Real-time genre classification for music digital libraries

Current Music Information Retrieval (MIR) systems focus on melody based retrieval with some tolerance for user errors in the melody specification. The system described here presents a novel method for theme retrieval: A theme is described as a list of musical events, containing melody and harmony features, which must be presented in a given order and within a given time frame. The system retrieves musical phrases that fit the description. A system of this type could serve musicians and listeners who wish to discover thematically similar phrases in music digital libraries. The prototype and underlying model have been tested on midi sequences of music by W.A. Mozart and have shown good performance results.

This paper deals with the Music/Speech classification problem, starting from a set of features extracted directly from compressed audio data. The proposed classification system is able to label audio sequences stored as compressed MPEG layer III files. Decoding and analyzing in a unique stage is a fundamental tool for audio streaming applications, such as real time classification. Moreover, the techniques described herein provide useful tools in the management (data tagging, summarization, etc.) of a digital music library. The adopted set of short-time features are computed from the spectral information available in the decoding stage. In this paper, we show that for the classification problem at hand this set of features is redundant and can be dramatically pruned. To this aim we used an optimization strategy based on principal component analysis and genetic algorithms. The results show a very interesting classification accuracy using just one short-time feature.

Metadata research for music digital libraries has traditionally focused on genre. Despite its potential for improving the ability of users to better search and browse music collections, music subject metadata is an unexplored area. The objective of this study is to expand the scope of music metadata research, in particular, by exploring music subject classification based on user interpretations of music. Furthermore, we compare this previously unexplored form of user data to lyrics at subject prediction tasks. In our experiment, we use datasets consisting of 900 songs annotated with user interpretations. To determine the significance of performance differences between the two sources, we applied Friedman's ANOVA test on the classification accuracies. The results show that user-generated interpretations are significantly more useful than lyrics as classification features (p < 0.05). The findings support the possibility of exploiting various existing sources for subject metadata enrichment in music digital libraries.

The design and implementation of Harbin Institute of Technology--Digital Music Library (HIT-DML) is presented in this paper. Firstly, a novel framework, a music data model, and a query language are proposed as the theoretical foundation of the library. Secondly, music computing algorithms used in the library for feature extracting and matching are described. In addition, indices are introduced for both mining themes of music objects and accelerating content-based information retrieval. Finally, experimental results on the indices and the current development of the library are provided. HIT-DML is distinguished by the following points. First, it is inherently based on database systems, and combines database technologies with multimedia technologies seamlessly. Musical data are structurally stored. Second, it has a solid theoretical foundation, from framework and data model to query language. Last, it can retrieve musical information based on content against different kinds of musical instruments. The indices used, also power the library.

We continue our work on the automatic mining of user-created music reviews towards the goal of connecting user opinions to music objects in Music Digital Libraries (MDL). We demonstrate an experimental system which automatically discovered the key descriptive patterns that differentiated positive from negative reviews which helps us to better understand our successful Phase I results. Comparison to an earlier study indicates an important consistency across projects that warrants further investigation.

This paper describes work in progress on an ethnographic study of the searching/browsing techniques employed by people in the researcher’s local community. The insights provided by this type of study can inform the development of searching/browsing support for music digital libraries.

MiDiLiB is a six year research project on digital music libraries funded by the German Research Foundation (DFG) as a part of the Distributed Processing and Delivery of Digital Documents (V3D2) research initiative. MiDiLiB’s main focus is the development of content-based retrieval algorithms for both score- and waveform-based music. In this paper we give an overview of our research results, describe several prototypical systems for content-based music retrieval which have been developed during the project, and discuss applications of the presented techniques in the context of today’s and future digital music libraries.

Automatic music genre classification is critical for enhancing user experience in streaming platforms and recommendation systems. This study proposes a Convolutional Neural Network (CNN)-based approach using the GTZAN dataset, which contains ten music genres. The original 30-second audio tracks were segmented into overlapping 3-second chunks, then preprocessed and converted into three feature representations: Mel-Spectrogram, Chroma, and Spectral Contrast. CNN model consisting of four convolutional layers with increasing filters (32–256). The model was trained over 13 epochs using the Adam optimizer. The proposed model achieved 91% accuracy, outperforming previous approaches based on single-feature extraction. The integration of diverse spectral and harmonic features enabled the model to better distinguish between similar genres and improved its generalization. This method offers practical value for real-time music classification, automatic tagging, and intelligent audio indexing in music streaming services and digital libraries.

Musicologists and musicians often would like to search by keys in a digital music library. In this paper, we introduce a new key-finding algorithm that can be applied to music in both symbolic and audio formats. The algorithm, which is based on a Hidden Markov Model (HMM), provides two stages of key-finding output; the first one referring to local keys and the second one to the global key. We describe the input, the two output stages, and the parameters of the model. In particular, we describe two configurable parameters, the transition probability distributions, which are based on a matrix of neighbouring keys, and the emission probability distributions, which make use of established key profiles. We discuss the local key-finding capabilities of the algorithm, presenting an example analysis of the Prelude Op. 28 No. 20 in C minor by Chopin, showing the local key regions obtained using different key profiles. We evaluate the global key-finding capabilities of the model, using an existing dataset and six well-known key profiles as different model parameters. Since different key profiles will tend to err or misclassify in different ways and across different pieces, we train an ensemble method with the predictions from all the key profiles (6) through our model. We show that the ensemble method achieves state-of-the-art performance for major and overall keys, however, it still underperforms the state-of-the-art for minor keys.

The aim of the study is to conduct subjective tests on audio excerpt assignment to music genre and to carry out automatic classification of musical genres with the use of decision algorithms. First, the musicology background of classifying music into styles and genres is discussed. Then, an online survey is created to perform subjective tests with a group of listeners, whose task is assigning audio samples to selected music genres. Next, a set of music descriptors is proposed and all music excerpts are parametrized. For checking parameter redundancy the Principal Component Analysis (PCA) is performed. The created database containing feature vectors is then utilized for automatic music genre classification. Two classifiers, namely: Belief Networks and SMO (Sequential Minimal Optimization Algorithm) are employed for the purpose of music genre classification. The last step of this study is to compare the efficiency of the listeners classification with the automatic music genre classification system designed by the authors. Conducted tests show to what extent listeners’ assignment and the automatic classification results agree. It is also observed that very known performers are often rated without problems. Contrarily, songs of less known artists are more difficult to assign to the given genre.

The Naval Weapons Center (NWC) is currently developing automatic target classification systems for future surveillance and attack aircraft and missile seekers. Target classification has been identified as a critical operational capability which should be included on new Navy aircraft and missile developments or systems undergoing significant modifications. The objective for the Automatic Classification Infrared Ship Imagery System is to provide the following new capablities for surveillance and attack aircraft and antiship missiles: near real-time automatic classification of ships in day and night at long standoff ranges with a wide area coverage imaging infrared sensor. The sensor applies classical pattern recognition technology to automatically classify ships using Forward Looking Infrared (FLIR) images. Automatic Classification of Infrared Ship Imagery is based on the extraction of features which uniquely describe the classes of ships. These features are used in conjunction with decision rules which are established during a training phase. Conventional classification techniques require labeled samples of all expected targets, threats and non-threats for this training phase. To overcome the resulting need for the collection of an immense data base, NWC developed a Generalized Classifier which, in the training phase, requires signals only from the targets of interest, such as high value combatant threats. In the testing phase, the signals from the combatants are classified and signals from other ships, which are sufficiently different from the training data, are classified as &quot;other&quot; targets. This technique provides a considerable savings in computer processing time, in memory requirements and data collection efforts. Since sufficient IIR images of the appropriate quality and quantity were not available for investigating automatic IIR ship classification, TV images of ship models were used for an initial feasibility demonstration. The initial investigation made use of the experience gained with preprocessing and classifying ROR and ISAR data. For this reason, the most expedient method was to collapse the 2-dimensional TV ship images onto the longitudinal axis by summing the amplitude data in the vertical ship axis. The resulting 128 point 1-dimensional profiles show the silhouette of the ship and bear an obvious similarity with the radar data. Based on that observation, a 128 point Fourier transform was computed and the ten low order squared amplitudes of the complex Fourier coefficients were then used as feature vectors for the Generalized Classifier. In contrast to the radar data, the size of TV or IIR images of ships changes as a function of range. It is therefore necessary to develop feature extraction algorithms which are scale invariant. The central moments, which have scale and rotational invariant properties were therefore implemented. This method was suggested in 1962 by M. K. Hu (IRE Transactions on Information Theory). Using the moments alone resulted in unsatisfactory classification performance and indicated that edge enhancement was necessary and that the background needed to be rejected. The images were therefore processed with the Sobel nonlinear edge enhancement algorithm, which also has the desirable property that it works for images with low signal-to-noise ratios and poorly defined edges. Satisfactory results were obtained. In another experiment, the feature vector was composed of the five lower-order invariant moments and the five lower-order FFT coefficient squared magnitudes, excluding the zero frequency coefficient. This paper will describe the data base, the processing and classification techniques, discuss the results and addresses the topic of &quot;Processing of Images and Data Optical Sensors.&quot;

Notes for Notes Deborah Campana During the 2018 Music Library Association annual meeting in Portland, Oregon, the following awards were given: The MLA Citation, MLA's highest honor for distinguished service to music librarianship, was conferred upon Geraldine ("Gerry") E. Ostrove in recognition of her career-long contributions, specifically in clarifying the complex nature and application of music subject headings, as well as the separate treatment of music form and genre in subject cataloging. A vocal advocate for music and music librarianship throughout her long and distinguished career, and passionately devoted to the Music Library Association, her style, grace, eloquent and persuasive prose have graced many pages of publications, and her broad vision of the application of music subject headings and form/genre information will be influential for generations to come. Highly visible and active within various library and scholarly organizations, Ostrove has served MLA, most notably as president from 1985 to 1987; and in IAML, as vice president of its Bibliography Commission and president of the U.S. Branch. She was the founding author of the "Recent Publications in Music" column, which appeared in Fontes Artis Musicae (1989–2012), a global bibliography of music publications that many found useful for collection development purposes. MLA grants the A. Ralph Papakhian Special Achievement Award to a member who has provided extraordinary service to the profession of music librarianship in a focused time frame or in a focused area. The 2018 A. Ralph Papakhian Special Achievement Award went to Judy Tsou and Michael Colby. Diversity and inclusion are critical goals across our profession. Within music librarianship, MLA was fortunate to participate in an IMLS-funded program supporting ARL-MLA Diversity and Inclusion Initiative Scholars. As with all grants, the challenge intensifies as the program draws to a close. Judy Tsou and Michael Colby refused to let MLA's momentum in this important area dissipate. They chose to step forward personally and professionally, insisting on the priority of continuing a Diversity Scholarship program, and volunteered of themselves to ensure that there would be leadership to accomplish the task. Judy and Michael gave generously of their time, energy, personal resources, professional connections, vision, and creativity. Their commitment [End Page 64] galvanized a movement that resulted in no gap in support between the grant-funded program and MLA's own. The board wishes to recognize and thank Judy and Michael, and applauds what can be accomplished for MLA, and for the profession, through the commitment of individual members with undeterred dedication and passionate confidence in their colleagues. The Richard S. Hill Award for the best article on music librarianship or of a music-bibliographic nature was awarded to Kevin Kishimoto and Tracey Snyder for their article, "Popular Music in FRBR and RDA: Toward User-Friendly and Cataloger-Friendly Identification of Works," Cataloging & Classification Quarterly 54, no. 1 (2016): 60–86. The awards committee wrote that Kishimoto and Snyder's article about cataloging popular music and music outside of the Western Art tradition is a thorough, well-written, and often humorous discussion of how strict adherence to cataloging models can lead to decisions that do not align with user expectations, while also causing workflow inefficiencies for the cataloger. The authors give not only an overview of the topic, but do so in accessible and approachable language. By discussing and considering multiple facets of the issues at hand, the authors suggest solutions that consider cultural practices within music cataloging, fit within user expectations, and provide for more practical efficiencies for the cataloger. The Vincent H. Duckles Award for the best book-length bibliography or other research tool in music was presented to Jane Gottlieb for her book Music Library and Research Skills, 2d edition (New York; Oxford: Oxford University Press, 2016). As the Publication Awards Committee wrote: Gottlieb provides readers with a thorough and meticulously crafted guide to music libraries, research methodology, and print and electronic resources. In describing music research tools, she offers invaluable information on their history, scope, and contents. Gottlieb also provides sage advice on bibliographic style, copyright, and resources for exploring careers in music. Drawing from her extensive experience in music librarianship, Gottlieb elegantly captures the current state of the art in music...

The scenarios opened by the increasing availability, sharing and dissemination of music across the Web is pushing for fast, effective and abstract ways of organizing and retrieving music material. Automatic classification is a central activity to model most of these processes, thus its design plays a relevant role in advanced Music Information Retrieval. In this paper, we adopted a state-of-the-art machine learning algorithm, i.e. Support Vector Machines, to design an automatic classifier of music genres. In order to optimize classification accuracy, we implemented some already proposed features and engineered new ones to capture aspects of songs that have been neglected in previous studies. The classification results on two datasets suggest that our model based on very simple features reaches the state-of-art accuracy (on the ISMIR dataset) and very high performance on a music corpus collected locally.

Automatic modulation classification (AMC) is a challenging topic in the development of cognitive radio, which can sense and learn surrounding electromagnetic environments and help to make corresponding decisions. In this paper, we propose to complete the real-time AMC through constructing a lightweight neural network MobileViT driven by the clustered constellation images. Firstly, the clustered constellation images are transformed from I/Q sequences to help extract robust and discriminative features. Then the lightweight neural network called MobileViT is developed for the real-time constellation image classification. Experimental results on the public dataset RadioML 2016.10a with edge computing platform demonstrate the superiority and efficiency of MobileViT. Furthermore, the extensive ablation tests prove the robustness of the proposed method to the learning rate and batch size. To the best of our knowledge, this is the first attempt to deploy the deep learning model to complete the real-time classification of modulation schemes of received signals at the edge.

ABSTRACTIn the last couple of decades, data analytics‐based pattern classification methods for disease detection have gained much traction in healthcare research and applications. The current study builds linear programming (LP) models for detecting disease incidence. We propose sequential steps of a convex programming algorithm to construct decision boundary functions to classify patterns in disease detection data. We compare the performance of our LP‐based classifier with others (neural network, decision tree, k‐nearest‐neighbor, logistic regression, naïve‐Bayes, and support‐vector‐machine) on four datasets: two different ones for breast cancer, and one each for diabetes and diabetic retinopathy. Statistical tests reveal that the LP classifier did significantly better than the other methods in five out of eight false‐positive and false‐negative test cases. There is not a statistically significant difference in performance in the remaining three tests between the LP classifier and the best alternative method. Most importantly, the LP classifier has significantly superior performance in both diabetes detection and diabetic retinopathy data. The success of the proposed LP classifier results from avoiding “modeling noise” and “memorization of training data.” We recommend that our proposed LP classifier be among the set of classifiers for use in disease detection analytics.