The BEATS Dataset: Towards Understanding Music-Physiology Dynamics in Real-World Scenarios

arXiv:2611.15837 · eess.SP, cs.AI · Submitted 2026-11-11 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.

Rosa: I'm Rosa.

Dev: And I'm Dev, here with Taro.

Rosa: We're looking at the paper: "The BEATS Dataset".

Dev: Here's the gist of the paper. The BEATS dataset introduces a robust compilation of psycho-physiological data collected in free-movement environments from 30 participants listening to curated playlists intended to promote well-being This dataset aims to support research on wellbeing, music psychology, affective computing and signal processing By situating data collection outside of lab settings, the BEATS dataset offers the opportunity to study real-world music-listening scenarios Dataset Composition The BEATS dataset consists of four main components: (1) sensor data collected during music listening sessions using a research-graded wearable device in free-movement environments, (2) music listening activity records, stimulus metadata, acoustic features, and self-reports associated with the listening sessions, (3) individual characteristics, such as demographics, music preference, musical skills, and personality traits, and (4) self-reported psychological measures before and after their experimental participation spanning for a number of weeks.

Rosa: The dataset includes 21-day in-the-wild recordings of multimodal physiological data from 30 participants, with over 10,000 hours of physiological recordings from 30 participants (n = 30).

Paper summary: Rosa: Looking at the whole BEATS Dataset paper, it seems the authors are arguing that this collection is a significant step forward for understanding how music listening actually impacts people in their normal lives.

Dev: They highlight that by capturing data in free-movement environments, they provide an assessment of emotions and physiological responses that isn't limited to short-term, semi-controlled studies <ref:2611.15837#pg2>.

Rosa: And the sheer size of the dataset, with those over ten thousand hours from thirty participants over twenty-one days, is presented as something valuable for advancing music psychology and affective computing <ref:2611.15837#pg3>.

Dev: The implication is that this resource should be used to develop more effective personalized music-based interventions tailored to real-world contexts, rather than just theoretical models.

Rosa: They are positioning it as a resource for the community to advance our understanding of music and physiology dynamics in these real-world scenarios <ref:2611.15837#pg3>.

Dev: It shifts the focus from isolated experiments to studying how music consumption becomes a dispersed practice that enriches other main activities <ref:2611.15837#pg2>.

Taro: What I hear is that this moves the conversation toward testing how these systems handle the messy reality of real-world data and movement, which is crucial for autonomy research <ref:2611.15837#pg3>.

Rosa: That's right, it’s about taking those physiological measurements and linking them to actual behaviors like listening while driving or cooking <ref:2611.15837#pg3>.

Conclusion: Rosa: So we’ve looked at how this BEATS dataset is built, and now we're coming to the end of this paper, "The BEATS Dataset: Towards Understanding Music-Physiology Dynamics in Real-World Scenarios."

Dev: It really boils down to taking all this collected stuff—the sensor data, the music logs, the personal details—and putting it together so researchers can actually study how people listen to music when they're just living their normal lives.

Rosa: Exactly. The authors are arguing that moving data out of a controlled lab and into these free-movement settings is super important for seeing if these physiological responses actually mean anything in the real world.

Dev: And they’ve gathered a huge amount of information, over ten thousand hours from thirty different people, which gives researchers a lot to work with for music psychology and affective computing stuff.

Rosa: It suggests that we can get a much clearer picture of how music influences our bodies when things are actually happening in the wild, not just in some quiet testing room.

Dev: It opens the door for developing tools that help people personalize their music experiences based on where they are and what they’re doing right then.

Rosa: If you only listen to this show, it means thinking about how we can use these real-world listening patterns to build better systems for well-being and maybe even how we design music for specific activities like driving or cooking.

Dev: It’s a big step because it validates studying music through the lens of messy, real-life human behavior instead of just clean, artificial setups.

Rosa: So now we have this massive collection sitting there, ready for people to start asking really complex questions about what it all means for how we experience music daily.

University of Glasgow · OmniBuds Ltd

eess.SP, cs.AI

Submitted: 2026-11-11

Updated: 2026-11-11

Comments: Under review in Nature Scientific Data

Code: https://github.com/lciti/cvxEDA

Project page: https://neuropsychology.github.io/NeuroKit

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 82/100

The gist: The BEATS dataset introduces a robust compilation of psycho-physiological data collected in free-movement environments from 30 participants listening to curated playlists intended to promote…

Key concepts

BEATS Dataset
A compilation of psycho-physiological data gathered from 30 participants listening to curated playlists in free-movement environments. It integrates sensor data, music activity logs, and self-reports to examine real-world music consumption and its effects on well-being.
Sensor Data
Physiological measurements collected using a research-graded wearable device during music listening sessions. This data captures real-time bodily responses related to the participants' physical states while they are engaged in listening activities.
Free-Movement Environments
The setting where the data was collected, meaning outside of a laboratory. This context is crucial because it allows researchers to study how music and physiology interact during everyday, unscripted music listening situations.
Multimodal Data Integration
The dataset combines four distinct types of information: physiological sensor readings, detailed music activity records (including acoustic features), participant characteristics, and psychological self-reports. This comprehensive approach enables deeper analysis of complex relationships.

Terminology

Summary

The BEATS dataset introduces a robust compilation of psycho-physiological data collected in free-movement environments from 30 participants listening to curated playlists intended to promote well-being This dataset aims to support research on wellbeing, music psychology, affective computing and signal processing By situating data collection outside of lab settings, the BEATS dataset offers the opportunity to study real-world music-listening scenarios

Dataset Composition

The BEATS dataset consists of four main components: (1) sensor data collected during music listening sessions using a research-graded wearable device in free-movement environments, (2) music listening activity records, stimulus metadata, acoustic features, and self-reports associated with the listening sessions, (3) individual characteristics, such as demographics, music preference, musical skills, and personality traits, and (4) self-reported psychological measures before and after their experimental participation spanning for a number of weeks. The dataset includes 21-day in-the-wild recordings of multimodal physiological data from 30 participants, with over 10,000 hours of physiological recordings from 30 participants (n = 30). p. 8

Where the paper says it

Altogether, the BEATS dataset includes data from 30 participants monitored over 21 days, yielding 10,725 hours of raw physiological recordings from PPG, EDA, skin temperature, and triaxial accelerometer. Figure 3 illustrates an example music listening session aligned with multimodal physiological signals.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. multimodal physiological data integration allows AI systems to infer affective states using traditional machine learning approaches in real-world settings, moving beyond short-term, semi-controlled studies.

  2. The system can analyze music listening activity alongside raw physiological signals (PPG, EDA, skin temperature, triaxial acceleration) to derive minute-level digital biomarkers related to cardiovascular activity and physical activity.

  3. AI systems can perform personalized music-based interventions by matching curated playlists to specific musical functions like affect regulation or goal attainment based on participant profiles (personality, music preference).

  4. The system can validate the acoustic profiles of different music functions using supervised learning classifiers, identifying discriminative features such as spectral contrast for Affective music and average onset frequency for Goal-Attainment music.

  5. AI systems can assess the temporal evolution of physiological markers by comparing window-level features across pre-, during-, and post-listening periods, identifying shifts in HRV, EDA, and temperature relative to individual baselines.

Abstract

We introduce the BEATS dataset, a robust compilation of psycho-physiological data collected in the wild from 30 participants while they listened to curated playlists intended to promote well-being. An acronym for Biomarkers, Experience and Affect from Therapeutic Soundscapes, the BEATS dataset aims to support research on wellbeing, music psychology, affective computing and signal processing. More specifically, this dataset includes: (1) sensor data collected during music listening sessions using a research-grade wearable device in free-movement environments; (2) music listening activity records, stimulus metadata, acoustic features, and self-reports associated with the listening sessions (3) individual characteristics, such as demographics, music preference, musical skills, and personality traits; and (4) self-reported psychological measures before and after their experimental participation spanning for a number of weeks. By situating data collection outside of lab settings, the BEATS dataset offers the opportunity to study real-world music-listening scenarios, allowing for a richer exploration of the physiological, subjective and therapeutic effects of music under real-world conditions.

Sources

Related papers