An Efficient Identification Methodology for Improved Access to Music Heritage Collections

A comprehensive methodology for automatic music identification is presented. The main application of the proposed approach is to provide tools to enrich and validate the descriptors of recordings digitized by a sound archive institution. Experimentation has been carried out on three different datasets, including a collection of digitized vinyl discs, although the methodology is not linked to a particular recording carrier. Automatic identification allows a music digital library to retrieve metadata about music works even if the information was incomplete or missing at the time of the acquisition. Automatic segmentation of digitized material is obtained as a byproduct of identification, allowing the music digital library to grant access to individual tracks, even if discs are digitized using a single file for a complete disc side. Results show that the approach is both efficient and effective.