CENG SP-FORM1
This form should be used for all CENG 407 – 408 Senior Project Proposals. A topic can be jointly proposed by faculty, company and/or student, signed by at least one of the partners.
Part I. Project Proposer
Names (supervisor, company, student) and organizations
Advisor: Assist. Prof. Dr. Murat SARAN
Institution: MINISTRY OF CULTURE AND TOURISM
DIRECTORATE GENERAL OF STATE OPERA AND
BALLET (ŞÜKRİYE AYDEMİR, 0532 641 67 39, sukriye.aydemir@gmail.com)
Students: Çağrı GENÇER (201411208), Serenay
GÜNEŞ (201111024), Ayçanur ODAMAN (201111037)
Mobile
Signature
5548484497
(Serenay)
5536059073
(Çağrı) serenygns@gm gencercagr@g ail.com mail.com
5077308344
(Ayçanur) aycanilu@gmail
.com
Part II. Project Information, to be completed by the proposer (Faculty, Student and/or Company)
Starting Term
2 0 1 5 / 2 0 1 6
Fall ▢ Spring
Title
Printed Sheet Music Archive System Integrated with Optical Music Notation Recognition for
Directory General of State Opera and Ballet
Description (extra sheets can be added)
In this project, printed sheet music archive system integrated with optical music notation recognition will be developed for Directory General of State Opera and Ballet. First we analyze graphical properties of music notation. And then we explain optical music recognition (OMR). Optical music recognition (OMR) is the application of optical character recognition to interpret sheet music or printed scores into editable or playable form. Also, optical music recognition is a form which is symbol overlaid with image analysis. It enables musical notes transmit to digital symbol music representation. Optical music recognition systems generally have the following four stages in their recognition process.
1) Stave line identification: The position of the staves is identified, and (usually) the lines are removed, leaving the superimposed musical symbols. A new image is then constructed in which the lines are removed.
2) Musical object location: The symbols that were on and around the stave are located
3) Symbol identification: The type of each symbol is determined.
4) Semantics of music notation: The relationship between symbols is determined and the information is stored in a form that programs such as sequencers or music editors can use.
The system includes a selection of methods for identifying staves and isolating objects, methods for describing and identifying musical shapes and specifying the relationships between the shapes which is recognized.
Before optical music recognition system, a web interface is developed to reach music archive confidently. The web interface aims at convenient, easy, simple and especially securable interface for musicians. In database part, kind of music notes are categorized in terms of their relationships such as wind instruments, pulsatile instruments or stringed instruments etc. Hence, musicians can reach easily their music archive system what they want with a password.
Furthermore, in this system can be processed music analysis, music editing and music information. We try to work specifically based on accuracy of tools and their capability to adapt optical music recognition.
The system will help the musicians who work at Directory General of State Opera and Ballet to reach whole musical note documentations with digital systems clearly and safely. One innovative aspect of our recognition technique is that the staves are considered as sequences of symbols. This means that we can imagine building models of these sequences that can be integrated into the recognition process, enabling us to evaluate the validity or probability of a given musical sequence. Moreover, it is flexible to work with different music notation and symbol sets. Therefore, it checks reconciliation of musical note chiefly. Also, sometimes musician can damage original book of printed sheet music. So, this project can mainly provide fast way for musicians
Version: July 2015
CENG SP-FORM1
Novelty
Justification
Although there are optical music recognition systems in the market, there is no music sheet archive system integrated with optical music notation recognition. Most commercial tools today are non-adaptive, acting as black-boxes that do not improve their performance through usage: when a symbol is misread, it continues to be misread in the subsequent pages, even if the user corrects the results by hand on every page. With this project, musicians able to reach confidently and easily music archive system.
Complexity
One important different is that music is two-dimensional (pitch vertically and time horizontally), while text is essentially one-dimensional. Another significant complication in music is that the symbols are usually overlaid on a five-line stave, so it is difficult to isolate symbols. Also, unlike
OCR, symbols are made up of components that can be combined in many ways. For example, a stem may hold a number of note heads, and may be beamed to another stem. A further difference is that the same symbol may appear in different forms. For example, the lengths of slurs, ties, beams, and stems will depend on the context in which they are used. OMR is complex and must incorporate many rules about what can appear in musical notation.
Constraints: economics, sustainability, environment, ethics, security, health, social and political issues,
Since all printed music sheets are copyright protected, the information security is the most important constraint in this project. The archive system should not allow unauthorized people to access to the system.
Risks involved
Because of the different format types of musical notes, the system may confuse musical notes.
For example, consider which of the two musical shapes should be processed first: the beamed notes or the accidental? Logically the accidental and second note form a subgroup, but the recognition system will more naturally link the two notes as a group, since they are physically linked, and then treat the accidental in isolation.
Version: July 2015