Music Information Retrieval (MIR) is a rapidly advancing field dedicated to the scientific analysis of music across different cultures. This research explores the application of an Error-Correcting Output Codes (ECOC) framework combined with a Multilayer Perceptron (MLP) for the classification of Iranian classical music, an area that has been less explored in computational musicology. The novel approach leverages the NAVA dataset, a new standard in MIR specifically structured for this task, to address the challenge of distinguishing complex musical modes known as Dastgahs, which are integral to Iranian music. The proposed model reduces computational complexity while achieving higher classification accuracy compared to traditional deep learning models. This study demonstrates the potential of MLP-ECOC models to handle high feature dimensionality and various musical complexities effectively. This research contributes to the broader application of MIR technologies in non-Western musical contexts, aiming to enhance personalised music discovery and provide a deeper understanding of the cultural heritage embedded in music. This work sets a new standard in the field, promoting simpler, yet effective, ensemble learning techniques in music classification with a highest recorded accuracy of 96.32%.

This article presents a research platform for musicians' interaction on room stages. It is intended for stage acoustical quality and psychophysical experiments on music performance including considerations for hearing one's own and others' instruments. The platform is a real-time auralisation system; hence, its quality depends on the system's calibration and latency, among other factors. Therefore, the presented system focuses on the acoustic considerations for laboratory implementations. The calibration is implemented as a set of filters accounting for the microphone-instrument distances and the directivity factors, as well as the transducers' frequency responses. Moreover, sources of errors are characterised using both state-of-the-art information and derivations from the mathematical definition of the calibration filter. In order to compensate for hardware latency without cropping parts of the simulated impulse responses, the virtual direct sound of musicians hearing themselves is skipped from the simulation and addressed by letting the actual direct sound reach the listener through open headphones. The required latency compensation meets both the minimum distance requirement between musicians for the interactive part (i.e. hearing others) and the minimum ray path requirement for the floor reflection for the individual part (i.e. hearing oneself), which is 2 m for the implemented system. Finally, a proof of concept is provided that includes objective and subjective experiments, which give support to the feasibility of the proposed setup.

Discovering the emotional–semantic dimensions underlying music description is central to music psychology and widely applied in sonic branding practice. Academic work typically relies on dimension reduction approaches such as principal components analysis (PCA), which (i) require subjective reinterpretation of latent components, (ii) often yield uneven component importances when a fixed number of dimensions is imposed, and (iii) offer limited guidance for selecting practically manageable subsets of descriptors. Addressing this gap, we evaluate whether an existing feature-selection algorithm — Diversity-Induced Self-Representation (D-ISR; Liu et al. [2017]) — can serve as an objective and scalable alternative for identifying concise yet representative sets of emotional–semantic attributes. Using a large real-world dataset (NParticipants = 55,593; NResponses = 5,820,188; NAudioTracks = 251), we compare D-ISR and PCA within a unified experimental framework. D-ISR selects 14 core attributes from an industry-scale pool of 212 attributes and reconstructs the original 212-dimensional space with good accuracy. Direct comparison with PCA demonstrates how D-ISR provides a more balanced trade-off between interpretability, reconstruction fidelity, and the need for a practically small set of descriptors. Our findings (i) document and analyse a large-scale emotional–semantic music dataset rarely accessible in the public domain, (ii) demonstrate a principled framework for comparing feature-selection and component-extraction methods for music-descriptor research, and (iii) illustrate how large attribute sets can be reduced to flexible, task-appropriate representations of the emotional–semantic music space. This contributes a clear methodological foundation for both scientific studies of musical meaning and applied work such as sonic branding.

Traditional sound synthesis methods offer distinct advantages but often face challenges in either computational efficiency or realism. Nonlinear dynamical systems present a compelling middle ground, especially with recent advances in the numerical integration of ordinary differential equations, allowing efficient implementation while capturing the essential dynamics of acoustic instruments. Here, we explore the application of nonlinear dynamical systems, particularly the theory of normal forms, to sound synthesis, offering a novel perspective for creating virtual instruments. The theory of normal forms provides a rigorous mathematical framework for understanding how oscillations can be generated, destroyed, or how they interact in nonlinear systems, making it a powerful tool in sound design. Despite its potential, this approach remains underexplored in the context of sound synthesis. We present an overview of relevant concepts in dynamical systems such as phase space, fixed points, and bifurcations. Three specific dynamical systems are then examined to analyze their suitability for sound synthesis. Finally, we present a package for the Max programming environment for real-time synthesis and show that it is possible to develop virtual instruments based on the examples presented before.

Musical stimuli provide a valuable approach to understanding and eliciting emotions. This research aimed to evaluate the validity and reliability of a musical stimuli set, composed to elicit happiness, fear/anger, sadness, and serenity, developing a novel tool for non-verbal emotion assessment. Furthermore, a comparative analysis explored variations in assessing perceived emotions, valence, and arousal dimensions among musicians and non-musicians. A sample of 200 individuals, aged 18 to 43 years (M = 27.27; SD = 5.90), and comprising 100 musicians and 100 non-musicians, evaluated the stimuli by reporting their perceived emotions through a forced-choice identification task, and by rating valence and arousal using the Self-Assessment Manikin. Among the original 116 stimuli designed, 38 musical stimuli were selected based on item response theory and confirmatory factor analysis for developing the Musical Emotion Evaluation Test. Additionally, musicians and non-musicians performed similarly when identifying emotions and perceiving emotional valence and arousal in musical stimuli. The validation of musical stimuli to evaluate emotions contributes to increasing the reliability and validity of the assessment in the research and clinical contexts. This study also provides insights into how musical expertise influences the perception of emotions in music.

Music performance is one of the most refined forms of skilled human behaviour, combining high-level motor and cognitive demands with expressive intentions. We examined how musical features and movement expression affect postural sway in expert saxophone players. Twenty participants (nine female) performed excerpts varying in tempo, rhythmical density, articulation, and technical demands, in standing position, under movement-restricted and expressive-movement conditions. Generalised linear mixed models were used to assess the effects of these factors on centre-of-mass measurements. Results showed that participants swayed faster and travelled longer distances in music with faster tempo and increased rhythmical density, but slower and with reduced mediolateral range when performing tonguing technique (staccato). When limited to technical motion, participants still showed increased mediolateral sway during staccato passages, suggesting compensatory postural adjustments. Sway frequency was unaffected by movement condition, possibly reflecting an unconscious, task-related motor response. Performances were longer when movement was constrained, highlighting the role of body motion in temporal regulation. Our findings help understand how saxophone players accommodate technical and expressive goals, offering new insights into motor control and multisensory integration during performance.

Theories of accent production lack empirical data supporting how notes are marked as accented in performance. The present study used percussion performance as a lens to investigate how movements of the performer’s mallets, hands, and wrists contribute to marking notes as accented and how such movements may be refined via training. Experienced percussionists completed a single experimental session where they practiced an excerpt scored for multiple drums. During training, an instructor provided theoretically informed coaching prompts designed to improve performance quality and effectiveness of accent production between successive performances of the excerpt. Motion capture technology measured movements of the mallets, hands, and wrists along specific phases of the accent trajectory: the preparatory upstroke, accent downstroke, post-accent upstroke, and following-note downstroke. Analyses revealed that at post-training, the position of both mallets during the accent downstroke were higher than pre-training. Mallet velocity was also greater in post-training vs. pre-training for the accent downstroke. The average hand position was higher in post-training vs. pre-training for the preparatory upstroke. These changes in movement kinematics coincided with increased effectiveness of accent production in post- vs. pre-training performances, as evaluated by trained judges. The results were interpreted with regards to unique mechanisms that give rise to accent production in relation to mallet and upper-limb movements, and how such mechanisms can be applied towards translating theory into practice for improving accent production in percussion.

This paper proposes group delay-based schemes for automatic tonic identification in Indian and Turkish classical music. Modified group delay functions (MODGD) have been employed for the proposed task. The tonic in Indian art music is the base pitch chosen by a performer in a concert. Tonic in Turkish music is one of the stable pitches of the performance, which serves as the reference throughout the performance. In the proposed work, tonic pitch in Indian art music is estimated by analysing the drone instrument in the non-melodic audio segments (segments with no lead melody line). Two schemes have been introduced for Indian art music. While the first scheme employs melodic pitch as an attribute to identify non-melodic segments, the second relies on audio descriptors. Modified group delay functions computed from the flattened music spectrum of these frames are bin-wise summed to form a summary-MODGD-gram. The location of the peak in the tonic range of summary-MODGD-gram is mapped to the tonic pitch. The experiment is also extended to estimate the tonic in Turkish makam music. The performance is evaluated on the Indian and Turkish classical music datasets, and the results show the potential of MODGD in tonic pitch estimation.

While several perceptual procedures exist to assess pitch accuracy of singing, computer-assisted methods are scarce. We developed and tested a method that combines automated frequency analysis with manual note segmentation. PitchAnalyzer2.2 software was used to extract the fundamental frequency and the musical pitches of the audio samples. The onsets, offsets, and steady-state sections of the sung syllables were detected manually to identify the keys or root notes and modulations that were dominant in a sample. Finally, the difference between the singing-sample melody and the target melody was computed to determine the overall scores for each performance. We used samples obtained from Chinese primary school children to test the new method. The samples were collected before and after the children participated in an extracurricular training programme on music versus foreign-language learning. Replicating previous findings, girls outperformed boys in singing accuracy. There were no effects of training programme or age. Testing the new method for analysis of singing-pitch accuracy for a target melody revealed that it is appropriate to generate overall performance scores for samples of children’s singing, particularly when their singing sample contains multiple modulations.

A fundamental task for research on music cognition is to pinpoint the exact roles of nature and nurture in the perception of consonance/dissonance, i.e., the relative agreeableness and stability versus disagreeableness and instability of concurrent pitch combinations. In this experiment listeners (N = 133) from two cultural groups (Indian and Western) with two levels of musical expertise (musicians and non-musicians) rated perceived tension across all 12 harmonic intervals ranging from the minor second to the perfect octave, played with just intonation on the harmonium timbre. The results show that the psychoacoustic phenomena of roughness and harmonicity predict perceived tension in intervals best for Western musicians, followed by somewhat smaller explanatory rates for Indian musicians and Western non-musicians. Psychoacoustics was a weaker predictor of tension ratings for non-musicians compared to musicians across both cultural groups, but especially in the case of Indian listeners. The obtained results are discussed in the light of previous consonance/dissonance research.