piazza della Scienza” at the University of “Milano-Bicocca” during daytime. Beyond traditional noise measurements, such as the A-weighted equivalent sound pressure level (L_Aeq), the study also calculates eco-acoustic indices to evaluate the relationship between biophony and anthrophony and sound perception in different areas of the square, a novel approach for the field. A survey, including ad hoc questionnaires and environmental sound recordings, was conducted with students in the area. The final goal was to compare acoustic and eco-acoustic measurements with perceived sound quality, ultimately establishing design principles for the urban space outside the university buildings. We expect eco-acoustic indices to correlate with people’s sound perception and provide additional insights beyond those offered by traditional acoustic indices. Understanding the role of eco-acoustic indices in shaping people’s soundscape perceptions could drive the development and use of indices that more accurately explain soundscape perception outcomes, ultimately enhancing urban public health and spatial planning. The main questions of this study are as follows:

I.

Do students’ perceptions of soundscapes differ within sites of the “Piazza della Scienza”, considering site functions and user behaviors?

II.

Do acoustic data (e.g., L_Aeq, Delta L_A10–L_A90) explain students’ perceptions of sound quality?

III.

Do eco-acoustic indices explain students’ perceptions of sound quality?

2. Materials and Methods

2.1. Study Area

“Bicocca” was a former industrial district in the northern periphery of Milan, Italy, and, in the nineties, after a drastic urban renovation, became one of the main university centers in the city, hosting the newborn University of “Milano-Bicocca”. The study took place in “Piazza della Scienza” (Figure 1), which is one of the main squares of the district and is delimited by four university buildings (U1, U2, U3 and U4). The square is car-free and crossed by the tram rails, dividing the space into two sides. The limits of the square are the surrounding university buildings. The square is connected to two dual carriageway roads with moderate traffic, both to the east and to the west. The area of the square is mainly frequented by students, researchers, professors and university workers.

Data collection was performed during May 2023 in eight sites of the square: four paved sites in front of each of the buildings and their four corresponding underground courtyards (Figure 1). Eight sites were chosen to represent the diversity of the acoustic environment within the square, focusing on areas differing in sound sources, spatial configuration, and surrounding infrastructure. The square is arranged into two levels, surrounded by six-floor buildings (U1–U4), experiencing significant anthropogenic noise. The ground level (orange polygons) is characterized by a fully cemented floor, whereas in the sunken courtyards (green polygons) grass and trees are present, namely in Sites 6, 7 and 8, while Site 5 has no grass (see Figure 1).

2.2. Acoustic Data Collection

Acoustic measurements took place during one week of May 2023 (22 May 2023 to 30 May 2023), using a Class 1 sound level meter that had been previously calibrated (Svantek Italia Srl, Caurgate, Italy and Accon Italia Srl, San Genesio ed Uniti, Italy). The sound level meters were calibrated at the beginning of the measurement period, with additional calibrations performed weekly to ensure accuracy and consistency across the dataset. The instruments were positioned strategically at 5 points of the Piazza (Figure 1—red dots). As this study was conducted in the context of an urban renewal project, and Site 5 is not scheduled for any specific intervention, acoustic data were not collected at this site, only eco-acoustic and survey data. The SLMs were used to continuously acquire noise levels and assess the acoustic clime of the square. At the same time, 8 Song Meter Micro (WildLife Acoustic Inc., Maynard, MA, USA) instruments were used, hanging from the level meters (Figure 2), to monitor the soundscape (Figure 1—blue dots). These instruments were programmed to record the week’s soundscape for 1 min for every 2 mins of pause, using a sampling rate of 48 kHz and an amplitude gain of +12 dB; their output is an uncompressed 16-bit WAV file.

2.3. Field Survey

An on-site survey was designed to investigate the soundscape of the Piazza through students and university workers’ perceptions, opinions and assessments. Fourteen questions were elaborated following the ISO guidelines [4] to make the survey as replicable as possible. The following questions delved into the 8 sites’ sound environments:

-

Students’ and university workers’ general opinions on the soundscape (from extremely negative, 0, to extremely positive,4, how would you describe the sound environment).

-

Appropriateness of the sound environment (how appropriate is the sound environment from not at all, 0, to completely appropriate, 4).

-

Perceived types of sounds (what is the level of the intensity of traffic, technical installations, people and natural sounds from very low, 1, to very high, 4).

-

Most annoying sound source (what is the most bothersome sound source between traffic, technical installations, people and nature).

-

Soundscape perceived features (from 0, not at all, to 4, definitely, to what extent do you agree with the following adjectives to describe the sound environment: pleasant, chaotic, stimulating, boring, calming, bothersome, lively and monotonous).

The survey also included questions on age, gender, purpose (for what reasons do you visit the square) and frequency of visit (how often do you visit the square), time spent in the area (few minutes, some hours, all day), sensitivity to noise (to what extent are you sensible to noise) and self-reported psychophysical condition (to what extent did you feel sad, listless, energetic, calm and active in the past two weeks) to enhance the analysis of the soundscape in relation to respondents’ sociodemographic characteristics and habits. One question on the appraisal of the visual environment of the site was also added to the survey (in general, how would you describe the visual environment of the site from very negative,0, to very positive, 4).

Following the ISO guidelines [4], answers to the soundscape perception were given on a 5-point Likert scale with “0” corresponding always to the worst condition and “4” to the optimal. The questionnaires were administered at the 8 sites of the “Piazza” (see map) throughout seven sunny days from 9:00 to 18:00. A convenience sampling was used, as researchers approached passers-by and people who were sitting in the sites—mainly students—in order to ask them to participate in the survey. Respondents voluntarily decided to answer the questions posed by the researchers, who briefly explained the purpose of the study. Before starting the questionnaire, they were asked to focus for a few seconds on the sounds and on the environment of the site. The questionnaires were filled out with interviewees’ phones, by scanning the QR code provided by the interviewers and navigating through the questions on the ArcGIS online website accessible via Android and IOS mobiles, as the survey was set up with the ArcGIS Survey 123 software version 3.19.114. Before submitting their answers, the respondents had the possibility to see and sign the privacy policy notice for the protection of personal data. The questionnaires were deliberately brief—approximately 4 or 5 min—and the researchers were always available for clarifications on the meaning of the questions.

2.4. Acoustic Data Analysis

The acoustic data collected were extracted from the sound level meters using their corresponding software (Svan PC++ version 3.4.19 and NoiseDataView version 5.63 Beta). Data exported included the continuous equivalent sound level L_Aeq referred to the day period/hour (6:00–19:00) of the week of the survey. The L_Aeq and the octave band sound pressure level L_A were determined by SVANPC++, VibRum PLUS+ and NoiseDataView. Records during adverse weather conditions (i.e., rain greater than 0.20 mm and wind speed greater than 5 m/s) were removed from the analysis. The weather data were obtained by the weather station, Davis Vantage Pro 2, installed on the roof of the U4 building, and downloaded via the Weatherlink software 6.0.3.

In this study, the “Piazza” sites were described using sound recordings collected throughout the week of data collection during the daytime survey period (6:00–19:00). For the analysis of participants’ sound perception, only recordings captured during the half-hour corresponding to their responses’ time were used.

2.5. Acoustic Indicators

The determined acoustic indicators of the sonic environment in the “Piazza” are listed in the following:

L_Aeq, A-weighted equivalent continuous sound pressure level;

L_A10, the A-weighted sound level exceeded 10% of the measurement time, usually representative of the highest sound levels;

L_A90, the A-weighted sound level exceeded 90% of the measurement time, generally taken as representative of background noise;

Sound climate, L_A10–L_A90, often taken as indicator of the time variability of the sonic environment.

For the description of the “Piazza” we used the levels of the full week of survey while for the comparisons with individual’s answers we used only the level during the hour of each participant answer. The differences of acoustic indicators across the measurement sites were statistically tested using a non-parametric test (Kruskal–Wallis) with pairwise comparisons.

2.6. Microphone Calibrations for Eco-Acoustic Analysis

Soundscape recorders are not calibrated and do not show a flat frequency response, which introduces biases in the analysis of recordings and hinders comparisons with data collected by other devices, like sound level meters. To address this issue, the Song Meter Micro recordings were calibrated according to the procedure outlined in [27,28]. This calibration ensured that eco-acoustic indices derived from the recordings were directly comparable to data from calibrated sound level meters, eliminating biases introduced by frequency response inconsistencies. Specifically, the filter was derived by comparing the power spectral density of a reference signal (white noise) recorded in an anechoic chamber by a sound level meter (LD831-C) and the Song Meter Micro. A detailed explanation of the procedure is provided in [28].

2.7. Eco-Acoustic Indices

Eco-acoustic indices are metrics in soundscape ecology used to summarize audio data and classify soundscapes by analyzing pitch, saturation, and amplitude [29]. These indices facilitate the comparison of time steps or frequency bins to identify patterns and variations [15]. Moreover, they provide a quick assessment of long-term audio recordings and help to extract biological insights about the environment (Llusia, 2024) [30]. In this study, eco-acoustic indices were calculated using the Soundecology package [31] in R software 2024.09.0 [32].

The considered indices were as follows:

(a)

The acoustic complexity index (ACI) quantifies the vocalizations of avifauna by analyzing the sound intensity modulation, which varies rapidly over time in the case of biophony but is very constant for numerous anthropogenic noises [33,34]. It is calculated by performing the amplitude difference of adjacent time samples within a frequency band, relative to the total amplitude of that band [34].

(b)

The acoustic diversity index (ADI) provides a measure of the intensity distribution diversity by calculating the Shannon index on the time intervals divided spectrum [29,34]. Low values are associated with intensity distribution diversity (i.e., nocturnal insects [13]) while high values to distribution evenness (i.e., high levels of geophony and anthrophony [13]).

(c)

The acoustic evenness index (AEI) uses the same rationale as ADI but applies the Gini coefficient instead of the Shannon index and consequently measures the inequality of intensities [35].

(d)

Acoustic entropy (H) estimates the total entropy or heterogeneity of the recordings by calculating the product of Shannon’s spectral entropy and temporal entropy [36]. H values range in [0, +1]; +1 refers to an even signal (i.e., silent recording or faint bird calls) while 0 refers to a pure tone (i.e., insect vocalizations) [34].

(e)

The dynamic spectral centroid (DSC) provides information about the typology of sound events using the spectral centroid of the recording [Hz]; the SC is calculated by dividing the spectrum into time intervals and computing the gravity center [29].

(f)

The zero-crossing rate (ZCR) measures the noisiness of the recording by calculating the number of times per second that the acoustic signal crosses the null value of the instantaneous sound pressure; high values of crossing rate are associated with noisy recordings and biophony presence, whereas low values are linked to tonal sounds [37,38].

(g)

The number of peaks (NP) counts the major frequency peaks obtained on a mean spectrum scaled between 0 and 1. It is linked to animal sound activity level (defined as the number of different song types), which is less sensible to ambient noise [39] and inversely correlated to dB (A) [40].

To analyze the structure of the eco-acoustic dataset and assess the differences between sites in the “Piazza”, a principal components analysis (PCA) was carried out. This multivariate test aims at extracting from the dataset the most important information, simplifying its description while reducing its size and analyzing its structure [41]. The difference between sites has been evaluated by performing a multiple factor analysis on the first two principal components, which analyzes a set of observations (indices values for each audio recording) described by several groups of variables (indices); only the first two components have been considered, as they retain the largest possible variance [41]. In this study, PCA was performed using the Stats and the Factorextra [42] functions in R software [32].

2.8. Questionnaire Data Analysis

After removing the observations with invalid and empty answers, 398 valid cases out of 420 responses were considered for the analysis. A descriptive analysis of the sample was carried out to delineate the socio-demographic characteristics of the respondents. Spearman’s correlations were used to analyze relationships between perception variables, ensuring robust statistical insights into the interplay between soundscapes and psychophysiological conditions. Frequencies, descriptive analyses and contingency tables were then carried out to study the general soundscape perception and the differences between respondents’ groups. The cases’ breakdown was based on gender, the site where the interview took place, participants’ self-reported level of sensibility to noise, number of hours usually spent in the square, and self-reported level of psychophysical conditions. Spearman’s correlations and Chi-squared tests were then performed with IBM SPSS 29 software to delve into the relationships between the survey’s variables. A preliminary analysis of the results motivated us to make some direct observations in the field, highlighting the differences between the 8 sites in the square, in terms of the facilities and of the physical and social environments.

Furthermore, according to the ISO/TS 12913-3:2019 [43,44], a circumplex model was used to represent the affective responses in a two-dimensional model. The main dimension is related to how pleasant or unpleasant the environment was judged, and was therefore noted as pleasantness; meanwhile, the second dimension is related to the amount of human and other activity [44].

2.9. Acoustic, Eco-Acoustic and Questionnaire Data Analysis

The results of the questionnaire on the soundscape perceptions were then analyzed in relation to acoustic (L_Aeq, L_A10–L_A90) and eco-acoustic indices, measured during the survey (hour of each participant). Spearman correlations were performed to assess the relation between the perception of sounds—such as sound appropriateness, level of appreciation of soundscape, pleasantness, eventfulness and perceived intensity of sound sources—with acoustic measures, such as L_Aeq and L_A10–L_A90. Acoustic, eco-acoustic and questionnaire data were also analyzed using Spearman correlations site by site to highlight the spatial differences of the soundscape in “Piazza della Scienza”.

Plots and graphic designs in this study were performed using GIS (QGIS.org, 2024), with the R packages “dyplr”, “tidyr”, and “ggplot2” [45,46,47].

3. Results

3.1. The Respondents’ Sample: Sociodemographics and Participants’ Psychophysical Conditions

The majority of respondents (approximately 95%) were university students aged between 18 and 30, while the remaining 5% were primarily researchers or professors over 30 years old. The sample was well balanced in terms of gender, with 45.6% male and 51.7% female participants. Most respondents were familiar with the location, as over 63% visited the square more than four times a week, while only 3% were there for the first time. Moreover, 60% of the participants typically spent only a few minutes in the square, whereas four out of ten reported spending hours there regularly.

Analysis of responses to questions about self-reported psychophysical states revealed strong correlations between feelings of unwillingness and sadness, as well as an inverse correlation between negative and positive states. To further explore these relationships, a factor analysis was conducted to identify underlying variables that could explain the observed patterns of correlation. This analysis yielded a single factor that accounted for more than 48% of the variance (KMO test = 0.773, p < 0.001). The factor score, calculated using the regression method in IBM SPSS, was interpreted as an index of psychophysical condition.

3.2. General Soundscape Perceptions

When participants were asked to characterize the current sound environment of “Piazza della Scienza”, most of them described it as neither distinctly negative nor positive. Statistical analysis showed no significant differences in responses between male and female participants. The average assessment of the square sound environment reflected this neutral perception, with a mean score of 2.05 on a scale from 0 to 4. According to the respondents, the primary source of acoustic annoyance was the university’s technical installations (43%), followed by road traffic (36%) and ambient noise from people (20%).

It is worth addressing that the soundscape quality perception varied with the duration of participants’ stays in the square. Those who spent less time in the area tended to find the sound environment less stimulating, less eventful, and more monotonous and duller (Figure 3). Correlation analyses revealed weak relationships between self-reported psychophysical conditions and soundscape appraisal, indicating that psychophysical status did not significantly influence perceptions of the soundscape. However, strong and significant correlations were identified between the overall assessment of the sound environment and certain aspects of its perceived affective quality. Specifically, there were high positive correlations between a favorable evaluation of the sound environment and perceptions of pleasantness (Spearman coefficient = 0.506, p < 0.01) and calmness (Spearman coefficient = 0.42, p < 0.01). Conversely, negative correlations were found between the assessment of the sound environment and perceptions of annoyance (Spearman coefficient = −0.496, p < 0.01) and chaos (Spearman coefficient = −0.39, p < 0.01). Additionally, the results show a significant correlation between perceptions of the visual and sound environments (Spearman coefficient = 0.273, p < 0.01).

3.3. The Differences Within the Square: Analyses per Site

The number of participants interviewed per site demonstrated a degree of homogeneity, with an average of 50 and a median of 49 cases per site (ranging from 34 to 65) (see

Table 1. Composition of participants groups at each site: number of participants, mean age, median age, minimum age and maximum age.

SITE	N	AGE (MEAN)	MIN AGE	MAX AGE
1	55	24.7	18	57
2	39	22.8	18	45
3	64	22.1	18	45
4	42	23.9	19	60
5	65	21.9	18	27
6	46	22.3	19	45
7	52	22.0	19	40
8	34	22.8	19	31
TOTAL	397	22.8	18	60

). There was a significant association between site and soundscape appreciation (Χ2 (14) = 38.82, p < 0.001). The assessment of the surrounding sound environment revealed that the grassed courtyards were perceived more favorably than the paved areas. Specifically, Site 2 received the lowest ratings, while Sites 6, 7, and 8 were rated the highest (Figure 4).

The analysis of sound sources revealed that road traffic noise was less noticeable in the courtyards, which are situated below ground level and shielded by buildings, whereas noise from technical installations was perceived as louder in Sites 3, 6, 7, and 8 (Figure 5). Contingency tables (see Table S1) show that road traffic noise was more frequently perceived as loud in Sites 1, 2, 3, and 5, with a decreasing trend in Sites 6, 7, and 8. The ANOVA (see Table S4) confirms a significant (sig. < 0.01) difference between the sites and the post hoc LSD analysis (Table S5) shows clear distinctions between the street-level sites (1 to 4) and the grassed underground sites (6 to 8).

Noise from technical installations was noticeable across all sites and namely in Sites 2,3 and 7 (see Table S2). The ANOVA (see Table S4) confirmed a significant (sig. < 0.0.1) difference between the sites and the LSD post hoc test (Table S5) shows significant differences between Sites 4 and 5—where the mean value is lower—and all of the other sites. People’s sounds (voices, laughter, steps, etc.) were perceived across all sites, but namely in Sites 5 and 8 (see Table S3). The ANOVA (Table S4) showed significant differences between sites and the LSD post hoc test (Table S5) confirmed significant differences between Sites 5 and 8 (with the highest mean values) and all of the other sites.

Technical installations were identified as the most bothersome in Sites 6 to 8, where road traffic noise was considered as less disturbing. In contrast, traffic was the predominant nuisance in Sites 1 and 4, while in Site 3, both traffic and technical installations were reported as equally bothersome (Figure 6). Notably, in Site 5, the sound of people was perceived as the most irritating. Using the responses regarding sensations induced by the sound environment in the square, a circumplex model (Figure 7 and Figure 8) was developed according to [43] and the ISO/TS 12913-3:2019 standard [44]. The results show that the paved sites (1–4) are clustered around the lower values of pleasantness and eventfulness, whereas sites in the grassed courtyard are clustered around the higher values.

3.4. Acoustics and Eco-Acoustics Description of “Piazza della Scienza”

Regarding the description of the acoustic and eco-acoustic characteristics of the “Piazza”, the Kruskal–Wallis and pairwise comparisons applied to L_Aeq levels across the sites (hourly levels of day period during the week of survey), showed significant differences between several sites (Table 1; Figure 9). Site 1 differs significantly from Sites 2 (p = 9.38 × 10⁻⁵), 3 (p = 9.38 × 10⁻⁵), 4 (p = 3.48 × 10⁻⁸), 6 (p = 4.77 × 10⁻³), and 8 (p = 1.17 × 10⁻¹²). Site 8 shows significant differences with nearly all sites: Site 2 (p = 0.037), Site 3 (p = 0.037), Site 5 (p = 0.0012), and Site 7 (p = 2.00 × 10⁻⁵). Meanwhile, Sites 2, 3, 4, 6, and 7 are similar to each other (p-values > 0.05), though Site 7 differs significantly from Site 4 (p = 0.019).

The pairwise comparisons of L_A10–L_A90 differences (Table 1; Figure 9) between sites showed that Site 1 is significantly different from Sites 2 (p = 7.70 × 10⁻³), 3 (p = 7.70 × 10⁻³), 4 (p = 9.92 × 10⁻⁴), 6 (p = 4.01 × 10⁻⁶), 7 (p = 1.91 × 10⁻¹⁰), and 8 (p = 6.15 × 10⁻⁵). Site 7 also exhibits significant differences with Sites 2 (p = 0.017) and 3 (p = 0.017) but does not differ from Site 4 (p = 0.094). Site 8 shows significant differences only with Site 1 (p = 6.15 × 10⁻⁵), but does not differ significantly from Sites 2, 3, 4, 6, or 7 (all p-values > 0.05). Sites 2, 3, 4, and 6 show no significant differences among themselves. Additionally, Site 6 does not significantly differ from Sites 7 and 8 (p = 1). For details on the frequency spectrum in ⅓ octaves bands in the different sites see Supplementary Figures S1 and S2.

Regarding the eco-acoustic indices of the different sites, the principal component analysis (PCA) performed with weekly data of Sites 1–3 and 5–8 showed some interesting differences. The PCA of eco-acoustic indices revealed that the first three dimensions explained a significant portion of the variance in acoustic characteristics (79.1%), highlighting the distinctiveness of the sites. The first dimension depends mainly on H and ZCR, while the second one on ADI and AEI, and the third dimension on DSC, NP, ZCR. The ellipses in the graph (Figure 10) show two groups characterized by two different inclinations of the focal axis (major axis). The first (Group A) includes Sites 1, 5, 6, 7 that have an inclination from left–top to right–bottom, these sites depend mainly on the ZCR, DSC and H indices. The second group (B), including Sites 2, 3 and 8, depends mainly on the ADI and AEI indices having an inclination opposite to the previous one. On the PCA biplot, Group A ellipses, including Sites 1, 5, 6, and 7, are distributed along a top–left to bottom–right inclination and are thus better represented by the zero-crossing rate (ZCR), dynamic spectral centroid (DSC), acoustic entropy (H), and number of peaks (NP) indices, which focus on the noisiness and frequency composition of the soundscape. In contrast, Group B ellipses, comprising Sites 2, 3, and 8, show an opposing inclination exhibiting a major representation by the acoustic diversity index (ADI) and acoustic evenness index (AEI), which evaluate the intensity distribution on the spectrum. Furthermore, the site centroids’ distribution on the biplot shows their site-specific usages and sound sources. Site 1 is the farthest from the plot center, showing distinct acoustic features (higher L_Aeq and disturbance by traffic). Sites 6, 7, 8, and 2 are relatively close, reflecting the similarities between lowered courtyards; in particular, Site 6 slightly gravitates toward Site 1, possibly due to its recorder’s higher positioning, which exposed it to noise from the Piazza’s ground level. Meanwhile, Site 7 is affected by noise from a student bar in the U3 building. Sites 2 and 8 appeared acoustically similar, likely because Site 2’s recorder, although on the first floor, faced a lowered courtyard (Site 8).

3.5. Acoustic Indicators and Soundscape Perception

Spearman correlation between L_Aeq values (mean values at the site in the timespan of the interviews) and soundscape perception items did not show any significant difference. Conversely, analyzing the relation between L_A10–L_A90 and sound perception, a negative correlation is observed with the eventful dimension (r² = −0.116; t = 2.1021, df = 383.08, p-value = 0.0362) and visual perception of the square (r² = −0.150). Meanwhile, a positive correlation was found with the perception of noisiness (r² = 0.128).

3.6. Eco-Acoustic Indices and Soundscape Perception Items

The correlation analysis between the eco-acoustic indices and the source of noise perceived by individuals showed that “technical installation” noise has a negative correlation with ACI (r² = −0.238) and AEI (r² = −0.158) and has a positive correlation with ADI (r² = 0.205) and NP (r² = 0.129). The anthropic noise (sound of people) showed a positive correlation with ACI (r² = 0.111) and negative with ADI (r² = −0.149). Finally, road traffic was positively correlated with ZCR (r² = 0.112). Regarding the emotions, the results reveal a significant negative correlation of ZCR with stimulation (r² = −0.141), while NP has a negative correlation with boredom (r² = −0.143) and a positive correlation with stimulation (r² = 0.147) and liveliness (r² = 0.179). Furthermore, there were no correlations between eventfulness and pleasantness and the eco-acoustic indices (p > 0.05).

4. Discussion

The findings of this study contribute to a deeper understanding of how urban soundscapes are perceived and the complex relationship between sound characteristics and human perception, specifically in a university campus environment. Traditional approaches to managing urban noise often focus solely on acoustic planning, primarily emphasizing the reduction and control of noise sources. However, the results from this study suggest that lower noise levels alone may not always lead to increased satisfaction among people in the area, as pointed out by [48]. As pointed out in [21], environmental factors such as the presence of natural features (i.e., trees) can significantly affect human perception of a place, even when sound pressure levels do not exceed noise limits.

The study on the sound environment of “Piazza della Scienza” reveals that participants generally perceived it as neutral, and that no significant gender differences have been observed in their responses (

Table 2. Acoustic indicators and eco-acoustic indices for each site.

Indicator	Site 1	Site 2	Site 3	Site 4	Site 5	Site 6	Site 7	Site 8
LAeq	60.0	60.0	61.1	59.6	_	60.1	60.9	58.0
LA10	61.1	61.1	61.3	60.9	_	61.4	62.2	59.1
LA90	57.4	58.2	57.7	57.8	_	58.5	59.1	56.7
LA10–LA90	3.6	2.9	3.7	3.2	_	2.9	3.1	2.4
ACI	149.56	147.31	146.59	150.25	146.29	145.18	145.16	145.16
ADI	7.02	7.00	6.99	6.99	7.03	7.03	6.99	6.99
AEI	0.09	0.13	0.14	0.14	0.07	0.07	0.14	0.14
DSC	0.95	0.84	0.76	0.84	0.91	0.85	0.93	0.93
H	0.662	0.627	0.644	0.642	0.656	0.637	0.650	0.634
ZCR	0.023	0.017	0.023	0.020	0.022	0.017	0.020	0.016
NP	14.82	11.88	12.19	10.98	17.26	12.13	11.18	11.18

). The primary sources of annoyance were technical installations and road traffic, while ambient noise from people was less prominent. Perceptions varied with the duration of stay, as short visits led to duller and more monotonous soundscape ratings. Correlation analyses showed weak relationships between self-reported psychophysical conditions and soundscape appraisal, indicating the minimal influence of personal states on perception. However, strong positive correlations were found between favorable evaluations and perceptions of pleasantness and calmness, while negative correlations were observed with annoyance and chaos. These findings highlight the importance of managing acoustic disturbances and fostering pleasant, calming soundscapes to improve overall environmental quality.

The role of space in shaping perceptions of pleasantness and eventfulness was further emphasized by the circumplex model, according to [17,43,44]. Overall, courtyard grassed sites, such as Sites 6, 7, and 8, were perceived as more pleasant and eventful, whereas street-level sites generally get low ratings for pleasantness and eventfulness, and transit-oriented spaces were rated more chaotic, noisy, and less enjoyable. The positive correlations between favorable evaluations of the sound environment and perceptions of pleasantness and calmness suggest that environments perceived as calming and pleasant contribute to a better overall auditory experience. This reinforces the notion that auditory environments conducive to relaxation can enhance well-being, particularly in high-traffic urban areas such as university campuses.

The soundscape analysis in “Piazza della Scienza” showed significant variability in the acoustic characteristics across different sites. Pairwise comparisons of L_Aeq levels and L_A10–L_A90 levels showed Sites 1 and 8 as the most distinct in terms of sound pressure levels (L_Aeq) and noise variability (L_A10–L_A90), highlighting the importance of spatial factors in shaping the auditory experience. The analysis of the perceived sound sources revealed the variability of the soundscape of the “Piazza (

Table 3. Summary of the Piazza della Scienza sites.

Sites	Description	Main Students’ Use	Bothersome Sound Sources	Overall Soundscape Appreciation
1, 2, 3, 4	An open square intersected by a tram line, encircled by four university buildings. A hub of motion and pause—where students linger between classes or await their public transport connection	Transit, waiting, pause	Traffic Technical installation (Site 3)	1,2 and 3 below moderate, 4 moderate
5	A combined study area and waiting space located beneath Site 1, conveniently accessible directly from the university building. Concrete pavement and tables.	Waiting, pause, study	People	Moderate
6, 7, 8	Places for relaxing and staying in between classes underneath Sites 2,3 and 4, accessible from the university building. Lawns and green pavements.	Rest and relax on the lawn, pause	Technical installations	Above moderate

)”. For instance, at street-level sites (Sites 1–4), traffic noise was identified as a prominent factor, often rated as moderate to high perceived intensity, and indicated to be the most bothersome source of noise. In contrast, in underground courtyards (Sites 6–8), where traffic noise is shielded by buildings, technical installations were identified as the most bothersome noise source. It is also important to consider that ground vibrations from trains passing at the nearby station (250 m away) and vehicular traffic might contribute to the low-frequency sounds detected in the Piazza, along with the technical installations, primarily located on the roof. These findings support the theory that urban soundscapes are not merely a reflection of equivalent sound pressure level (L_Aeq), which is also formed by the types of sounds and their contextual appropriateness in specific settings. Other studies highlight similar findings, emphasizing that a multidisciplinary approach is more effective than solely relying on cumulative noise indicators [49,50]. The sound of people was constant across sites, with Site 5 being the highest, due to its use for study activities. Furthermore, the presence of human voices in study-oriented environments (like Site 5) was perceived as disturbing, whereas technical sounds disrupted the seminatural atmosphere in more secluded areas (like Site 6).

The eco-acoustic indices that had the greatest explanatory power for psychoacoustic perception outcomes, according to [17], were analyzed in this study, providing additional insights into the complexity of the soundscape of the “Piazza della Scienza”. For instance, the acoustic complexity index (ACI) demonstrated a negative correlation with the perceived intensity of the sound generated by technical installations, suggesting that environments dominated by artificial sounds tend to exhibit lower acoustic complexity. Moreover, in our study, ACI positively correlates with perceived people’s sounds (e.g., speech, shouts and laughing) due to the complexity that speech adds to the soundscape in frequency modulation, intensity and time duration. The acoustic diversity index (ADI) was positively correlated with the perceived intensity of technical installation noise, probably due to a higher frequency occupancy in places with lower noise attenuation or a higher installation power. On the other hand, the acoustic evenness index (AEI) showed a negative association with it due to its anticorrelation with ADI. On the other hand, people’s sounds are negatively correlated with ADI; in fact, a higher presence of human voices, while adding complexity, reduces the frequency spread and the evenness of sound, leading to lower ADI values, in line with the findings of [51]. ZCR is positively correlated with the perceived presence of traffic as higher values of this index are related to noisy environments [38]. This relation is particularly interesting because of the high variability in the perceived traffic values between the survey sites. Regarding the NP index, the results showed a positive correlation with the perceived intensity of technical installations. Our results contrast those of [40] where NP is reported to be inversely correlated to dB (A); in our study, Sites 1 and 6 have higher levels of both L_Aeq (dB) and NP. This may be due to the difference of source composition and its contribution to the overall sound pressure level. Moreover, NP is positively correlated with stimulating soundscapes as it seems to be linked to animal sound activity level, which makes the sound environment stimulating [39]. Thus, it is important to consider not just the intensity of sound, but also its diversity and evenness in evaluating the overall quality of an environment.

The lack of correlation between eventfulness and pleasantness axes perception with eco-acoustic indices agrees with [17], which found limits to the association and predictive capacity between eco-acoustic indices and these psychoacoustics categories. However, when decomposing the axes, the NP index showed a positive correlation with stimulation and liveliness and a negative with boredom. The positive correlation between NP and liveliness highlights how distinct sound peaks, often associated with human or animal vocalizations, contribute to a dynamic and engaging soundscape. Conversely, the negative correlation between ZCR and stimulation may indicate that environments with high noisiness lack the tonal richness required to evoke positive engagement. Lawrence et al. [17] instead found high levels of NP correlate with calm or pleasant emotions.

5. Conclusions

The findings of this study have significant implications for urban soundscape management. They point out the importance of integrating both objective acoustic metrics and subjective perceptions into the design of urban environments. In further detail, the research shed light on the perception of soundscapes in a university environment highlighting the variability in sound perception across different spatial configurations. Interestingly, the perceived most bothersome source of noise varied between sites located in the same Piazza, confirming the importance of a thorough urban soundscape management that takes into consideration the social composition of the places (in this case, students’ practices, needs and opinions). As the most appreciated soundscapes were found in semi-enclosed green spaces, the research underlined the need for tranquil places that are away from traffic, in order to foster conviviality and creativity in university environments.

Moreover, the research delved into the relationship between acoustic and eco-acoustic data and perceived soundscapes, contributing to the studies on eco-acoustic and psycho-acoustic indices and to their application in the research on urban soundscape. We found significant correlations between eco-acoustic indices (i.e., ACI, ADI, AEI, ZCR, NP) and perceived sound sources (i.e., traffic, people, technical installations). The research also highlighted the limitations of relying solely on traditional noise metrics like L_Aeq, as the sound pressure level was not correlated to any indicator on soundscape perception. These findings pave the way for further exploration of soundscapes in other urban settings, offering insights that could guide interdisciplinary approaches to urban planning and socio-environmental studies.