This data release relates to the topic of interactions in music performance. The data consists of annotations, movement and audio descriptors, and computational predictions of interactions in jazz duos. The jazz duos are represented by 15 pulsed (standard jazz) improvisations and 15 non-pulsed improvisations (free jazz) examples, which have been documented previously (Moran et al., 2015) and the videos have been released separately. Here we release only the numerical data used in the evaluation of wavelet-based methods to estimate the interactive bouts within the performances. The data release is organized in two experiments, which are documented separately.Group music-making is a distinctive mode of human social interaction: it is a widespread activity that showcases the remarkable capacity for precision and creativity demonstrated in the coordination of rhythmic behaviour between individuals. Such coordination entails interpersonal entrainment, a process whereby two or more individuals interact with each other in a manner supporting the synchronization of body movements and musical sounds. Although musical entrainment is prevalent across the world's cultures, the way in which it is manifested appears to vary as a function of differences in social, ritual and musical conventions. A better understanding of the process of interpersonal entrainment and its cultural variation is therefore imperative. The main objective of this project is to investigate key aspects of interpersonal musical entrainment in a comparative study of a variety of cultural settings; it does so through the establishment of an international and interdisciplinary team, and by creating and analyzing a shared corpus of prepared and annotated performance data. Understanding musical entrainment requires contributions from several disciplines, in particular ethnomusicology, music cognition and computing. This project combines perspectives from each of these disciplines: it focuses on better understanding of culture- and genre- specific variations in interpersonal musical entrainment, addressed through both objective measures of entrainment and investigation of subjective perceptions of this coordination, interpreted in the light of ethnographic information about the functions and effects of the music and local aesthetic values. This ground-breaking, international and interdisciplinary project integrates three complementary approaches. First, we will apply and extend existing methods for the empirical study of interpersonal coordination in music-making. We will emphasize methods for analysing audio and video recordings, in order to carry out comparative analyses from several contrasting cultural settings - including India, Mali, Tunisia, Uruguay and Cuba - rather than being limited to laboratory studies. Secondly, we will conduct a series of experiments to measure listeners' perceptions and judgements of temporal coordination in recorded music extracts. The aim here is to measure both observers' sensitivity to variations in coordination, and their preferences, and to explore how these measures vary between cultures. Finally, we will explore all of these measures in relation to qualitative culture-specific factors: for instance, differences in preference for close coordination as opposed to individual freedom in group activities. The integration of these research strands will give us a much clearer picture of the shared and culturally-varying components of interpersonal musical entrainment. This collaborative research model is essential if we are to make properly informed comparative studies of a number of contrasting music cultures, and do so with a firm grounding in ethnomusicology, music cognition and the latest computational techniques in video and time series analysis. The aim is both to achieve a step-change in our understanding of musical entrainment - and therefore of social interaction and interpersonal coordination in general - and, by doing so, to establish long-term, international and interdisciplinary research collaborations bridging ethnomusicology, music cognition and computing.

The dataset contains data from two experiments. In experiment 1, the data collection method consist of (1) human annotations, (2) movement and audio descriptors based on computational feature-extraction techniques, and (3) computational predictions of interactions in jazz duos. The sample consists of jazz duos which are represented by 15 pulsed (standard jazz) improvisation and 15 non-pulsed improvisation (free improvisation) examples, which have been documented previously (Moran et al., 2015, see Related Resources). The sampling procedure is opportunity sampling and based on material already available and the population studied is very specific expert community of musicians, with no intention of generalisation across musical genres or situations. The videos used in the present work have been released separately, see Related Resources. Here we release only the numerical data used in the evaluation of wavelet-based methods to estimate the interactive bouts within the performances. In experiment 2, the data collection method is online rating task of short video clips. The sampling procedure is snowball sampling where the volunteers were invited to participants to the survey via social media. The population was mainly expert musicians and people interested in music. See separate documents for Experiments 1 and 2 explaining the format of the data and the meaning of the variables.