Melody categorization refers to the task of grouping a set of melodies into categories of similar items that originate from the same melodic contour. From a computational perspective, automatic melody categorization is of crucial importance for the automatic organization of databases, as well as for large-scale musicological studies—in particular, in the context of folk music and non-Western music traditions. We investigate methods starting from the raw audio file. For each recording contained in a collection, we extract a pitch sequence representing the main melodic line. We then estimate pairwise similarities and evaluate the discriminative power of the resulting similarity matrix with respect to ground-truth annotations. We propose novel evaluation methodologies, compare melody representations, and explore the potential of our approach in the context of two applications: interstyle and intrastyle categorization of flamenco music and tune-family recognition of folk-song recordings.

Fostered by the introduction of the Music Information Retrieval Evaluation Exchange (MIREX) competition, the number of systems that calculate symbolic melodic similarity has recently increased considerably. To understand the state of the art, we provide a comparative analysis of existing algorithms. The analysis is based on eight criteria that help to characterize the systems, highlighting strengths and weaknesses. We also propose a taxonomy that classifies algorithms based on their approach. Both taxonomy and criteria are fruitfully exploited to provide input for new, forthcoming research in the area.

In this article a number of musical features are extracted from a large musical database and these were subsequently used to build four composer-classification models. The first two models, an if–then rule set and a decision tree, result in an understanding of stylistic differences between Bach, Haydn, and Beethoven. The other two models, a logistic regression model and a support vector machine classifier, are more accurate. The probability of a piece being composed by a certain composer given by the logistic regression model is integrated into the objective function of a previously developed variable neighborhood search algorithm that can generate counterpoint. The result is a system that can generate an endless stream of contrapuntal music with composer-specific characteristics that sounds pleasing to the ear. This system is implemented as an Android app called FuX.

This article presents an offline method for aligning an audio signal to individual instrumental parts constituting a musical score. The proposed method is based on fitting multiple hidden semi-Markov models (HSMMs) to the observed audio signal. The emission probability of each state of the HSMM is described using latent harmonic allocation (LHA), a Bayesian model of a harmonic sound mixture. Each HSMM corresponds to one musical instrument’s part, and the state duration probability is conditioned on a linear dynamics system (LDS) tempo model. Variational Bayesian inference is used to jointly infer LHA, HSMM, and the LDS. We evaluate the capability of the method to align musical audio to its score, under reverberation, structural variations, and fluctuations in onset timing among different parts.

In this article, we explore the use of Bayesian networks for identifying the timbre of musical instruments. Peak spectral amplitude in ten frequency windows is extracted for each of 20 time windows to be used as features. Over a large data set of 24,000 audio examples covering the full musical range of 24 different common orchestral instruments, four different Bayesian network structures, including naive Bayes, are examined and compared with two support vector machines and a k -nearest neighbor classifier. Classification accuracy is examined by instrument, instrument family, and data set size. Bayesian networks with conditional dependencies in the time and frequency dimensions achieved 98 percent accuracy in the instrument classification task and 97 percent accuracy in the instrument family identification task. These results demonstrate a significant improvement over the previous approaches in the literature on this data set. Additionally, we tested our Bayesian approach on the widely used Iowa musical instrument data set, with similar results.

In many non-Western musical traditions, such as North Indian classical music (NICM), melodies do not conform to the major and minor modes, and they commonly use tunings that have no fixed reference (e.g., A = 440 Hz). We present a novel method for joint tonic and raag recognition in NICM from audio, based on pitch distributions. We systematically compare the accuracy of several methods using these tonal features when combined with instance-based (nearest-neighbor) and Bayesian classifiers. We find that, when compared with a standard twelve-dimensional pitch class distribution that estimates the relative frequency of each of the chromatic pitches, smoother and more continuous tonal representations offer significant performance advantages, particularly when combined with appropriate classification techniques. Best results are obtained using a kernel-density pitch distribution along with a nearest-neighbor classifier using Bhattacharyya distance, attaining a tonic error rate of 4.2 percent and raag error rate of 10.3 percent (with 21 different raag categories). These experiments suggest that tonal features based on pitch distributions are robust, reliable features that can be applied to complex melodic music.

Key changes are common in Western classical music. The precise segmentation of a music piece at instances where key changes occur allows for further analysis, like self-similarity analysis, chord recognition, and several other techniques that mainly pertain to the characterization of music content. This article examines the automatic segmentation of audio data into parts composed in different keys, using clustering on chroma-related spaces. To this end, the k -means algorithm is used and a methodology is proposed so that useful information about key changes can be derived, regardless of the number of clusters or key changes. The proposed methodology is evaluated by experimenting on the segmentation of recordings of existing compositions from the Classic-Romantic repertoire. Additional analysis is performed using artificial data sets. Specifically, the construction of artificial pieces is proposed as a means to investigate the potential of the strategy under discussion in predefined key-change scenarios that encompass different musical characteristics. For the existing compositions, we compare the results of our proposed methodology with others from the music information retrieval literature. Finally, although the proposed methodology is only capable of locating key changes and not the key identities themselves, we discuss results regarding the labeling of a composition's key in the located segments.

In this work, a probabilistic model for multiple-instrument automatic music transcription is proposed. The model extends the shift-invariant probabilistic latent component analysis method, which is used for spectrogram factorization. Proposed extensions support the use of multiple spectral templates per pitch and per instrument source, as well as a time-varying pitch contribution for each source. Thus, this method can effectively be used for multiple-instrument automatic transcription. In addition, the shift-invariant aspect of the method can be exploited for detecting tuning changes and frequency modulations, as well as for visualizing pitch content. For note tracking and smoothing, pitch-wise hidden Markov models are used. For training, pitch templates from eight orchestral instruments were extracted, covering their complete note range. The transcription system was tested on multiple-instrument polyphonic recordings from the RWC database, a Disklavier data set, and the MIREX 2007 multi-F0 data set. Results demonstrate that the proposed method outperforms leading approaches from the transcription literature, using several error metrics.

We present a method to recuperate fingerings for a given piece of violin music in order to recreate the timbre of a given audio recording of the piece. This is achieved by first analyzing an audio signal to determine the most likely sequence of two-dimensional fingerboard locations (string number and location along the string), which recovers elements of violin fingering relevant to timbre. This sequence is then used as a constraint for finding an ergonomic sequence of finger placements that satisfies both the sequence of notated pitch and the given fingerboard-location sequence. Fingerboard-location-sequence estimation is based on estimation of a hidden Markov model, each state of which represents a particular fingerboard location and emits a Gaussian mixture model of the relative strengths of harmonics. The relative strengths of harmonics are estimated from a polyphonic mixture using score-informed source segregation, and compensates for discrepancies between observed data and training data through mean normalization. Fingering estimation is based on the modeling of a cost function for a sequence of finger placements. We tailor our model to incorporate the playing practices of the violin. We evaluate the performance of the fingerboard-location estimator with a polyphonic mixture, and with recordings of a violin whose timbral characteristics differ significantly from that of the training data. We subjectively evaluate the fingering estimator and validate the effectiveness of tailoring the fingering model towards the violin.

In this article a multidimensional environment is defined to allow the exploration of musical content in a novel way by means of three-dimensional interfaces. The environment is created so that musical content can be located in a comprehensive space in which the world coordinates are related to music genres. To this end, the songs in a database are analyzed, and a description of their spectral content is obtained. These descriptions are then projected onto six vectors, previously determined, that represent six main genres defining a global space with six dimensions. These projected representations are useful to create a multidimensional world in which the relations, orientation, and motion will be readily intelligible to users.