Dr. Matthias Mauch from Queen Mary University of London will give a talk within statistical colloquium on Thursday, 08.11.2012 (room CDI 120, TU Dortmund)
„Teasing the Music out of Digital Data – my first five years of Music Informatics Research“
The first paper summarizes the final results of the project group „Mobile music worlds – explorative games for navigation through digital music collections“ and will be presented at the IEEE Conference on Computational Intelligence and Games (CIG) (Granada, Spain, 11-14.09.2012):
A. Jordan, D. Scheftelowitsch, J. Lahni, J. Hartwecker, M. Kuchem, M. Walter-Huber, N. Vortmeier, T. Delbruegger, Ü. Güler, I. Vatolkin, and M. Preuss – BeatTheBeat: Music-Based Procedural Content Generation In a Mobile Game
The second paper examines the possibilities to reduce the number of music training instances based on N-Gram sorting with the following removal of outliers and similar frequent instances. It will be presented at the Workshop on Knowledge Discovery, Data Mining and Machine Learning (KDML) (Dortmund, Germany, 12-14.09.2012):
I. Vatolkin, M. Preuss, and G. Rudolph – Training Set Reduction Based on 2-Gram Feature Statistics for Music Genre Recognition
MIR Toolbox plugin for AMUSE has been updated: MIR Toolbox 1.2.1 has been replaced by 1.4:
http://amuse-framework.svn.sourceforge.net/viewvc/amuse-framework/amusePluginMIRToolbox/
The results of a large study applying feature selection by SMS-EMOA for recognition of instruments in audio mixtures are available online and will be published in the special issue on evolutionary music of the journal Soft Computing – A Fusion of Foundations, Methodologies and Applications:
I. Vatolkin, M. Preuß, G. Rudolph, M. Eichhoff, C. Weihs: Multi-Objective Evolutionary Feature Selection for Instrument Recognition in Polyphonic Audio Mixtures.
Musikinformatik-Buch
Die zwei Kapitel „Foundations of Music“ und „Physics of Sound“ werden zu einem zusammen gefasst („The Musical Signal“). „Music Theory and Music Perception“ wird in „Musical Structures and their Perception“ umbenannt.
Das einführende Kapitel „Goals and Problems“ wird durch Beispiele ergänzt. Die Stichpunkte des Kapitels sind in Repository eingecheckt.
In der Datei book.tex sind Notationsbeispiele. Symbolliste wird vorne definiert, beim Hinzufügen neuer Symbole sollen die Personen den eigenen Namen dazu schreiben. Symbole, die übergreifend für mehrere Kapitel gelten, sollen mithilfe von \newcommand definiert werden.
Vektoren werden grundsätzlich als Spaltenvektoren definiert; transponierte Vektoren werden mit „T“ markiert.
Es wurden zwar keine Deadlines zum Schreiben von Kapiteln festgelegt, es wird allerdings erwartet, dass die ersten vollständigen Texte bis zu der nächsten Sitzung entstehen. Dabei kann gleichzeitig an allen Kapiteln gearbeitet werden.
Die nächste SIGMA-Sitzung
Der Link zu Umfrage für SIGMA #20: http://www.doodle.com/8p4c5d6ewpf6csnh (Die Umfrage wird Ende Juni abgeschlossen, bitte tragen Sie sich zeitnah ein!). Der Ort wird später festgelegt.
The 2-year research project ViSTA-TV funded by European Union is started on 01.06.2012. The target of the project is to apply the actual data mining methods for analysis and recommendation of TV streams.
The project group “Mobile music worlds – explorative games for navigation through digital music collections” is finished; the short summary of results is provided by a poster.
The forthcoming post-conference proceedings of GfKl 2011 will include the following publications:
Two updates providing easy extraction of new harmony-related audio features within AMUSE have been recently accomplished:
Update of Chroma Toolbox plugin
The AMUSE Chroma Toolbox plugin was updated to the version 2.0 (see http://www.mpi-inf.mpg.de/resources/MIR/chromatoolbox/ for more details on Chroma Toolbox).
The following features were added:
New Sonic Annotator NNLS plugin
This plugin enables the extraction of the following NNLS Chroma features (http://isophonics.net/nnls-chroma):
An interesting contest for music recommendation algorithms (the results will be presented at ISMIR 2012):
http://www.kaggle.com/c/msdchallenge
Since the data set is openly distributed, it can be also used for development / evaluation of methods not only for this competition.
