LARRIE listens to a live signal, recognises the patterns that recur within it, and infers what is coming next. From a single audio stream it produces timestamped event detections, recurring-pattern and structure maps, and real-time predictions for whatever sits downstream, from light to diagnosis.
Its home is the stage. LARRIE drives live autonomous light shows, reading the music as it plays and rendering synchronised lighting with no operator behind the desk. It has been used in sitting rooms and pop-up events around the world.
In development since 2020, LARRIE began as a college thesis project and has grown into a general engine for recurrence in sound. It has been used to predict heart attacks and save lives, and the same inference core can be applied to Formula 1 engine diagnostics.
It runs in two modes: entirely offline on your laptop, with no servers or internet needed, or as a cloud-served version with heavier, enhanced analysis.