The listening
forest.

Real idea. Playable metaphor.

This experience teaches one real AI use case: identifying birds from sound. It is independent of Cornell and Merlin. No audio is recorded or sent anywhere.

The forest, bird, sounds, spectrogram, candidate patterns, and result are crafted illustrations, not real recordings or model output. The pattern-matching challenge is a teaching metaphor: real machine-learning systems learn complex features from labeled examples, rather than simply matching these drawn lines. The six-second capture compresses a real recording session.

Explore the real technology

• Cornell: Merlin Sound ID project overview explains expert annotations, sound recordings, and model development.
• Cornell: How Sound ID works describes real-time suggestions, spectrograms, and comparisons with reference recordings.
• Cornell: Sound ID best practices recommends independent verification, quiet recording, and at least 30 seconds of recording.

All story text is captioned. Sound is optional. Turn motion off for a stiller experience. Move with ← / → or A / D; hold Space for the current action. Touch users can hold the on-screen controls. In the comparison chapter, choose with 1, 2, or 3. Escape closes this panel.