Long-duration, false-colour spectrograms for detecting species in large audio data-sets

Abstract

Long-duration recordings of the natural environment have many advantages in passive monitoring of animal diversity. Technological advances now enable the collection of far more audio than can be listened to, necessitating the development of scalable approaches for distinguishing signal from noise. Computational methods, using automated species recognisers, have improved in accuracy but require considerable coding expertise. The content of environmental recordings is unconstrained, and the creation of labelled datasets required for machine learning purposes is a time-consuming, expensive enterprise. Here, we describe a visual approach to the analysis of environmental recordings using long-duration false-colour (LDFC) spectrograms, prepared from combinations of spectral indices. The technique was originally developed to visualize 24-hour “soundscapes.” A soundscape is an ecoacoustics concept that encompasses the totality of sound in an ecosystem. We describe three case studies to demonstrate how LDFC spectrograms can be used, not only to study soundscapes, but also to monitor individual species within them. In the first case, LDFC spectrograms help to solve a “needle in the haystack” problem—to locate vocalisations of the furtive Lewin’s Rail (Tasmanian), Lewinia pectoralis brachipus. We extend the technique by using a machine learning method to scan multiple days of LDFC spectrograms. In the second case study, we demonstrate that frog choruses are easily identified in LDFC spectrograms because of their extended time-scale. Although calls of individual frogs are lost in the cacophony of sound, spectral indices can distinguish different chorus characteristics. Third, we demonstrate that the method can be extended to the detection of bat echolocation calls. By converting complex acoustic data into readily interpretable images, our practical approach bridges the gap between bioacoustics and ecoacoustics, encompassing temporal scales across three orders of magnitude. Using the one methodology, it is possible to monitor entire soundscapes and individual species within those soundscapes.

Spatio‐temporal heterogeneity in river sounds: Disentangling micro‐ and macro‐variation in a chain of waterholes

“…Just as species richness, diversity, and Shannon entropy are single numerical values thought to measure relevant attributes of an ecosystem; acoustic richness, diversity, and entropy of a recording can also be calculated to measure relevant attributes of soundscapes and ecosystems (Depraetere et al, ; Villanueva‐Rivera, Pijanowski, Doucette, & Pekin, ). Although these indices forego species identification and are designed to quantify specific attributes of the soundscape (Farina & Gage, ), they can describe species‐specific patterns if a species dominates a soundscape or a frequency band (Indraswari et al, ; Linke, Decker, Gifford, & Desjonquères, ; Towsey et al, ). These advances and other major advantages make PAM a viable option in freshwaters.…”

Assessing the value of acoustic indices to distinguish species and quantify activity: A case study using frogs

“…The three spectral indices are used to create LDFC spectrograms, a tool that facilitates rapid ecological analysis on broad temporal scales, such as visual identification and detection of target species call (Dema, Zhang, et al., ; Towsey, Zhang, et al., ), and observation of activity levels over long periods (Dema, Cappadonna, et al., ). However, most work on species level for anuran species has been qualitative (Towsey, Znidersic, et al., ); this has prompted us to conduct more quantitative work on the relationships between acoustic indices, species identity and call features. Spectral indices clearly generated species‐specific values, distinguishing U. inundata , L. bicolor , and L. rothii in the soundscape.…”

“…Litoria rothii has a multiple‐pulsed call in the form of stacked harmonics, with a narrow frequency bandwidth (Figure ). In addition, these three species called consistently and were visually distinct in the LDFC spectrogram images (Towsey, Znidersic, et al., ). Other species identified during the sample period were Litoria nasuta, Litoria caerulea, Platyplectrum ornatum , and Uperoleia lithomoda .…”

“…In this study, we aim to retain this information and reduce analysis time by using the LDFC spectrogram approach (Towsey, Zhang, et al., ) to create three indices that serve as different mathematical filters for interpreting the soundscape (Towsey, Truskinger, & Roe, ). It has been demonstrated that the ACI, the temporal ENT, and the EVN are generally the most useful acoustic indices for this purpose (Towsey, Wimmer, et al., ; Towsey, Znidersic, et al., ). A single index reveals limited information about the structure of biological activity and other acoustic events in a soundscape.…”

“…Furthermore, selected acoustic indices can be combined to create long‐duration, false‐colour (LDFC) spectrograms. The LDFC spectrogram approach is a soundscape visualisation technique that is useful for visually detecting species‐specific signals in various taxa, including anurans (Dema, Zhang, et al., ; Towsey, Znidersic, et al., ), by producing a single, 24‐hr spectrogram image on a computer screen while retaining detailed soundscape information (Towsey, Zhang, et al., ). This capacity to visualise and interpret acoustic data recorded over long periods of time demonstrates that acoustic indices and LDFC spectrograms are powerful tools for analysing ecoacoustic data.…”