JSD: A Dataset for Structure Analysis in Jazz Music

Abstract

Given a music recording, music structure analysis aims at identifying important structural elements and segmenting the recording according to these elements. In jazz music, a performance is often structured by repeating harmonic schemata (known as choruses), which lay the foundation for improvisation by soloists. Within the fields of music information retrieval (MIR) and computational musicology, the Weimar Jazz Database (WJD) has turned out to be an extremely valuable resource for jazz research. Containing high-quality solo transcriptions for 456 solo sections, the dataset opened up new avenues for the understanding of creative processes in jazz improvisation using computational methods. In this paper, we complement this dataset by introducing the Jazz Structure Dataset (JSD), which provides annotations on structure and instrumentation of entire recordings. The JSD comprises 340 recordings with more than 3000 annotated segments, along with a segment-wise encoding of the solo and accompanying instruments. These annotations provide the basis for training, testing, and evaluating models for various important MIR tasks, including structure analysis, solo detection, or instrument recognition. As an example application, we consider the task of structure boundary detection. Based on a traditional novelty-based as well as a more recent data-driven approach using deep learning, we indicate the potential of the JSD while critically reflecting on some evaluation aspects of structure analysis. In this context, we also demonstrate how the JSD annotations and analysis results can be made accessible in a user-friendly way via web-based interfaces for data inspection and visualization. All annotations, experimental results, and code for reproducibility are made publicly available for research purposes.

Self-Supervised Learning of Multi-Level Audio Representations for Music Segmentation

“…Boundary detection results on JSD given in Table II show that this dataset is much more challenging than SALAMI for any configuration of our system, as can be seen by the severe performance drops on all metrics. However, our results are still on-par with the baselines experimented in the original paper [24]. We can also observe that multi-level representations tend to outperform the single-level ones on a majority of boundary detection metrics (see rows M-level and Flat).…”

“…JSD: the Jazz Structure Dataset [24] gathers 340 jazz recordings that were manually annotated by 3 semiprofessional musicians with a background in jazz music. A key advantage of this dataset resides in the consistency of its annotations.…”

PiJAMA: Piano Jazz with Automatic MIDI Annotations

“…Another thing lacking from our dataset is annotations beyond MIDI. Having labels for beats, bars, chords, phrases, sections, and other structural aspects would enrich the dataset (see Eremenko et al, 2018;Balke et al, 2022). From a more musical perspective, our methodology focuses entirely on solo jazz piano.…”

Jazz Trio Database: Automated Annotation of Jazz Piano Trio Recordings Processed Using Audio Source Separation

“…The majority of the annotations were created manually: only dynamics and intonation were measured automatically. A subsequent project has since extended the WJD to include structure and instrumentation annotations for each piece (Balke et al, 2022). While the WJD is undoubtedly a landmark in the computational analysis of jazz, it only captures the performance of soloists, not their accompaniment.…”