Matching Large Music Datasets: A Path To A Cleaner Catalog

10/05/2017
3:00 pm - 3:20 pm
OCCC W315A

Objective: Guidance
Audience Level: Beginner/Intermediate
Session Type: Presentation

Normalizing digital music content can be challenging due to metadata differences across various record labels as well as some weaknesses of acoustic fingerprinting matching. Here at Pandora, we have built Matchstick, a system that matches incoming content from partners to our existing library, leveraging audio fingerprinting, metadata and performing rights analysis.

Speaker(s)

, Software Engineer, Pandora Media