Stridulant: A Python toolkit for detecting hidden ecological processes through sound

Passive acoustic monitoring has recently received a surge of interest due to its potential to be completely automated, non-invasive and generate continuous information that complements traditional ‘snap-shot’ manual inventories. Coupled with deep learning approaches, vocal species such as birds and bats are now routinely studied using sound. In contrast, the application of acoustic methods to non-vocal organisms remains underdeveloped, which may be due to the lack of accessible tools for processing audio in a non-subjective and time-efficient manner. Here, we present Stridulant , an open-access Python package containing signal processing tools designed to detect sounds associated with key ecological processes. Stridulant can be used as a stand-alone signal processing tool, or as a tool to facilitate construction of training data for deep learning models. We exemplify the signal processing pipeline with three case studies including acoustic signals from both plants and non-vocal soil fauna. We outline a workflow which starts by constructing sets of target and non-target sounds. From these sets, acoustic features are extracted and used to develop feature-based scans for identifying events in audio files that match a characterizing set of features and feature values. With this pipeline we successfully analyzed stridulation-like sounds from ants ( Formica rufa ), acoustic emissions from trees ( Picea abies ), and soil bioturbation sounds produced by earthworms ( Aporrectodea caliginosa and Lumbricus rubellus ).