Hi @loveness 

Passive acoustic monitoring (PAM) involves recording and analyzing sounds from the enviroment using acoustic recorder units (ARUs). Sound sources can be biological (e.g. sounds from wildlife), geophysical (e.g. rain, waves, wind, ice breaking) or human (e.g. ship traffic, seismic surveys, machine noise, wind farms). The combination of all sounds in a specific location and at a specific time period is called a soundscape.

Many animals produce sounds, and PAM is used to monitor a wide range of taxa -mammals, birds, amphibians, fish, insects -, across diverse hábitats, including terrestrial, marine, freshwater and even subterranean environments. By recording these sounds, researchers can gather data on various aspects of animal behavior, presence, population changes. They can also monitor changes in natural habitats and detect human-made noises that might affect wildlife, such as poaching, logging, marine sesimic activities, etc.

Givent the vast amount of data collected, the analysis can be very time-consuming so machine learning tools are often used to automatically detect and classify sounds. Also sounscape analysis often use acoustic índices – numerical metrics to summarize patterns in the audio recordings – as proxies for biodiversity metrics such as especies richness.

I hope this explanation helps clarify what acoustic monitoring entails.

Vanessa has given a great overview!! 

I would also just add that AI and machine learning are becoming integral parts of acoustic monitoring workflows. Given how small, cheap, and power-intensive the hardware (recorders) have become over the past few years, we are now able to gather unprecedented levels of recordings (in the millions of minutes!) that would be impossible to manually go and identify which species are calling. 

So, we can train AI models to do those species detection and classification steps for us in a much more automated way. We give the model a training dataset (a bunch of positives examples of species calls you want the model to find, as well as negative examples where the species is not calling), and it can learn to identify species. This enables us to process acoustic recordings much more quickly and be able to efficiently see which species are detected from which recorders and when! 

I love your learning spirit Loveness!! Keep pushing it.

Hi Loveness, I have some slides from a training course on the basics of acoustic monitoring that I can send you - ping me an email and I'll follow up!