Search

In this paper we investigate risk prediction of criminal re-offense among juvenile defendants using general-purpose machine learning (ML) algorithms. We show that in our dataset, containing hundreds of cases, ML models achieve better predictive power than a structured professional risk assessment tool, the Structured Assessment of Violence Risk in Youth (SAVRY), at the expense of not satisfying relevant group fairness metrics that SAVRY does satisfy. We explore in more detail two possible causes of this algorithmic bias that are related to biases in the data with respect to two protected groups, foreigners and women. In particular, we look at (1) the differences in the prevalence of re-offense between protected groups and (2) the influence of protected group or correlated features in the prediction. Our experiments show that both can lead to disparity between groups on the considered group fairness metrics. We observe that methods to mitigate the influence of either cause do not guarantee fair outcomes. An analysis of feature importance using LIME, a machine learning interpretability method, shows that some mitigation methods can shift the set of features that ML techniques rely on away from demographics and criminal history which are highly correlated with sensitive features. Funding was provided by Joint Research Centre (Grant No. HUMAINT).

Deep learning models have improved cutting-edge technologies in many research areas, but their black-box structure makes it difficult to understand their inner workings and the rationale behind their predictions. This may lead to unintended effects, such as being susceptible to adversarial attacks or the reinforcement of biases. There is still a lack of research in the audio domain, despite the increasing interest in developing deep learning models that provide explanations of their decisions. To reduce this gap, we propose a novel interpretable deep learning model for automatic sound classification, which explains its predictions based on the similarity of the input to a set of learned prototypes in a latent space. We leverage domain knowledge by designing a frequency-dependent similarity measure and by considering different time-frequency resolutions in the feature space. The proposed model achieves results that are comparable to that of the state-of-the-art methods in three different sound classification tasks involving speech, music, and environmental audio. In addition, we present two automatic methods to prune the proposed model that exploit its interpretability. Our system is open source and it is accompanied by a web application for the manual editing of the model, which allows for a human-in-the-loop debugging approach.

The use of machine learning (ML) approaches to target clinical problems is called to revolutionize clinical decision-making. The success of these tools is subjected to the understanding of the intrinsic processes being used during the classical pathway by which clinicians make decisions. In a parallelism with this pathway, ML can have an impact at four levels: for data acquisition, predominantly by extracting standardized, high-quality information with the smallest possible learning curve; for feature extraction, by discharging healthcare practitioners from performing tedious measurements on raw data; for interpretation, by digesting complex, heterogeneous data in order to augment the understanding of the patient status; and for decision support, by leveraging the previous step to predict clinical outcomes, response to treatment or to recommend a specific intervention. This paper discusses the state-of-the-art, as well as the current clinical status and challenges associated with each of these tasks, together with the challenges related to the learning process, the auditability/traceability, the system infrastructure and the integration within clinical processes.

The application of artificial intelligence (AI) to music stretches back many decades, and presents numerous unique opportunities for a variety of uses, such as the recommendation of recorded music from massive commercial archives, or the (semi-)automated creation of music. Due to unparalleled access to music data and effective learning algorithms running on high-powered computational hardware, AI is now producing surprising outcomes in a domain fully entrenched in human creativity&mdash, not to mention a revenue source around the globe. These developments call for a close inspection of what is occurring, and consideration of how it is changing and can change our relationship with music for better and for worse. This article looks at AI applied to music from two perspectives: copyright law and engineering praxis. It grounds its discussion in the development and use of a specific application of AI in music creation, which raises further and unanticipated questions. Most of the questions collected in this article are open as their answers are not yet clear at this time, but they are nonetheless important to consider as AI technologies develop and are applied more widely to music, not to mention other domains centred on human creativity.

Comunicació presentada a la 2016 8th International Conference on Games and Virtual Worlds for Serious Applications (VS-Games), celebrada els dies 7 a 9 de setembre de 2016 a Barcelona, Espanya. This paper combines Audio Signal Processing and Virtual Reality (VR) content to create novel immersive experiences for orchestral music audiences. In VR, the auralization of sound sources of recorded live content remains still a rather unexplored topic. We aim to build a multimodal experience, where visual and audio cues bring a sonic augmentation of the real scene. In the particular scenario of orchestral music content, our goal is to acoustically zoom on a particular instrument when the VR user stares at it. This work aims to improve the learning aspects of music listening, either for education or for personal enrichment. We use audio signal processing to separate different sound sources (instruments) in a acoustic scene (orchestral music recording). Given the signals captured by multiple microphones and the musical score of the piece, our system is able to isolate the different instruments. From the processed separated tracks, we use a binaural rendering technique to emphasize a give instrument. For these experiments we used original content from top European orchestras.

In order to better understand the musical properties which elicit an increased sensation of wanting to move when listening to music – groove – we investigate the effect of adding syncopation to simple piano melodies, under the hypothesis that syncopation is correlated to groove. Across two experiments we examine listeners’ experience of groove to synthesized musical stimuli covering a range of syncopation levels and densities of musical events, according to formal rules implemented by a computer algorithm that shifts musical events from strong to weak metrical positions. Results indicate that moderate levels of syncopation lead to significantly higher groove ratings than melodies without any syncopation or with maximum possible syncopation. A comparison between the various trans-formations and the way they were rated shows that there is no simple relation between syncopation magnitude and groove.

This paper presents a drum transcription algorithm adjusted to the constraints of real-time audio. We introduce an instance filtering (IF) method using sub-band onset detection, which improves the performance of a system having at its core a feature-based K-nearest neighbor classifier (KNN). The architecture proposed allows for adapting different parts of the algorithm for either bass drum, snare drum or hi-hat cymbals. The open-source system is implemented in the graphic programming languages Pure Data (PD) and Max MSP, and aims to work with a large variety of drum sets. We evaluated its performance on a database of audio samples generated from a well known collection of midi drum loops randomly matched with a diverse collection of drum sets. Both of the evaluation stages, testing and validation, show an improvement in the performance when using the instance filtering algorithm.

The classical music traditions of the Indian subcontinent,/nHindustani and Carnatic, offer an excellent ground on which/nto test the limitations of current music information research/napproaches. At the same time, studies based on these music/ntraditions can shed light on how to solve new and complex/nmusic modeling problems. Both traditions have very distinct/ncharacteristics, specially compared with western ones:/nthey have developed unique instruments, musical forms, performance/npractices, social uses and context. In this article,/nwe focus on the Carnatic music tradition of south India, especially/non its melodic characteristics. We overview the/ntheoretical aspects that are relevant for music information/nresearch and discuss the scarce computational approaches/ndeveloped so far. We put emphasis on the limitations of the/ncurrent methodologies and we present open issues that have/nnot yet been addressed and that we believe are important to/nbe worked on. The research leading to these results has received funding/nfrom the European Research Council under the European/nUnion’s Seventh Framework Programme (FP7/2007-2013)/n/ ERC grant agreement 267583 (CompMusic).

The issue of tuning in Indian classical music has been, historically,/na matter of theoretical debate. In this paper, we/nstudy its contemporary practice in sung performances of Carnatic/nand Hindustani music following an empiric and quantitative/napproach. To do so, we select stable fundamental frequencies,/nestimated via a standard algorithm, and construct/ninterval histograms from a pool of recordings. We then compare/nsuch histograms against the ones obtained for different/nmusic sources and against the theoretical values derived/nfrom 12-note just intonation and equal temperament. Our/nresults evidence that the tunings in Carnatic and Hindustani/nmusic differ, the former tending to a just intonation system/nand the latter having much equal-tempered influences. Carnatic/nmusic also presents signs of a more continuous distribution/nof pitches. Further subdivisions of the octave are/npartially investigated, finding no strong evidence of them. The research leading to these results/nhas received funding from the European Research Council/nunder the European Union’s Seventh Framework Programme/n(FP7/2007-2013) / ERC grant agreement 267583/n(CompMusic).

The thesis aims to develop a system of Hindustani tala automatic recognition which can be trained by building a labeled corpus of Hindustani songs with tabla accompaniment. Most of the research concerning rhythm in the North Indian classical music was developed around monophonic recordings and the scope was just recognizing the tabla strokes or modeling the expressiveness of tabla solos and not the metric cycles in which these strokes usually occur, the talas. The aspects researched were segmentation and stroke recognition in a polyphonic context, as recognizing the talas is a perceptually challenging task and the automatic detection proved to be even more dificult.

This paper proposes a system for score-informed audio source separation for multichannel orchestral recordings. The orchestral music repertoire relies on the existence of scores. Thus, a reliable separation requires a good alignment of the score with the audio of the performance. To that extent, automatic score alignment methods are reliable when allowing a tolerance window around the actual onset and offset. Moreover, several factors increase the difficulty of our task: a high reverberant image, large ensembles having rich polyphony, and a large variety of instruments recorded within a distant-microphone setup. To solve these problems, we design context-specific methods such as the refinement of score-following output in order to obtain a more precise alignment. Moreover, we extend a close-microphone separation framework to deal with the distant-microphone orchestral recordings. Then, we propose the first open evaluation dataset in this musical context, including annotations of the notes played by multiple instruments from an orchestral ensemble. The evaluation aims at analyzing the interactions of important parts of the separation framework on the quality of separation. Results show that we are able to align the original score with the audio of the performance and separate the sources corresponding to the instrument sections. This work is partially supported by the Spanish Ministry of Economy and Competitiveness under CASAS Project (TIN2015-70816-R).

Comunicació presentada a la conferència 14th Sound and Music Computing Conference, celebrada a Finlàndia del 5 al 8 de juliol de 2017. Deep learning approaches have become increasingly popular in estimating time-frequency masks for audio source separation. However, training neural networks usually requires a considerable amount of data. Music data is scarce, particularly for the task of classical music source separation, where we need multi-track recordings with isolated instruments. In this work, we depart from the assumption that all the renditions of a piece are based on the same musical score, and we can generate multiple renditions of the score by synthesizing it with different performance properties, e.g. tempo, dynamics, timbre and local timing variations. We then use this data to train a convolutional neural network (CNN) which can separate with low latency all the renditions of a score or a set of scores. The trained model is tested on real life recordings and is able to effectively separate the corresponding sources. This work follows the principle of research reproducibility, providing related data and code, and can be extended to separate other pieces. The TITANX used for this research was donated by the NVIDIA Corporation. This work is partially supported by the Spanish Ministry of Economy and Competitiveness under CASAS project (TIN2015-70816-R) and by the Spanish Ministry of Economy and Competitiveness under the Maria de Maeztu Units of Excellence Programme (MDM- 2015-0502). We thank Agustin Martorell for his help with Sibelius and Pritish Chandna for his useful feedback.