Computer supported Play-along music

Špica, Amela (author), Brodnik, Andrej (mentor)

Score following problem is the problem of a real-time matching of musical performance to the corresponding musical score. We also want to provide an algorithm which would follow the soloist, even when (s)he makes jumps in the score while performing. Using a string representation of the score, and an augmented suffix tree data structure built on that string, we are approaching this problem as solving an incremental approximate string matching problem. This will allow error of insertion, replacement, and deletion. We proposed three different approaches to the solution and gave their comparison and evaluation. Besides, our solution supports soloist following even if (s)he jumps to a completely arbitrary part of the composition.

• Language: English
• Keywords: musical accompanist, score following, approximate string matching, suffix tree
• Work type: Master's thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2018
• PID: 20.500.12556/RUL-105224
• Publication date in RUL: 12.11.2018

Secondary language

• Language: Slovenian
• Title: Računalniško podprto glasbena spremljava

Abstract

Sledenje partituri je problem, pri katerem želimo v stvarnem času poizkati mesto v notnem zapisu, ki se najbolje ujema s trenutno izvedenimi toni glasbenika. V magistrski nalogi smo problem prevedli na problem približnega iskanja niza v besedilu. Za temeljno podatkovno strukturo smo uporabili priponsko drevo nad višinami not skladbe. Drevo smo razširili tako, da omogoča učinkovito približno iskanje niza z možnimi dodanimi, zamenjanimi ali izpuščenimi notami. Zasnovali, primerjali in ovrednotili smo tri pristope tovrstnega preiskovanja priponskega drevesa.

• Keywords: korepetitor, sledenje partituri, približno iskanje niza, priponsko drevo