It starts by extracting the key signals from your recording: every drum hit, its timing, its strength, and the overall rhythmic flow. Background noise and decay are filtered out at this stage. Then it classifies in two stages. First: drumming or silence. Second: steady groove or fill. That layered check catches what single-pass systems miss. This layered approach helps reduce the impact of silence, decay, and low-level background noise on segmentation.

No fixed-time slicing. The first cut follows what it actually hears: grooves, fills, and silence. Then it tightens the edges. Boundaries snap to natural pauses. False fragments merge. Long sections split. True silence is restored. These steps help reduce unwanted audio at segment edges, false fragments, and overly long sections, resulting in cleaner and more stable segments.

Once the audio is segmented, the system analyzes the rhythmic structure of each section, including drum patterns, accents, density, and timing relationships between hits. It then compares segments to find similarities. Related grooves group together. Repeated patterns get flagged. It also locks in the BPM for each segment, correcting common half-time or double-time misreads against the full track. Every segment leaves tagged with its groove pattern and BPM—ready to organize, edit, or export.