Supervisor

Suitable for

Abstract

Acoustic monitoring using low-power embedded devices is increasingly valuable in environmental and biodiversity monitoring. One aspect of this work is to automatically detect and count acoustic events such as the calls of a bats, insects, birds or amphibians. This currently involves writing bespoke low-level code for each specific application. However, it is also possible to make effective detection algorithms by chaining together simple filters, threshold detectors, and finite state machines. This project will investigate the feasibility of automatically generating this code through a high-level description of this chain; either through a high-level language or through a graphical representation. The student will develop the key elements of a tool chain that will allow an ecologist or biologist to interactively design a detection algorithm, given an example of the sound to be detected, and to compile this algorithm for the AudioMoth acoustic sensor (www.openacousticdevices.info).

Prerequisites: Compilers and experience of embedded devise useful but not essential