Birdsong beats traffic: the largest analysis yet shows city sounds shape health and happiness

For decades, the science of urban sound has been a science of harm. Researchers measured decibels, tracked annoyance, and linked traffic noise to cardiovascular disease, disturbed sleep, and poor mental health. A landmark new synthesis argues that this noise-centric view captures only half the story. In a systematic review and meta-analysis published in npj Urban Sustainability, a team led by Xuan Guo of Leibniz University Hannover and Christian Albert’s environmental planning group, together with colleagues at Fujian Agriculture and Forestry University and Fuzhou University, pooled evidence from 194 studies to ask a deceptively simple question: what do the sounds of the city actually do to human well-being? The answer, drawn from 66 studies comprising 1658 effect sizes across 28 sound sources, is that sound is not merely something to be reduced. It is a resource to be designed.

The analytical backbone of the study is the ISO 12913-1 standard, which defines soundscape as the acoustic environment as perceived or experienced and understood by a person, in context. That people-centered definition matters technically because it reframes the acoustic environment from a physical exposure metric, sound pressure level, into a perceptual construct shaped by interactions between sounds, human interpretation, and environmental setting. The researchers adapted this framework into an operational model with three components: soundscape characteristics, covering both the perception of individual sound sources such as birdsong, flowing water, or conversation and the holistic interpretation of the auditory environment; well-being-related outcomes, organized into four domains of environmental experience, human health, behavioral responses, and social well-being; and contextual factors, spanning environmental conditions, personal attributes, and temporal variables such as exposure duration and time of day.

The descriptive mapping of the literature revealed a strikingly uneven evidence base. Natural sounds were the most frequently examined category, appearing in 69.07 percent of the 194 studies, yet the research concentrated on a narrow set of sources, dominated by unspecified birdsong and mixed natural sounds. Human-sound research centered largely on music and speech. Human health was the dominant outcome domain, also examined in 69.07 percent of studies, followed by environmental experience at 32.99 percent, while behavioral responses and social well-being were comparatively neglected. Equally notable is how rarely domains were examined together: only 16.49 percent of studies covered two outcome domains, and none covered three or more. Roughly one third of the studies, 36.08 percent, examined no testable contextual factor at all. The most frequently studied pathway, natural sounds in relation to human health, appeared in 101 cumulative instances, whereas associations between overall soundscape perception and social well-being or behavioral responses appeared in only four and three instances respectively, too few for robust cross-study synthesis.

Where the evidence permitted quantitative pooling, the meta-analysis delivered its headline findings. Natural sounds showed moderate positive effects across environmental experience, human health, and social well-being, with Hedges’ g ranging from 0.53 to 0.59, and the largest effect on human health at g = 0.59 with a 95 percent confidence interval of 0.42 to 0.76. Human sounds showed a more targeted profile: no significant effects on environmental experience or human health in the main-effects model, but moderate positive effects on behavioral responses at g = 0.60, 95 percent CI 0.24 to 0.97, and small positive effects on social well-being at g = 0.38, 95 percent CI 0.08 to 0.68. In plain terms, birdsong and flowing water reliably make people feel better, while the sounds of other people, conversation, music, and play, more selectively shape how people behave in and feel about public space.

Drilling into the health domain revealed a differentiated pattern with direct relevance to viral claims about nature sounds. Natural sounds showed positive effects across all health subtypes, but mental health stood out as the strongest at g = 0.83, 95 percent CI 0.64 to 1.02, significantly stronger than physical health, with a difference of 0.53, and cognitive health, with a difference of 0.35. Human sounds showed a significant effect only for mental health, at g = 0.46. For environmental experience, both sound types improved overall environmental appraisal, natural sounds at g = 0.66 and human sounds at g = 0.54, whereas effects on specific sensory sub-dimensions, visual, thermal, and olfactory experience, did not differ significantly from zero. The authors suggest that soundscape effects are typically expressed at the level of integrated environmental appraisal, where auditory information interacts with visual, thermal, olfactory, spatial, and social cues, though they caution that weaker effects for specific sensory variables may partly reflect how rarely and inconsistently those experiences are measured.

Source-specific analyses added crucial nuance, undercutting any simple prescription that biological sounds are always good. Mixed birdsong showed particularly large effects on environmental experience at g = 1.27, 95 percent CI 0.69 to 1.85, and on social well-being at g = 0.57. Music showed the most consistent positive effects among human sounds, especially for environmental experience at g = 1.20, 95 percent CI 0.81 to 1.58, and the interaction model even revealed a significant positive effect of music on thermal comfort at g = 0.57 that was not apparent in the main-effects model. But dog barking, a biological sound by category, showed a strongly negative effect on environmental experience at g = -1.21, 95 percent CI -1.78 to -0.64. The meaning a sound carries, its intensity, source identity, and compatibility with the setting, matters as much as its broad category. Water sounds may be restorative when perceived as natural and clean, but less effective when artificial, excessively loud, or visually disconnected from the environment.

Contextual moderation proved to be the study’s most technically consequential contribution. Environmental setting type moderated soundscape-well-being relationships in opposite directions for the two sound types: natural sounds were effective for environmental experience in outdoor settings at g = 0.57 but not in no-visual settings, whereas human sounds showed the reverse pattern, significant in no-visual settings at g = 0.59 but not outdoors. Yet the interaction model revealed that within no-visual settings, where the auditory pathway is isolated from competing cues, natural sounds produced their largest effects on overall environmental experience at g = 0.96 and mental health at g = 1.17, an estimate of what sound alone can achieve. Control group type mattered as well: effect sizes were generally larger when soundscapes were compared with noise exposure than with silence, and human sounds showed significant positive effects primarily as alternatives to noise. Relative to noise, both sound types improved environmental experience and mental health, natural sounds reaching g = 1.18 for mental health. The authors interpret this carefully: part of the observed benefit of human sound may reflect relief from an adverse acoustic baseline rather than an intrinsic benefit of the sound itself.

Exposure duration and geography added further layers of conditionality. Effects did not increase monotonically with exposure. For human health, the largest effect sizes for both natural sounds, g = 0.73, and human sounds, g = 0.53, occurred at just one to five minutes of exposure. Short exposure to natural sounds affected mental and physical health, effects on mental health persisted beyond five to ten minutes, and a significant effect on cognitive health emerged only after more than ten minutes, at g = 1.00. Geographical context significantly moderated relationships, particularly for natural sounds, with larger effects on environmental experience in Europe, g = 0.78, than in Asia, g = 0.46, while significant health effects for both sound types were observed only in studies conducted in Asia. The authors stress that these four moderators are proxies for context rather than measurements of it, recording where a study was conducted or how it was designed, not the perceived meaning, appropriateness, or social fit that the patterns implicate. That context-dependence is visible at all through instruments this coarse suggests the true dependence of human-sound effects on listening conditions is likely stronger than the estimates indicate.

The limitations are candidly acknowledged. Primary studies varied substantially in design, stimuli, control conditions, outcome measures, and exposure duration, and the multilevel model, which nests effect sizes within studies using the metafor package in R, partitions heterogeneity into between-study and within-study components but does not resolve these conceptual differences. Sparse reporting of demographic, cultural, and temporal detail forced coarse contextual coding, and given the multiplicity of models and post-hoc pairwise comparisons, moderator findings remain exploratory even with Benjamini-Hochberg false-discovery-rate adjustment. Blinding of participants or experimenters is rarely feasible in auditory interventions. The review was confined to English-language, peer-reviewed, quantitative studies, excluding qualitative methods, conference papers, and scholarship strands such as acoustic ecology and technical acoustic modeling, so the evidence base represents a specific subset of the field. Publication bias testing detected potential small-study or time-lag effects in three models, with no significant evidence in the remaining models.

The planning implications are nonetheless concrete. For health outcomes, the consistent positive effects of natural sounds support protecting and enhancing restorative acoustic resources, birdsong, water, and wind-induced vegetation sounds, alongside continued reduction of intrusive mechanical noise. Because such sounds depend on the ecological and spatial conditions that generate them, this implies habitat management, ecologically appropriate water features, vegetation structures, and spatial buffers against traffic-noise intrusion, making the identification and protection of existing high-quality soundscapes as important as creating new ones. For social well-being, certain human sounds may support social presence, perceived safety, and place attachment in settings designed for interaction, but should not be added indiscriminately; in quiet or restorative settings the same sounds may be unnecessary or intrusive. Interventions, the authors argue, should be evaluated not only by decibel reduction but also by perceived appropriateness, pleasantness, restoration, behavioral use, and social meaning. The dual strategy is clear: keep reducing harmful noise exposure while explicitly recognizing, protecting, and designing for the positive contributions of sound. The city of the future, on this evidence, will not be silent. It will be composed.

Subject of Research: Urban Sustainability

Article Title: Beyond noise and silence: urban soundscapes and their contribution to human well-being—a systematic review and meta-analysis

Article References: Guo, X., Schmidt, S., Chen, Z., Yang, L., Liu, J., & Albert, C. (2026). Beyond noise and silence: urban soundscapes and their contribution to human well-being—a systematic review and meta-analysis. npj Urban Sustainability, 6(1), Article 130. https://doi.org/10.1038/s42949-026-00482-2

Image Credits: AI Generated

DOI: 10.1038/s42949-026-00482-2

Keywords: soundscapes, urban planning, well-being, noise pollution, meta-analysis, environmental psychology, public health, birds, Beyond, noise, silence, urban