Abstract: Swimming with the tide of deep learning, the field of music information retrieval (MIR) experiences parallel development and a sheer variety of feature-learning models has been applied to music classification and tagging tasks. Among those learning techniques, the deep convolutional neural networks (CNNs) have been widespreadly used with better performance than the traditional approach especially in music genre classification and prediction. However, regarding the music recommendation, there is a large semantic gap between the corresponding audio genres and the various aspects of a song that influence user preference. In our study, aiming to bridge the gap, we strive to construct an automatic music aesthetic annotation model with MIDI format for better comparison and measurement of the similarity between music pieces in the way of harmonic analysis. We use the matrix of qualification converted from MIDI files as input to train two different classifiers, support vector machine (SVM) and Decision Tree (DT). Experimental results in performance of a tag prediction task have shown that both learning algorithms are capable of extracting high-level properties in an end-to end manner from music information. The proposed model is helpful to learn the audience taste and then the resulting recommendations are likely to appeal to a niche consumer.
Abstract: Choosing good features is an essential part of machine learning. Recent techniques aim to automate this process. For instance, feature learning intends to learn the transformation of raw data into a useful representation to machine learning tasks. In automatic audio classification tasks, this is interesting since the audio, usually complex information, needs to be transformed into a computationally convenient input to process. Another technique tries to generate features by searching a feature space. Genetic algorithms, for instance, have being used to generate audio features by combining or modifying them. We find this approach particularly interesting and, despite the undeniable advances of feature learning approaches, we wanted to take a step forward in the use of genetic algorithms to find audio features, combining them with more conventional methods, like PCA, and inserting search control mechanisms, such as constraints over a confusion matrix. This work presents the results obtained on particular audio classification problems.
Abstract: Rapid progress in audio compression technology has contributed to the explosive growth of music available in digital form today. In a reversal of ideas, this work makes use of a recently proposed efficient audio compression scheme to develop three important applications in the context of Music Information Retrieval (MIR) for the effective manipulation of large music databases, namely automatic music recommendation (AMR), digital rights management (DRM) and audio finger-printing for song identification. The performance of these three applications has been evaluated with respect to a database of songs collected from a diverse set of genres.
Abstract: Music Information Retrieval (MIR) and modern data mining techniques are applied to identify style markers in midi music for stylometric analysis and author attribution. Over 100 attributes are extracted from a library of 2830 songs then mined using supervised learning data mining techniques. Two attributes are identified that provide high informational gain. These attributes are then used as style markers to predict authorship. Using these style markers the authors are able to correctly distinguish songs written by the Beatles from those that were not with a precision and accuracy of over 98 per cent. The identification of these style markers as well as the architecture for this research provides a foundation for future research in musical stylometry.
Abstract: Music segmentation is a key issue in music information
retrieval (MIR) as it provides an insight into the
internal structure of a composition. Structural information about
a composition can improve several tasks related to MIR such
as searching and browsing large music collections, visualizing
musical structure, lyric alignment, and music summarization.
The authors of this paper present the MTSSM framework, a twolayer
framework for the multi-track segmentation of symbolic
music. The strength of this framework lies in the combination of
existing methods for local track segmentation and the application
of global structure information spanning via multiple tracks.
The first layer of the MTSSM uses various string matching
techniques to detect the best candidate segmentations for each
track of a multi-track composition independently. The second
layer combines all single track results and determines the best
segmentation for each track in respect to the global structure of
the composition.