31 S. Dixon M. Mauch A. Anglade 2011 R. Kronland-Martinet S. Ystad K. Jensen M. Aramaki CMMR 2010 Post Symposium Proceedings Berlin / Heidelberg Springer 0 Amélie Anglade Emmanouil Benetos Matthias Mauch Simon Dixon 2010 Journal of New Music Research 39 4 349-361 2010 We present a new genre classification framework using both low-level signal-based features and high-level harmony features. A state of-the-art statistical genre classifier based on timbral features is extended using a first-order random forest containing for each genre rules derived from harmony or chord sequences. This random forest has been automatically induced, using the first-order logic induction algorithm TILDE, from a dataset, in which for each chord the degree and chord category are identified, and covering classical, jazz and pop genre classes. The audio descriptor-based genre classifier contains 206 features, covering spectral, temporal, energy, and pitch characteristics of the audio signal. The fusion of the harmony-based classifier with the extracted feature vectors is tested on three-genre subsets of the GTZAN and ISMIR04 datasets, which contain 300 and 448 recordings, respectively. Machine learning classifiers were tested using 5x5-fold cross-validation and feature selection. Results indicate that the proposed harmony-based rules combined with the timbral descriptor-based genre classification system lead to improved genre classification rates. 3 Amelie Anglade Rafael Ramirez Simon Dixon 2009 Proceedings of the 10th International Conference on Music Information Retrieval (ISMIR 2009) Kobe, Japan 669-674 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. 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 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 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 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.