Show HN: An e-ink frame that hears birds and draws them as 1800s illustrations

Introducing the e-ink bird frame that employs real-time bird detection through audio, utilizing fully local AI to render the detected birds as authentic 1800s illustrations. The project, currently in its early stages, can be found at fugleramme.arnegiacomo.dev, showcasing the birds currently heard in the garden located in Bergen, Norway.

The frame is constructed using a Raspberry Pi 5, an Inky Impression 13.3 Spectra 6 e-ink panel, a microphone, and an A4 frame. The hardware implementation and operational documentation can be found at arnegiacomo.dev/fugleramme. BirdNET-Go, responsible for the bird detection, listens to the microphone and executes the classifier.

Upon detecting a bird species, the frame connects to the BirdNET-Go API, matches the species to its corresponding hand-cut illustration, and packs the illustration onto a textured paper page. The birds are displayed with larger species towards the center of the page, sized according to their body mass. An empty window displays a bare perch when no birds are detected.

The illustrations are sourced from public-domain natural-history materials, with over 800 cut-outs covering more than 400 species. These illustrations are all genuine, hand-curated from original plates, and not AI-generated. The birds depicted in the frame are primarily from Scandinavian, British, and central European regions, with plans underway to broaden the coverage to broader European and North American areas.

The project's design inspiration derives from a WWF Verdens naturfond poster by Axel Thorenfeldt, the live-frame concept from AvianVisitors on Instagram, and the bird detection system from BirdNET-Go. The e-ink panel is not mandatory, as the frame can also operate as a web kiosk on a display connected via HDMI, or be accessed from any device on the network.

The frame's user interface includes an admin page for configuration, automatic updates, and other essential settings. The complete parts list, recommendations, and alternative options can be found at the hardware section of the project's documentation. Contributions of various kinds, including fixes, documentation, and artwork, are highly welcomed and encouraged.