Institute of Formal and Applied Linguistics

After being established in early 2024, the Prague Music Computing Group is organizing its first workshop. The workshop will take place on Friday April 19th from 10:45 to 13:15, in room S3 of the Malá Strana building.

Most importantly, we have two invited speakers from the leading Optical Music Recognition (OMR) team, part of the Pattern Recognition and Artificial Intelligence Group, University of Alicante:

 

Juan C. Martínez-Sevilla: "Towards Universal Optical Music Recognition"

Abstract. Optical Music Recognition (OMR) has become a popular technology to retrieve information present in musical scores in conjunction with the increasing improvement of Deep Learning techniques, which represent the state-of-the-art in the field. However, its effectiveness is limited to cases where only one type of notation or texture is taken into account. To be able to make robust OMR systems, we explore different methods to handle diverse notations and textures seamlessly. This involves developing new vision and language algorithms in conjunction with new symbolic formats that allow systems to learn in a unified manner. Thanks to this research we move closer towards universal optical music recognition, facilitating users to effortlessly access and interact with the vast music cultural heritage.

 

Antonio Ríos Vila: "End-to-end Optical Music Recognition beyond staff-level transcription"

Abstract. Optical Music Recognition (OMR) is a field that has progressed significantly recently, bringing accurate systems that transcribe effectively music scores into digital formats. Despite this, there are still several limitations that hinder OMR from achieving its full potential. Specifically, state of the art OMR still depends on multi-stage pipelines for performing full-page transcription, as well as it has only been demonstrated in monophonic cases, leaving behind very relevant engravings. In this talk, we dive into research that aims to break this glass ceiling for OMR, from seeking solutions to adress beyond monophony music structures to the FP-SMT, a segmentation-free end-to-end model that is able to transcribe full-page music scores.

These will be preceded by a block of lightning talks by PMCG students, both on topics related to Optical Music Recognition (OMR) and others (computational Gregorian chant research, music generation and other topics; detailed programme below). We are looking forward to seeing you there!

 

Speaker	Topic	Starts at
Jan Hajič	Introducing the Prague Music Computing Group	10:45
Emre Rasimgil	Unlimited Muzak: background music generation on a budget	10:50
Patrik Backo	Generating an endless drum kit	10:55
Šimon Libřický	Integrating symbolic processing in notation software	11:00
Reut Tal	With a Smile and a Song: collecting data on music emotion through a game	11:05
Jan Borecký	Open-world game audio navigation: diver rescue	11:10
Adam Štefunko	Parsimento! A data-driven model for basso continuo.	11:15
	break	11:20
Anna Dvořáková	Mapping Gregorian chant repertoire: communities, clusters and topic models	11:30
Vojtěch Lanz	Chant melody segmentation using Bayesian nonparametrics	11:35
Vojtěch Dvořák	Music notation layouts: look fast.	11:40
Kristýna Harvanová	Synthetic realistic sheet music images	11:45
Jiří Mayer	Optical Music Recognition directly using MusicXML	11:50
	break	12:05
Juan Carlos Martinez-Sevilla	Towards Universal Optical Music Recognition	12:15
Antonio Ríos Vila	End-to-end Optical Music Recognition beyond staff-level transcription	12:45

 

 

The Prague Music Computing Group is proud to collaborate with TAH Centre for Innovation in Technology, Arts and Humanities: https://tah-centre.org/