Meta Popular Music Corpus Project

Popular Music Corpus Project connects existing symbolic datasets of popular music into a searchable, web-based database. Currently, publicly available popular music datasets are in many different formats, ranging from an excel spreadsheet to .musicXML, .krn, or even a text file that must be compiled through a C program. Many music scholars without a programming background cannot access these datasets, which prevents further scholarly discussion. Furthermore, relevant metadata of each songs—artist name, title, release date, producer names—are often missing, or inconsistently spelled. Harmonic data in these datasets are often encoded differently, which can result in conflicting analysis of the same repertoire. Such conflicting also prevents comparing or searching existing datasets efficiently.

UROPs worked with Jinny Park on automation of spell-checking metadata and updating missing metadata, querying MusicBrainz unique ID for each song based on artist name and song title. This project aims to culminate in an interactive website, where anyone can browse the database with a user-friendly interface, along with a public API that researchers can use.