Listening with Merlin: How Sound ID Works and What It Can (and Can’t) Do

Merlin Sound ID has made it possible for anyone to identify birds by sound, but how does it actually work, and how should birders use it? In this talk, I’ll give an accessible overview of how Merlin Sound ID listens to audio and turns it into species suggestions. I’ll describe the kinds of recordings that teach Merlin, how the system continues to improve, and where its strengths and limitations lie.

About Our Speaker

Grant Van Horn is a machine learning researcher at the Cornell Lab of Ornithology, where he works on the Sound ID and image-recognition models used in the Merlin Bird ID app. His work focuses on building practical tools that help people engage with nature. Grant received his PhD from Caltech, where he studied how to train and deploy large-scale machine learning models on mobile devices, work that contributed to technologies now used in iNaturalist, Seek, and Merlin. He is currently on leave from the University of Massachusetts Amherst, where he is an assistant professor of computer science. Learn more about Grant’s work here.

Date: Thursday, February 19 @ 6pm

Location: Online via Zoom

Link: https://us06web.zoom.us/j/87377145453?pwd=Jt8aYrFqyD9zTBGl09jH58J5s9gbev.1

Passcode: 052289