Razpoznavanje enostavnega notnega zapisa

diplomsko delo

Miha Vučko (Author), Damjan Zazula (Mentor)

Abstract

Namen te diplomske naloge je izdelati program za optično razpoznavanje glasbenih simbolov na enostavnih notnih zapisih in predvajanje razbrane melodije. Naloga spada na področje optičnega razpoznavanja glasbe. Uporabljeni sta metodi Houghove transformacije in polaganja šablon. Pri razpoznavanju notnih črt smo dosegli stoodstotno uspešnost z izločanjem črt iz Houghovega prostora. S polaganjem šablon odkrivamo taktnice, ključe, vrste takta, celinke, polovinke, četrtinke ter celinske, polovinske in četrtinske pavze. Razpoznamo tudi piko za podaljšano trajanje tonov. Vse glasbene simbole smo s polaganjem šablon razpoznali z uspešnostjo nad osemdeset odstotkov,lepri določanju vrste taktov nismo bili tako uspešni. Za predvajanje razpoznane melodije smo uporabili knjižnico midi-dot-net, ki posnema zvok klavirja. Program je pisan strukturirano, tako da se lahko razširi z drugimi metodami za razpoznavanje vzorcev, razpoznavanjem večjega nabora glasbenih simbolov in drugimi predvajalniki razpoznane melodije.

Keywords

optično razpoznavanje glasbe;Houghova transformacija;polaganje šablon;

Data

Language:	Slovenian
Year of publishing:	2013
Typology:	2.11 - Undergraduate Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	[M. Vučko]
UDC:	004.8(043.2)
COBISS:	17517590
Views:	1509
Downloads:	277
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	RECOGNITION OF SIMPLE MUSICAL NOTATION
Secondary abstract:	The purpose of this diploma thesis is to build an application for optical recognition of simple musical notations and play back the recognized melodies. The thesis covers the field of optical music recognition. Methods we use are Hough transform and template matching. We achieved one hundred percent efficiency in removing the staff lines from the Hough space. We use template matching to detect bar-lines, clefs, time signatures, whole, half, and quarter notes and whole, half, and quarter rests. We also detect dotted notes. We recognized all musical symbols with above eighty percent efficiency, with only exception being time signatures. To play the recognized music, we took advantage of the midi-dot-net library which imitates the piano sound. A flexible structure of our application allows for an expansion with advanced pattern recognition methods, additional musical symbols, and different melody players.
Secondary keywords:	optical music recognition;Hough transform;template matching;
URN:	URN:SI:UM:
Type (COBISS):	Bachelor thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko
Pages:	VI, 46 str.
ID:	8727621