Journal of New Music Research Volume 47, Issue 3, June 2018
The Journal of New Music Research (JNMR) publishes material which increases our understanding of music and musical processes by systematic, scientific and technological means. Research published in the journal is innovative, empirically grounded and often, but not exclusively, uses quantitative methods. Articles are both musically relevant and scientifically rigorous, giving full technical details. No bounds are placed on the music or musical behaviours at issue: popular music, music of diverse cultures and the canon of western classical music are all within the Journal’s scope. Articles deal with theory, analysis, composition, performance, uses of music, instruments and other music technologies. The Journal was founded in 1972 with the original title Interface to reflect its interdisciplinary nature, drawing on musicology (including music theory), computer science, psychology, acoustics, philosophy, and other disciplines.
Convolution-based classification of audio and symbolic representations of music
Gissel Velarde, Carlos Cancino Chacón, David Meredith, Tillman Weyde & Maarten Grachten
We present a novel convolution-based method for classification of audio and symbolic representations of music, which we apply to classification of music by style. Pieces of music are first sampled to pitch–time representations (spectrograms or piano-rolls) and then convolved with a Gaussian filter, before being classified by a support vector machine or by k-nearest neighbours in an ensemble of classifiers. On the well-studied task of discriminating between string quartet movements by Haydn and Mozart, we obtain accuracies that equal the state of the art on two data-sets. However, in multi-class composer identification, methods specialised for classifying symbolic representations of music are more effective. We also performed experiments on symbolic representations, synthetic audio and two different recordings of The Well-Tempered Clavier by J. S. Bach to study the method’s capacity to distinguish preludes from fugues. Our experimental results show that our approach performs similarly on symbolic representations, synthetic audio and audio recordings, setting our method apart from most previous studies that have been designed for use with either audio or symbolic data, but not both.
Learning a well-formed microtonal scale: Pitch intervals and event frequencies
Yvonne Leung & Roger T. Dean
This study investigates learning interval structure and pitch occurrence frequency of a microtonal scale by two groups of musicians (one experienced in Western tonal music only, the other in several microtonal systems) and non-musicians. While musically untrained participants could rapidly learn the pitch occurrence frequency of this scale, learning microtonal pitch intervals was slow in musicians. Interestingly, microtonal musicians were the slowest in responding to deviant pitch intervals and timbre changes in microtonal melodies amongst the musicians. These results extend our recent observation of non-musicians’ ability to learn aspects of microtonal pitch intervals, suggesting that paradoxically, musicians do not adjust their learned expectations to microtonal systems as quickly as non-musicians.
Generative statistical models with self-emergent grammar of chord sequences
Hiroaki Tsushima, Eita Nakamura, Katsutoshi Itoyama & Kazuyoshi Yoshii
Generative statistical models of chord sequences play crucial roles in music processing. To capture syntactic similarities among certain chords (e.g. in C major key, between G and G7 and between F and Dm), we study hidden Markov models and probabilistic context-free grammar models with latent variables describing syntactic categories of chord symbols and their unsupervised learning techniques for inducing the latent grammar from data. Surprisingly, we find that these models often outperform conventional Markov models in predictive power, and the self-emergent categories often correspond to traditional harmonic functions. This implies the need for chord categories in harmony models from the informatics perspective.
A supervised classification approach for note tracking in polyphonic piano transcription
Jose J. Valero-Mas, Emmanouil Benetos & José M. Iñesta
In the field of Automatic Music Transcription, note tracking systems constitute a key process in the overall success of the task as they compute the expected note-level abstraction out of a frame-based pitch activation representation. Despite its relevance, note tracking is most commonly performed using a set of hand-crafted rules adjusted in a manual fashion for the data at issue. In this regard, the present work introduces an approach based on machine learning, and more precisely supervised classification, that aims at automatically inferring such policies for the case of piano music. The idea is to segment each pitch band of a frame-based pitch activation into single instances which are subsequently classified as active or non-active note events. Results using a comprehensive set of supervised classification strategies on the MAPS piano data-set report its competitiveness against other commonly considered strategies for note tracking as well as an improvement of more than in terms of F-measure when compared to the baseline considered for both frame-level and note-level evaluations.
Violin mpulse esponse ength and erceptions of cceptability
T. Lloyd, P. Gaydecki, J. Ginsborg & C. Yates
The design, execution and analysis of a double-blind listening study is described, in which participants gave preference ratings for nine versions of the same piece of music, obtained by convolving violin impulse responses, of different length, with the original piece played on an electric instrument and stored in digital form. The original impulse response, with a length of 91.4 ms was measured from a Stradivarius violin and progressively degraded by shortening its length. The participants, who were all trained musicians, were asked to record their preferences based on personal taste, not perceived measures of quality. Analysis of the data revealed a sigmoid relationship between preference and length of impulse response. Emulated music generated using short impulse responses was the least preferred, and this aversion was reasonably constant for responses shorter than 2.7 ms. However, and perhaps surprisingly, impulse responses of only 5.78 ms were deemed acceptable. Extending the length beyond this value had little effect on the attributed preferences. Between these two values, there was a steep increase in the assigned scores, suggesting a sigmoid relationship. The findings are of theoretical interest for psychoacoustics and can be applied to the development of electronic devices that emulate stringed instruments in real time.
Listener preference towards a real and emulated violin
T. Lloyd, P. Gaydecki, H. Johannsson, J. Ginsborg & C. Yates
An account is provided of a double-blind experiment in which musically trained and untrained participants listened to recordings of a real and emulated (virtual) violin and rated them according to their personal preference. Post-experiment the participants were also asked to identify which one was the real instrument. The emulated violin was developed by convolving the impulse response from an acoustic instrument with the output from an electric violin. The real violin was preferred by both groups with no significant difference between scores for each group. However, more of the trained musicians correctly identified the real violin.