Structure and segmentation

librosa.segment provides functions for structural analysis and segmentation of audio.

This includes recurrence (self-similarity) and cross-similarity matrix computation, filtering and transformations of similarity matrices, and temporal segmentation by clustering.

cross_similarity	Compute cross-similarity from one data sequence to a reference sequence.
recurrence_matrix	Compute a recurrence matrix from a data matrix.
recurrence_to_lag	Convert a recurrence matrix into a lag matrix.
lag_to_recurrence	Convert a lag matrix into a recurrence matrix.
timelag_filter	Apply a filter in the time-lag domain.
path_enhance	Multi-angle path enhancement for self- and cross-similarity matrices.
agglomerative	Bottom-up temporal segmentation.
subsegment	Sub-divide a segmentation by feature clustering.