Date of Publication

8-13-2011

Document Type

Master's Thesis

Degree Name

Master of Science in Computer Science

Subject Categories

Computer Sciences

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Nelson Marcos

Defense Panel Chair

Nelson Marcos

Defense Panel Member

Gregory Cu
Jocelyn Co

Abstract/Summary

Written music or music transcriptions are useful mediums of music for learning and sharing, which usually come in the form of musical score, which are generally the standard transcription format, and the tablature format, which are focused on the guitar. Creating music transcription however can be quite a tedious pro- cess. Though various systems exist to aid in creating transcribed music, almost all generate musical scores, lacking a focus for the guitarists community. This re- search aims to develop a system that will help in automatically generating guitar tablatures and musical scores based on musical audio data. Information gathered from the audio consist of pitch, onsets and durations, chords, and beat and tempo. Major issues that were encountered during the research were harmonics for pitch detection, thresholding for onset detection, chord distinction, similar chord struc- tures for chord labeling, and the subjective quality of tempo. Results are generally acceptable, performed on a data set that contains 22 files with varying elements. 70% accuracy was gathered from pitch detection, 60% accuracy from onset de- tection, 86% accuracy for chord distinction, 85% accuracy for chord labeling, and 81% accuracy for beat and tempo.

Abstract Format

html

Language

English

Format

Print

Accession Number

TG05023

Shelf Location

Archives, The Learning Commons, 12F Henry Sy Sr. Hall

Physical Description

x, 98 leaves ; 28 cm. + 1 computer optical disc.

Keywords

Music; Guitar; Transcription

Upload Full Text

wf_yes

Embargo Period

2-22-2022

Download

Share

COinS