A Raspberry Pi bird frame that listens before it draws

A Norwegian developer has combined several open-source components into fugleramme, a picture frame that identifies nearby birds by their calls and displays them using 19th-century natural-history illustrations. Every stage of the machine-learning pipeline runs locally on a Raspberry Pi, so recorded audio never needs to leave the device.

Fugleramme is a compact Python renderer built around BirdNET-Go, which manages microphone input, classification, and detection settings. The renderer reads recent detections, composes the corresponding birds on an Inky Impression e-ink panel, and serves the same view as a web kiosk. Without an e-ink panel, it can display the kiosk over HDMI or through a browser on the local network.

A field guide that updates itself

Each refresh resembles a page from a Victorian ornithology book. The project uses hand-curated, public-domain natural-history drawings with their backgrounds removed. Detected species are arranged on a textured paper background, with larger birds placed near the centre and scaled according to real body mass.

Species lacking a prepared illustration are omitted from the composition. When no illustrated species have been detected during the configured window, the frame shows a bare perch.