Characterizing amplitude and frequency modulation cues in natural soundscapes: A pilot study on four habitats of a biosphere reserve

Abstract

Natural soundscapes correspond to the acoustical patterns produced by biological and geophysical sound sources at different spatial and temporal scales for a given habitat. This pilot study aims to characterize the temporalmodulation information available to humans when perceiving variations in soundscapes within and across natural habitats. This is addressed by processing soundscapes from a previous study [Krause, Gage, and Joo. (2011). Landscape Ecol. 26, 1247] via models of human auditory processing extracting modulation at the output of cochlear filters. The soundscapes represent combinations of elevation, animal, and vegetation diversity in four habitats of the biosphere reserve in the Sequoia National Park (Sierra Nevada, USA). Bayesian statistical analysis and support vector machine classifiers indicate that: (i) amplitude-modulation (AM) and frequency-modulation (FM) spectra distinguish the soundscapes associated with each habitat; and (ii) for each habitat, diurnal and seasonal variations are associated with salient changes in AM and FM cues at rates between about 1 and 100 Hz in the low (<0.5 kHz) and high (>1-3 kHz) audio-frequency range. Support vector machine classifications further indicate that soundscape variations can be classified accurately based on these perceptually inspired representations.

Auditory hemispheric asymmetry as a specialization for actions and objects

“…Using neurally relevant spectro-temporal representations (the STM framework), these works show that different subspaces of the STM encode distinct information types: temporal modulations for meaning (speech), spectral modulations for melodies, or very fast temporal modulations for alarms (screams). These latter are also critical to discriminate between different soundscapes of natural spaces (Thoret et al, 2020). Here, by analyzing corpora of natural environmental sounds, we show that sounds are mostly composed of purely temporal and purely spectral modulations (Figure 1), at ranges also observed in speech and music.…”

Double-pass consistency for amplitude- and frequency-modulation detection in normal-hearing listeners

“…The auditory perception of amplitude modulation (AM) and frequency modulation (FM) has received substantial interest over the last decades because of the repeated demonstration of the crucial role played by AM and FM cues in robust speech perception (e.g., Shannon et al, 1995;Zeng et al, 2005) and in environmental sound perception (Singh and Theunissen, 2003;Thoret et al, 2020). It is generally assumed that auditory processing of AM and FM overlaps because of "FM-to-AM conversion" in the cochlea (the frequency-dependent attenuation of the FM caused by the tuned cochlear filters resulting in AM cues; Zwicker, 1952;Saberi and Hafter, 1995).…”

Human Auditory Ecology: Extending Hearing Research to the Perception of Natural Soundscapes by Humans in Rapidly Changing Environments

“…These include open and closed habitats from tropical, sub-tropical, temperate and arctic biomes that are often recorded with a relatively high temporal resolution (in most cases, 1 min every 15 min) for many months and seasons, and sometimes for several years (e.g., Gage & Axel, 2013). Figure 2 shows two-dimensional amplitude-modulation (2D-AMi) spectra computed by a model of human auditory processing (Thoret et al, 2020; Varnet et al, 2017) for a corpus of five natural soundscapes recorded in distinct terrestrial biomes on different continents at dawn or early morning: a boreal forest, a tropical forest, a temperate forest, a desert, and a savannah. These 2D-AMi spectra were obtained by passing the recordings through two successive filterbanks (see Varnet et al, 2017, and Thoret et al, 2020, for more details) simulating the spectro-temporal analysis performed by the human auditory system.…”

“…Because of the large size of the databases collected by ecoacousticians, it is now possible to train machine-learning algorithms to assess the best performance for the behavioral tasks under study, and subsequently evaluate how much and what information is lost, missed or ignored by (real) human observers in these tasks because of internal noise, limited attentional and memory capacities and/or suboptimal decision strategies. As an example, testing architectures where machine-learning algorithms are driven by the output of peripheral or mid-level auditory processing models (e.g., Apoux et al, 2023; Thoret et al, 2020) should prove useful in establishing the importance of low-level sensory cues and the contribution of more central mechanisms in natural soundscape perception. This model-driven approach offers a unique opportunity to map in greater depth the information-processing architecture of the human auditory system in the case of natural sounds and test for the existence of central mechanisms selectively tuned for biological sound-source detection and biodiversity assessment.…”