A5012131168- Music and Audio Processing
- Music Technology and Sound Studies
- Speech and Audio Processing
- Diverse Musicological Studies
- Neuroscience and Music Perception
- Diversity and Impact of Dance
- Speech Recognition and Synthesis
- Animal Vocal Communication and Behavior
- Innovative Human-Technology Interaction
- Musicology and Musical Analysis
- Human Motion and Animation
- Diverse Music Education Insights
- Music Therapy and Health
- Classical Antiquity Studies
- Blind Source Separation Techniques
- Aesthetic Perception and Analysis
- Advanced Data Storage Technologies
- Video Analysis and Summarization
- Human Pose and Action Recognition
- Cultural Industries and Urban Development
- Asian Culture and Media Studies
- Ancient Mediterranean Archaeology and History
- Music Education and Analysis
- Artistic and Creative Research
- Creativity in Education and Neuroscience
KTH Royal Institute of Technology
2017-2024
Universitat de les Illes Balears
2022
Interaction Design (United Kingdom)
2021-2022
Austrian Research Institute for Artificial Intelligence
2016-2017
Boğaziçi University
2013-2016
Universitat Pompeu Fabra
2012-2016
Bahçeşehir University
2013-2014
University of Crete
2007-2012
INESC TEC
2012
Instituto de Engenharia de Sistemas e Computadores Investigação e Desenvolvimento
2011
Abstract Music is present in every known society but varies from place to place. What, if anything, universal music cognition? We measured a signature of mental representations rhythm 39 participant groups 15 countries, spanning urban societies and Indigenous populations. Listeners reproduced random ‘seed’ rhythms; their reproductions were fed back as the stimulus (as game ‘telephone’), such that biases (the prior) could be estimated distribution reproductions. Every tested group showed...
In this paper, we propose a method that can identify challenging music samples for beat tracking without ground truth. Our method, motivated by the machine learning "selective sampling," is based on measurement of mutual agreement between sequences. calculating show critical influence different evaluation measures. Using our approach demonstrate how to compile new dataset comprised difficult excerpts and examine difficulty in context perceptual musical properties. Based tag analysis indicate...
The fields of music, health, and technology have seen significant interactions in recent years developing music for health care well-being. In an effort to strengthen the collaboration between involved disciplines, workshop “Music, Computing, Health” was held discuss best practices state-of-the-art at intersection these areas with researchers from psychology neuroscience, therapy, information retrieval, technology, medical (medtech), robotics. Following discussions workshop, this article...
Nonnegative matrix factorization (NMF) is used to derive a novel description for the timbre of musical sounds. Using NMF, spectrogram factorized providing characteristic spectral basis. Assuming set spectrograms given genre, space spanned by vectors obtained bases modeled statistically using mixtures Gaussians, resulting in base this genre. This shown improve classification results up 23.3% compared MFCC-based models, while compression performed decreases training time significantly....
This article examines ethical dimensions of Music Information Retrieval (MIR) technology. It uses practical ethics (especially computer and engineering ethics) socio-technical approaches to provide a theoretical basis that can inform discussions in MIR. To help ground the discussion, engages with concrete examples discourse drawn from MIR field. argues technology is not value-neutral but influenced by design choices, so has unintended ethically relevant implications. These be invisible...
Dance requires skillful composition of complex movements that follow rhythmic, tonal and timbral features music. Formally, generating dance conditioned on a piece music can be expressed as problem modelling high-dimensional continuous motion signal, an audio signal. In this work we make two contributions to tackle problem. First, present novel probabilistic autoregressive architecture models the distribution over future poses with normalizing flow previous well context, using multimodal...
In this paper, we suggest a novel group delay based method for the onset detection of pitched instruments. It is proposed to approach problem by examining three dimensions separately: phase (i.e., delay), magnitude and pitch. The evaluation suggested detectors phase, pitch performed using new publicly available fully annotated database monophonic recordings which balanced in terms included instruments samples per instrument, while it contains different performance styles. Results show that...
It has long been assumed that rhythm cognition builds on perceptual categories tied to prototypes defined by small-integer ratios, such as 1:1 and 2:1. This study aims evaluate the relative contributions of both generic constraints selected cultural particularities in shaping rhythmic prototypes. We experimentally tested musicians’ synchronization (finger tapping) with simple periodic rhythms at two different tempi participants Mali, Bulgaria, Germany. found support for classic assumption...
Music is present in every known society, yet varies from place to place. What, if anything, universal music cognition? We measured a signature of mental representations rhythm 39 participant groups 15 countries, spanning urban societies and indigenous populations. Listeners reproduced random ‘‘seed’’ rhythms; their reproductions were fed back as the stimulus (as game “telephone”), such that biases (the prior) could be estimated distribution reproductions. Every tested group showed sparse...
As a special case of the Mellin transform, scale transform has been applied in various signal processing areas, order to get description that is invariant changes. In this paper, autocorrelation sequences derived from music signals. It shown two such sequences, when similar rhythms with different tempo, differ mainly by scaling factor. By using proposed descriptors are robust tempo changes, and specially suited for comparison pieces tempi but rhythm. characteristics widely encountered...
The subject of this paper is the conversion a given speaker's voice (the source speaker) into another identified target one). We assume we have at our disposal large amount speech samples from and with least part them being parallel. proposed system built on mapping function between spectral envelopes followed by frame selection algorithm to produce final envelopes. Converted produced basic LP analysis synthesis using converted compared three types conversion: without mapping, excitation...
This text targets a review of the computational analysis literature for Turkish makam music, discussing in detail challenges involved and presenting perspective further studies. For that purpose, basic concepts music description melodic, rhythmic timbral aspects are considered detail. Studies on tuning analysis, automatic transcription, melodic usul detection reviewed. Technological data resource needs advancement discussed available sources presented.
Sounds in a piece of music form rhythmic patterns on the surface signal, and metered these stand some relation to underlying mode or meter. In this paper, we investigate how rhythm is related usul, which are modes compositions Turkish makam music. On large corpus notations vocal pieces short usul observe ways notes distributed usul. We differences distributions between Eurogenetic music, imply less accentuated stratification meter changes style two composers who represent different...
AbstractThe most relevant representations of music are notations and audio recordings, each which emphasizes a particular perspective promotes different approximations in the analysis understanding music. Linking these two analysing them jointly should help to better study many musical facets by being able combine complementary methodologies. In order develop accurate linking methods, we have take into account specificities given type this paper, present method for musically sections score...
AbstractAbstractThe aim of this paper is to identify and discuss various methods in computational rhythm description Carnatic Hindustani music India, Makam Turkey. We define describe three relevant annotation tasks for these cultures—beat tracking, meter estimation, downbeat detection. then evaluate several methodologies from the state art Music Information Retrieval (MIR) tasks, using manually annotated datasets Turkish Indian music. This evaluation provides insights into nature cultures...
Ai technologies are increasingly used by artists and creatives, they as any other (technological) artefacts embedded with the values practices part of their historical development. These, for example, include desensitized embodied orientations being in world, involving such disregard, abuse, exploitation non-human (e.g. climate) human gender racialization relation to power). Our world how get designed formed our socio-cultural practices, norms - predominant human-centered value sets (with...
Musical worlds, not unlike our lived realities, are fundamentally fragmented and diverse, a fact often seen as challenge or even threat to the validity of research in Music Information Research (MIR). In this article, we propose treat characteristic musical universe(s) an opportunity enrich re-orient MIR. We that time has arrived for MIR reflect on its ethical cultural turns (if they have been initiated at all) take them step further, with goal profoundly diversifying discipline beyond...
This paper introduces a new way to measure rhythmic similarity between two musical pieces using periodicity spectra. In order detect for of different tempi, the linearity warping path their spectra serves as similarity. Using modified kNN classification approach on datasets, proposed provides comparable accuracy (82.1%) best widely used measures (85.5%) first dataset; For second dataset, which is characterized by large variance outperforms all reference measures, reaching an 69.0%, while...
This paper introduces scale transforms to measure rhythmic similarity between two musical pieces. The rhythm of a piece music is described by the transform magnitude, computed transforming sample autocorrelation its onset strength signal domain. Then, pieces can be compared without impact tempo differences using simple distances these descriptors like cosine distance. A widely used dance dataset has been chosen for proof concept. On this data set, proposed method based on achieves...
In this paper, we propose a new state-of-the-art particle filter (PF) system to infer the metrical structure of musical audio signals. The inference method is designed overcome problem PFs in multi-modal probability distributions, which arise due tempo and phase ambiguities rhythm representations. We compare with hidden Markov model (HMM) several other PF schemes terms performance, speed scalability on datasets. demonstrate that using proposed computational complexity can be reduced...
Automatic music transcription, a central topic in signal analysis, is typically limited to equal-tempered and evaluated on quartertone tolerance level. A system proposed automatically transcribe microtonal heterophonic as applied the makam of Turkey. Specific traits this that deviate from properties targeted by current transcription tools are discussed, collection instrumental vocal recordings compiled, along with aligned reference pitch annotations. An existing multi-pitch detection...
As music generated using artificial intelligence ({\em AI music})becomes more prevalent -- originating not only from individualsbut also services or businesses centered around scalable content generation --the need to study it and its impacts grow.How can this material sources be meaningfully studied critically engaged with, however?The paper begins answer principal question by considering six aspects of music,discussing each with reference a contemporary service: {\em Boomy.com}.The are:...