0 Matthias Mauch Simon Dixon 9998 IEEE Transactions on Audio, Speech, and Language Processing Chord labels provide a concise description of musical harmony. In pop and jazz music, a sequence of chord labels is often the only written record of a song, and forms the basis of so-called lead sheets. We devise a fully automatic method to simultaneously estimate from an audio waveform the chord sequence including bass notes, the metric positions of chords, and the key. The core of the method is a 6-layered dynamic Bayesian network, in which the four hidden source layers jointly model metric position, key, chord, and bass pitch class, while the two observed layers model low-level audio features corresponding to bass and treble tonal content. Using 109 different chords our method provides substantially more harmonic detail than previous approaches while maintaining a high level of accuracy. We show that with 71% correctly classified chords our method significantly exceeds the state of the art when tested against manually annotated ground truth transcriptions on the 176 audio tracks from the MIREX 2008 Chord Detection Task. We introduce a measure of segmentation quality and show that bass and meter modelling are especially beneficial for obtaining the correct level of granularity. 3 Amelie Anglade Rafael Ramirez Simon Dixon 2009 Proceedings of the 10th International Conference on Music Information Retrieval (ISMIR 2009) Kobe, Japan October, 2009 We present an automatic genre classification technique making use of frequent chord sequences that can be applied on symbolic as well as audio data. We adopt a first-order logic representation of harmony and musical genres: pieces of music are represented as lists of chords and musical genres are seen as context-free definite clause grammars using subsequences of these chord lists. To induce the contextfree definite clause grammars characterising the genres we use a first-order logic decision tree induction algorithm. We report on the adaptation of this classification framework to audio data using an automatic chord transcription algorithm. We also introduce a high-level harmony representation scheme which describes the chords in term of both their degrees and chord categories. When compared to another high-level harmony representation scheme used in a previous study, it obtains better classification accuracies and shorter run times. We test this framework on 856 audio files synthesized from Band in a Box files and covering 3 main genres, and 9 subgenres. We perform 3-way and 2-way classification tasks on these audio files and obtain good classification results: between 67% and 79% accuracy for the 2-way classification tasks and between 58% and 72% accuracy for the 3-way classification tasks. 3 Amelie Anglade Rafael Ramirez Simon Dixon 2009 Proceedings of the 6th Sound and Music Computing Conference (SMC 2009) Porto, Portugal 309-314 July, 2009 We present an approach for the automatic extraction of transparent classification models of musical genres based on harmony. To allow for human-readable classification models we adopt a first-order logic representation of harmony and musical genres: pieces of music are represented as lists of chords and musical genres are seen as context-free definite clause grammars using subsequences of these chord lists. To induce the context-free definite clause grammars characterising the genres we use a first-order logic decision tree induction algorithm, Tilde. We test this technique on 856 Band in a Box files representing academic, jazz and popular music. We perform 2-class and 3-class classification tasks on this dataset and obtain good classification results: around 66% accuracy for the 3-class problem and between 72% and 86% accuracy for the 2-class problems. A preliminary analysis of the most common rules extracted from the decision tree models built during these experiments reveals a list of interesting and/or well-known jazz, academic and popular music harmony patterns. 3 Magas, M. Proutskova, P 2009 ECDL, WEMIS Corfu 27/09/2009 3 Tidhar, D. Fazekas, G. Kolozali, S. Sandler, M. 2009 10th International Conference on Music Information Retrieval, Kobe, Japan 26/10/2009 We describe the process of collecting, organising and publishing a large set of music similarity features produced by the SoundBite playlist generator tool. These data can be a valuable asset in the development and evaluation of new Music Information Retrieval algorithms. They can also be used in Web-based music search and retrieval applications. For this reason, we make a database of features available on the Semantic Web via a SPARQL end-point, which can be used in Linked Data services. We provide examples of using the data in a research tool, as well as in a simple web application which responds to audio queries and finds a set of similar tracks in our database. 3 Kurt Jacobson Mark Sandler 2009 Web Science Athens, Greece WSRI 1 1 17/03/2009 In an effort to move towards intuitive visual interfaces for faceted browsing of structured data about music, we develop a visualization technique called k-pie}. Derived from a network visualization technique know as $k$-cores decomposition, k-pie layout accounts for the semantic labels or `colors' associated with each vertex. Vertices of a graph are arranged in a 2 dimensional circle where `slices' in the circle correspond to a specific vertex label and the most connected vertices are found in the center of the visualization. We describe the k-pie algorithm and demonstrate how it can be useful in the context of Semantic Web technologies. http://journal.webscience.org/110/ 3 Fazekas, G. Sandler, M. 2009 Unlocking Audio 2 - Connecting with Listeners London, UK 16/03/2009 3 Fazekas, G. Cannam, C. Sandler, M. 2009 IEEE/ACM Joint Conference on Digital Libraries JCDL’09 Workshop on Integrating Digital Library Content with Computational Tools and Services Austin, Texas, USA 15/06/2009 3 Fazekas, G. Sandler, M. 2009 126th Convention of the AES Munich, Germany 07/05/2009 10 Fazekas, G. Cannam, C. Sandler, M. 2009 C4DM White Paper 04/2009 We describe the construction of SAWA a simple Web-based system for automated audio analysis. This system is capable of calculating an easily extended set of musically meaningful features such as beat, tempo, and key estimates from uploaded audio files, returning the results as rich RDF data suitable for interlinking on the Semantic Web. Unlike existing systems, our application is built on open and reusable components and provides an example of quick and straightforward development. 3 Fazekas, G. Sandler, M. 2009 12th International Conference on Digital Audio Effects, Como, Italy 01/09/2009 This paper discusses architectural aspects of a software library for unified metadata management in audio processing applications. The data incorporates editorial, production, acoustical and musicological features for a variety of use cases, ranging from adaptive audio effects to alternative metadata based visualisation. Our system is designed to capture information, prescribed by modular ontology schema. This advocates the development of intelligent user interfaces and advanced media workflows in music production environments. In an effort to reach these goals, we argue for the need of modularity and interoperable semantics in representing information. We discuss the advantages of extensible Semantic Web ontologies as opposed to using specialised but disharmonious metadata formats. Concepts and techniques permitting seamless integration with existing audio production software are described in detail. 31 Amelie Anglade Simon Dixon 2009 London Hopper Colloquium The explosion of the size of personal and commercial music collections has left both content providers and customers with a common dif�culty: organising their huge musical libraries in such a way that each song can be easily retrieved, recommended and included in a playlist with similar songs. Because classifying large amounts of data is expensive and/or time-consuming, people are gaining interest in the automatic characterisation of songs. We present the �rst step towards a framework able to automatically induce rules characterising songs by various musical phenomena (e.g. rhythm, harmony, structure, etc.). For this study we are interested in the automatic extraction of harmony patterns. We analyse manually annotated chord data available in RDF and interlinked with web identi�ers which themselves give access to a detailed description of the chords. We pre-process these data to obtain high-level information before passing them to an Inductive Logic Programming software which extracts the harmony rules underlying them. This framework is tested over the Real Book (jazz) songs and the Beatles' (pop) music. It generates a total over several experiments of 12,450 harmony rules characterising and differentiating these two datasets. An analysis of the most common rules reveals a list of well-known pop and jazz patterns. This poster won the first prize of the poster competition (open to PhD students and postdoctoral researchers). 3 Magas, M. Sewart, R. Fields, B. 2009 ACM SIGGRAPH 10 Matthias Mauch Chris Cannam Matthew Davies Christopher Harte Sefki Kolozali Dan Tidhar M. Sandler 2009 10th International Conference on Music Information Retrieval Late-Breaking Session, Kobe, Japan 3 Matthias Mauch Katy Noland Simon Dixon 2009 10th International Conference on Music Information Retrieval, Kobe, Japan Chord extraction from audio is a well-established music computing task, and many valid approaches have been presented in recent years that use different chord templates, smoothing techniques and musical context models. The present work shows that additional exploitation of the repetitive structure of songs can enhance chord extraction, by combining chroma information from multiple occurrences of the same segment type. To justify this claim we modify an existing chord labelling method, providing it with manual or automatic segment labels, and compare chord extraction results on a collection of 125 songs to baseline methods without segmentation information. Our method results in consistent and more readily readable chord labels and provides a statistically significant boost in label accuracy. 0 Katy Noland Mark Sandler 2009 Computer Music Journal 33 3 Amélie Anglade Simon Dixon 2008 Proceedings of the 9th International Conference on Music Information Retrieval (ISMIR 2008) Philadelphia, Pennsylvania, USA 63-68 September, 2008 We present an approach for the automatic characterisation of the harmony of song sets making use of relational induction of logical rules. We analyse manually annotated chord data available in RDF and interlinked with web identifiers for chords which themselves give access to the root, bass, component intervals of the chords. We pre-process these data to obtain high-level information such as chord category, degree and intervals between chords before passing them to an Inductive Logic Programming software which extracts the harmony rules underlying them. This framework is tested over the Beatles songs and the Real Book songs. It generates a total over several experiments of 12,450 harmony rules characterising and differentiating the Real Book (jazz) songs and the Beatles’ (pop) music. Encouragingly, a preliminary analysis of the most common rules reveals a list of well-known pop and jazz patterns that could be completed by a more in depth analysis of the other rules. 3 Kurt Jacobson Matthew Davies Mark Sandler 2008 125th AES Convention San Francisco, CA, USA October, 2008 In large audio collections, it is common to store audio content with perceptual encoding. However, encoding parameters may vary from collection to collection or even within a collection - using different bit rates, sample rates, codecs, etc. We evaluate the effect of various audio encodings on the onset detection task. We show that audio-based onset detection methods are surprisingly robust in the presence of MP3 encoded audio. Statistically significant changes in onset detection accuracy only occur at bit-rates lower than 32kbps. 3 Kurt Jacobson Ben Fields Mark Sandler Michael Casey 2008 124th AES Convention Amsterdam, Netherlands May, 2008 In large audio collections, it is common to store audio content using perceptual encoding. However, encoding parameters may vary from collection to collection or even within a collection - using different bit rates, sample rates, codecs, etc. We evaluate the effect of various lossy audio encodings on the application of audio spectrum projection features to the automatic genre classification tasks. We show that decreases in mean classification accuracy, while small, are statistically significant for bit-rates of 96kbps or lower. Also, a heterogeneous collection of audio encodings has statistically significant decreases in mean classification accuracy compared to a pure PCM collection. 3 Kurt Jacobson Z. Huang Mark Sandler 2008 NetSci 2008 Norwich, UK 79-80 June, 2008 We are interested in using online social networks to automatically determine relationships between musicians and artists. We hope to leverage such information for computational musicology studies and for designing new music recommendation systems. Myspace has become the de-facto standard for web-based music artist promotion. It is estimated there are around 7 million artist pages on Myspace. These pages typically include some media (streaming audio) and a list of “friends� specifying social connections. This combination of user-authored media and a user-specified social network provides a unique data set that is unprecedented in scope and scale. We sample a portion of the Myspace artist network – only including artist pages in our sample (pages that include user-authored audio files). We also collect audio data from these pages. We show this network conforms in many respects to the topologies expected in social networks. A variation on the concept of assortativity is used to examine the network structure in the context of musical genre. Community structure is also evaluated with respect to musical genre. Finally, audio-based analysis is used to as a means of agglomerative community detection. The network statistics for our sample are summarized in Table 1. #nodes #edges ave degree ave shrt pth diameter clstr coeff undirected 15478 91326 11.80 4.47 9 .219 directed 15478 120487 15.57 6.42 16 - Similar values are commonly reported in social networks [Costa 2007]. However our sample size is insufficient to assume these values are indicative of the entire network. The cumulative degree distributions for the network sample suggest close to a scale-free topology. However, the power-law fit breaks down for high and low values of degree. Similar “broad-scale� degree distributions have been reported for citation networks and movie actor networks [Amaral 2000]. Assortativity We use the concept of genre to evaluate how the network structure relates to music. Generally, an artist or a song is associated with one or more musical genres (i.e. rock, pop, rap, etc.). On Myspace, the artist-user is given the option to specify a genre association. The result is each artist page is associated with between 0 and 3 genre labels selected from a static set of 119 genre labels. We are interested in the network assortativity with respect to genre – if there is a high degree of assortative mixing with respect to genre, this suggests the network structure could be meaningful in the context of musicology and music recommendation. However, we are confronted with the unique problem of multiple labels – each node is associated with between 0 and 3 node types. We propose two minor modifications to the assortativity calculation proposed by [Newman 2003]. In one method, we simply truncate the list of genre labels so each node is only associated with one label. This results in a value of r=0.35. In the second method, we preserve all genre labels and consider two nodes to be of the same type if they share one or more genre labels. This results in a value of r=0.78. Both methods suggest some level of assortative mixing, although on nearly opposite ends of the spectrum. It is also shown that the number of shared genre labels between artist pairs decreases as geodesic distance between artists increases. These results suggest that the structure of this musician network is meaningful in the context of music-related studies. 3 Amélie Anglade Simon Dixon 2008 Proceedings of the International Workshop on Machine Learning and Music (MML2008 held in conjunction with ICML/COLT/UAI 2008) Helsinki, Finland July, 2008 We present a logic-based framework using a relational description of musical data and logical inference for automatic characterisation of music. It is intended to be an alternative to the bag-of-frames approach for classification tasks but is also suitable for retrieval and musical knowledge discovery. We present the first results obtained with such a system using Inductive Logic Programming as inference method to characterise the Beatles and Real Book harmony. We conclude with a discussion of the knowledge representation problems we faced during these first tests. 3 B. Fields K. Jacobson M. Casey M. Sandler 2008 Proc. of ICMC August music networks 3 Matthias Mauch Simon Dixon 2008 ISMIR 2008 Conference Proceedings, Philadelphia, USA 2008 3 Mauch, M Müllensiefen, D. Dixon, S, Wiggins, G. 2008 Proceedings of the 10th International Conference on Music Perception and Cognition, Sapporo, Japan 2008 3 Stewart, Rebecca Levy, Mark Sandler, Mark 2008 DAFX 2008 Previous interfaces for large collections of music have used spatial audio to enhance the presentation of a visual interface or to add a mode of interaction. An interface using only audio information is presented here as a means to explore a large music collection in a two or three-dimensional space. By taking advantage of Ambisonics and binaural technology, the application presented here can scale to large collections, have flexible playback requirements, and can be optimized for slower computers. User evaluation reveals issues in creating an intuitive mapping between between user movements in physical space and virtual movement through the collection, but the novel presentation of the music collection has positive feedback and warrants further development. 0 Casey, M. A. Rhodes, C. Slaney, M. 2008 IEEE Transactions on Audio, Speech and Language Processing 16 1015–1028 0 Casey, M. A. Veltkap, R. Goto, M. Leman, M. Rhodes, C. Slaney, M. 2008 Proceedings of the IEEE 96 668–696 3 Magas, M. Casey, M. Rhodes, C. 2008 ACM SIGGRAPH New Tech Demo 3 Fields, B. Jacobson, K. Rhodes, C. Casey, M. 2008 International Symposium on Music Information Retrieval 559–564 0 Kurt Jacobson Ben Fields Mark Sandler 2008 Proc. of ISMIR sampling networks http://ismir2008.ismir.net/papers/ISMIR2008_118.pdf 3 Gyorgy Fazekas Yves Raimond Mark Sandler 2008 125th Convention of the AES, San Francisco, USA 31 Yves Raimond Christopher Sutton Mark Sandler 2008 Linked Data on the Web workshop matching music 3 K. Jacobson M. Sandler 2008 Proc. of CMMR 306-314 3 Amélie Anglade Marco Tiemann Fabio Vignoli 2007 Proceedings of the 8th International Conference on Music Information Retrieval (ISMIR 2007) Vienna, Austria 95-100 September, 2007 We present an approach to automatically create virtual communities of users with similar music tastes. Our goal is to create personalized music channels for these communities in a distributed way, so that they can for example be used in peer-to-peer networks. To find suitable techniques for creating these communities we analyze graphs created from real-world recommender datasets and identify specific properties of these datasets. Based on these properties we select and evaluate different graph-based community-extraction techniques. We select a technique that exploits identified properties to create clusters of music listeners. We validate the suitability of this technique using a music dataset and a large movie dataset. On a graph of 6,040 peers, the selected technique assigns at least 85% of the peers to optimal communities, and obtains a mean classification error of less than 0.05 over the remaining peers that are not assigned to the best community. 3 Amélie Anglade Marco Tiemann Fabio Vignoli 2007 Proceedings of the 1st ACM International Conference on Recommender Systems (RecSys 2007) Minneapolis, Minnesota, USA ACM 41-48 October, 2008 This article presents an approach to automatically create virtual communities of users with similar music preferences in a distributed system. Our goal is to create personalized music channels for these communities using the content shared by its members in peer-to-peer networks for each community. To extract these communities a complex network theoretic approach is chosen. A fully connected graph of users is created using epidemic protocols. We show that the created graph sufficiently converges to a graph created with a centralized algorithm after a small number of protocol iterations. To find suitable techniques for creating user communities, we analyze graphs created from real-world recommender datasets and identify specific properties of these datasets. Based on these properties, different graph-based community-extraction techniques are chosen and evaluated. We select a technique that exploits identified properties to create clusters of music listeners. The suitability of this technique is validated using a music dataset and two large movie datasets. On a graph of 6,040 peers, the selected technique assigns at least 85% of the peers to optimal communities, and obtains a mean classication error of less than 0.05% over the remaining peers that are not assigned to the best community. 3 Kurt Jacobson Matthew Davies Mark Sandler 2007 123rd AES Convention New York, USA October, 2007 Music information retrieval encompasses a complex and diverse set of problems. Some recent work has focused on automatic textual annotation of audio data, paralleling work in image retrieval. Here we take a narrower approach to the automatic textual annotation of music signals and focus on rhythmic style. Training data for rhythmic styles are derived from simple, precisely labeled drum loops intended for content creation. These loops are already textually annotated with the rhythmic style they represent. The training loops are then compared against a database of music content to apply textual annotations of rhythmic style to unheard music signals. Three distinct methods of rhythmic analysis are explored. These methods are tested on a small collection of electronic dance music resulting in a labeling accuracy of 73\%. 3 Ben Fields Micheal Casey 2007 Proc. Audio Engineering Society 123rd Int. Conv. New York, NY, USA October, 2007 audio classification, music similarity 3 Benjamin Fields 2007 Proc. Audio Engineering Society 122nd Int. Conv. Vienna, Austria Audio Engineering Society May, 2007 music classification 3 Rhodes, C. Casey, M. 2007 International Symposium on Music Information Retrieval 41–46 3 Matthias Mauch Simon Dixon Christopher Harte Michael Casey Benjamin Fields 2007 8th International Conference on Music Information Retrieval, ISMIR 2007 Vienna, Austria 255-258 Modern collections of symbolic and audio music content provide unprecedented possibilities for musicological research, but traditional qualitative evaluation methods cannot realistically cope with such amounts of data. We are interested in harmonic analysis and propose key-independent chord idioms derived from a bottom-up analysis of musical data as a new subject of musicological interest. In order to motivate future research on audio chord idioms and on probabilistic models of harmony we perform a quantitative study of chord progressions in two popular music collections. In particular, we extract common subsequences of chord classes from symbolic data, independent of key and context, and order them by frequency of occurrence, thus enabling us to identify chord idioms. We make musicological observations on selected chord idioms from the collections. 3 M. Levy K. Noland M. Sandler 2007 Proceedings of the 2007 {IEEE} International Conference on Acoustics, Speech and Signal Processing ({ICASSP}) Honolulu, Hawai'i 3 Katy Noland Mark Sandler 2007 Proceedings of AES 122nd Convention Vienna 31 Y Raimond S Abdallah M Sandler F Giasson 2007 music_ontology 3 M. Levy M. Sandler M. Casey 2006 Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on 5 V-V 1-4244-0469-X musical_structure A method for segmenting musical audio with a hierarchical timbre model is introduced. New evidence is presented to show that music segmentation can be recast as clustering of timbre features, and a new clustering algorithm is described. A prototype thumbnail-generating application is described and evaluated. Experimental results are given, including comparison of machine and human segmentations http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1661200 3 K. Jacobson 2006 Proc. of ISMIR music similarity, feature extraction 3 Katy Noland Mark Sandler 2006 Proceedings of the 7th International Conference on Music Information Retrieval Victoria 3 Samer Abdallah Katy Noland Mark Sandler Michael Casey Christophe Rhodes 2005 Proceedings of the 6th International Conference on Music Information Retrieval London