« All Events

Paper Session 10b: Interactive Media II

May 15 @ 9:00 am - 10:30 am
Session Chair: Felipe Otondo

Paper abstracts

Fabian Ostermann: “BbMuse: A Blackboard-Driven Framework for Real-Time Interactive Music”

Interactive music systems are frequently built as ad hoc multi-agent architectures with custom communication protocols and project-specific execution models, while recent machine-learning approaches often encapsulate behavior in monolithic, computationally expensive black boxes. This paper revisits blackboard architectures for real-time interactive music generation and argues that composition and musical interaction can be modeled as distributed decision-making processes operating on shared musical state. We introduce BbMuse (BlackBoard MUSic Engine), an open-source, platform-independent Python framework that implements a dataflow-oriented blackboard variant inspired by real-time robotics. System state is encoded as typed representations on a global blackboard, while modules explicitly declare required and provided information, enabling automatic scheduling via topological sorting. As a result, system development becomes incremental and module-focused, since no inter-module dependencies must be specified. Further, the framework supports concurrent execution and demonstrates that real-time performance is possible with Python using native-library acceleration. We provide a growing collection of example projects, discuss diverse use cases and outline future features for learning-based module replacement as well as a GUI editor.

Eun Ji Oh, Jun Woo Beck and Alexandria Smith: “The Singing Skin: An Audience-Centered Biofeedback System for Musical Interaction Based on Galvanic Skin Response”
Music can evoke measurable physiological responses, yet these responses have been predominantly explored from the performer’s perspective in interactive and biofeedback-based music systems. In contrast, the sonification of audience physiology remains relatively underexplored in live music contexts. We present The Singing Skin, a real-time, audience-centered biofeedback system for live performance that integrates listeners’ physiological responses into musical control. The system measures galvanic skin response (GSR) and uses the phasic component of the GSR signal as an index of moment-to-moment audience engagement. This phasic GSR–based control signal is normalized and mapped to the rhythmic subdivision of a monophonic lead line generated by a wavetable synthesizer. Rather than directly modifying tempo or pitch, the control signal modulates the cutoff rate of a low-pass filter, producing an indirect pacing effect that influences perceived musical drive and energy. The system is demonstrated in a live performance setting involving a violinist and a listener equipped with GSR sensors. This work contributes a novel approach to audience-inclusive musical interaction by extending audience physiology as an active control source in live music
performance.
Penelope Bekiari and Anastasia Georgaki: “Hyponoia: An Affective Computing System for Augmented Musical Performance — A Case Study”

This paper investigates how EEG-driven biofeedback systems influence performability and listening strategies in contemporary electroacoustic performance. We introduce Hyponoia (Hyper-Observational Neuro-Oscillation Interactive Agency), a real-time interactive system that translates performers’ neurophysiological activity into state-based compositional behaviours. Unlike conventional EEG-based musical interfaces that map signals to discrete parameters, Hyponoia operates at the level of musical processes, structuring sonic form through inferred neuro-affective states. The system integrates EEG and heart-rate data within a closed biophysical feedback loop, in which performers’ internal cognitive and affective states dynamically interact with the evolving sonic environment. A comparative case study was conducted with expert and non-expert musicians performing in open-form electroacoustic contexts. We hypothesise that performers with sound-based expertise exhibit distinct patterns of neural engagement and interaction with the system. Results indicate that expert performers demonstrate richer theta activity, more coherent alpha modulation, and greater neural variability, associated with enhanced internal auditory imagery and anticipatory listening. In contrast, non-expert performers exhibit more constrained neural responses and reduced sensitivity to spectromorphological change. These findings suggest that performability in biofeedback-driven systems depends less on instrumental technique than on listening literacy and embodied sonic awareness. Rather than acting as an autonomous agent, the system functions as a responsive mediator that amplifies differences in perceptual and cognitive engagement, contributing to an emerging performance aesthetic grounded in physiological feedback and real-time interaction.

 

Details

Organizer

Venue