TESTING ACOUSTIC DETECTION OF SPECIES WITH AI | |||
| Brian Schretzmann; Westervelt Ecological Services; brians@westervelt.com; | |||
There is increasing pressure to adopt artificial intelligence (AI) and other automated tools across all facets of research, government, and business. Some AI tools are very user-friendly and have seen widespread adoption despite their potential to lead the user astray. This presentation will briefly compare two AI tools that assist in bird identification: BirdNET-Analyzer - Cornell University’s desktop PC equivalent to their mobile app Merlin - and Arbimon, an online acoustic detection platform. During the winter of 2024-2025, Westervelt Ecological Services (WES) set up acoustic monitors at a site in the Delta to compare the two platforms to see how the tools fared in detecting a specific target species: California black rail (Laterallus jamaicensis coturniculus). This lightning talk will highlight some acoustic monitor hardware pitfalls encountered, as well as the pros and cons of using Merlin/BirdNET-Analyzer. It will also give a brief overview of the powerful Arbimon platform and reveal the results of processing nearly a year of daily acoustic data to see if either system could detect the elusive black rail. | |||
|