Nadine Kroher

Nadine Kroher, Helena Cuesta, Aggelos Pikrakis

We are investigating the broader concept of using AI-based generative music systems to generate training data for Music Information Retrieval (MIR) tasks. To kick off this line of work, we ran an initial experiment in which we trained a genre classifier on a fully artificial music dataset created with MusicGen. We constructed over 50 000 genre- conditioned textual descriptions and generated a collection of music excerpts that covers five musical genres. Our preliminary results show that the proposed model can learn genre-specific characteristics from artificial music tracks that generalise well to real-world music recordings.

Nadine Kroher, Steven Manangu, Aggelos Pikrakis

Most contemporary music tagging systems rely on large volumes of annotated data. As an alternative, we investigate the extent to which synthetically generated music excerpts can improve tagging systems when only small annotated collections are available. To this end, we release GTZAN-synth, a synthetic dataset that follows the taxonomy of the well-known GTZAN dataset while being ten times larger in data volume. We first observe that simply adding this synthetic dataset to the training split of GTZAN does not result into performance improvements. We then proceed to investigating domain adaptation, transfer learning and fine-tuning strategies for the task at hand and draw the conclusion that the last two options yield an increase in accuracy. Overall, the proposed approach can be considered as a first guide in a promising field for future research.

Nadine Kroher, Aggelos Pikrakis

We present computational tools that we developed for the analysis of a large corpus of flamenco music recordings, along with the related exploratory findings. The proposed computational backend is based on a set of Convolutional Neural Networks that provide the structural annotation of each music recording with respect to the presence of vocals, guitar and hand-clapping ("palmas"). The resulting, automatically extracted annotations, allowed for the visualization of music recordings in structurally meaningful ways, the extraction of global statistics related to the instrumentation of flamenco music, the detection of a cappella and instrumental recordings for which no such information existed, the investigation of differences in structure and instrumentation across styles and the study of tonality across instrumentation and styles. The reported findings show that it is feasible to perform a large scale analysis of flamenco music with state-of-the-art classification technology and produce automatically extracted descriptors that are both musicologically valid and useful, in the sense that they can enhance conventional metadata schemes and assist bridging the semantic gap between audio recordings and high-level musicological concepts.

Nadine Kroher, Emilia Gómez

Automatic note-level transcription is considered one of the most challenging tasks in music information retrieval. The specific case of flamenco singing transcription poses a particular challenge due to its complex melodic progressions, intonation inaccuracies, the use of a high degree of ornamentation and the presence of guitar accompaniment. In this study, we explore the limitations of existing state of the art transcription systems for the case of flamenco singing and propose a specific solution for this genre: We first extract the predominant melody and apply a novel contour filtering process to eliminate segments of the pitch contour which originate from the guitar accompaniment. We formulate a set of onset detection functions based on volume and pitch characteristics to segment the resulting vocal pitch contour into discrete note events. A quantised pitch label is assigned to each note event by combining global pitch class probabilities with local pitch contour statistics. The proposed system outperforms state of the art singing transcription systems with respect to voicing accuracy, onset detection and overall performance when evaluated on flamenco singing datasets.

Nadine Kroher, José-Miguel Díaz-Báñez, Joaquin Mora et al.

Flamenco is a music tradition from Southern Spain which attracts a growing community of enthusiasts around the world. Its unique melodic and rhythmic elements, the typically spontaneous and improvised interpretation and its diversity regarding styles make this still largely undocumented art form a particularly interesting material for musicological studies. In prior works it has already been demonstrated that research on computational analysis of flamenco music, despite it being a relatively new field, can provide powerful tools for the discovery and diffusion of this genre. In this paper we present corpusCOFLA, a data framework for the development of such computational tools. The proposed collection of audio recordings and meta-data serves as a pool for creating annotated subsets which can be used in development and evaluation of algorithms for specific music information retrieval tasks. First, we describe the design criteria for the corpus creation and then provide various examples of subsets drawn from the corpus. We showcase possible research applications in the context of computational study of flamenco music and give perspectives regarding further development of the corpus.