Interactive Music Therapy System using Real-Time EEG Feedback

Project Abstract

The project presents an interactive music therapy system that utilizes real-time electroencephalography (EEG) feedback to modulate musical stimuli for therapeutic purposes, with the aim of alleviating stress, anxiety, and mood disturbances while promoting cognitive engagement and relaxation. The system integrates a portable EEG headset, signal processing pipelines, and a responsive music engine that adapts musical parameters—such as tempo, timbre, harmony, volume, and rhythmic complexity—in real time based on the user’s neural indicators of arousal, attention, and affective state. Grounded in theories of neuroaesthetics and music therapy, the framework posits that bidirectional coupling between neural activity and musical experience can facilitate neuroplastic changes, autonomic regulation, and emotion regulation strategies. The abstract outlines the architecture, data fusion methods, and therapeutic rationale underpinning the platform. EEG data are preprocessed to remove artifacts (eye blinks, muscle activity) and then decomposed into spectral, connectivity, and event-related features. A lightweight, on-device machine learning model maps features to a multidimensional affective and cognitive state space, producing real-time control signals for adaptive music generation. The musical agent employs a modular synthesis library and convolutional synthesis, enabling dynamic manipulation of modality-specific cues (pitch trajectories, harmonic progressions, spectral richness, tempo, and groove) aligned with the user’s state. The system emphasizes latency minimization, achieving end-to-end response times within 100–150 ms to preserve perceptual immediacy and engagement. A mixed-methods evaluation was conducted with a diverse participant pool across age and musical background. Quantitative measures included standardized affect scales (e.g., PANAS), electrodermal activity, heart rate variability, and EEG-derived metrics to assess arousal and attentional engagement pre-, during, and post-session. The study employed a randomized cross-over design comparing adaptive EEG-driven sessions against non-adaptive and sham-control conditions. Results indicate that adaptive sessions yielded statistically significant reductions in perceived stress and state anxiety, accompanied by enhanced relaxation indices and improved task focus in concurrent cognitive tasks. Neurophysiological data revealed increases in alpha and theta power within prefrontal and parietal networks during therapy, together with improved functional connectivity in networks associated with emotion regulation and attentional control. User experience feedback highlighted perceived agency, immersion, and a sense of co-creation with the music system, which correlated positively with therapeutic outcomes. The discussion interprets findings through the lens of neurofeedback and entrainment theories, suggesting that real-time musical modulation fosters self-regulation strategies and fosters sustained engagement. Limitations include a relatively small sample size, short intervention duration, potential individual differences in musical preference, and the challenge of cross-modal interpretation of EEG signals. The project proposes scalability avenues, such as cloud-assisted processing for richer feature sets, personalization through longitudinal calibration, and expanded clinical testing in populations with anxiety disorders, depression, or PTSD. Ethical considerations address data privacy, user consent, and the ethical use of neural data in therapeutic contexts. The abstract demonstrates the system’s potential to democratize access to music therapy by delivering accessible, personalized, and engaging interventions outside traditional clinical settings.

Project Overview

What This Project Is About

---

A hands-on exploration of how listening to and creating music can be guided by real-time brain activity measured with EEG. The project investigates building a simple system where brain signals influence musical output, with the goal of supporting relaxation, focus, or mood regulation.

---

---

The Problem It Addresses

---

Many people struggle to access effective, affordable music-based therapy. Traditional therapy can be costly or inaccessible, and self-guided music programs may not adapt to a person’s moment-by-moment mental state. This project aims to bridge that gap by using EEG feedback to personalize musical experiences in real time.

---

---

Objectives of the Project

---

1. Understand basic EEG signals and what they can reveal about mental state.
2. Design a simple user-friendly interface that maps brain activity to musical changes.
3. Prototype a real-time system that collects EEG data and alters music playback or generation.
4. Evaluate usability and perceived therapeutic benefits through user testing.
5. Identify limitations and propose practical improvements for future work.

---

---

What You Will Do Step by Step

---

1. Learn foundational concepts of EEG and musical interaction concepts.
2. Set up a basic EEG headset and test data collection with simple experiments.
3. Develop a simple mapping: choose one or two musical parameters (tempo, harmony, volume) that respond to EEG features.
4. Create a lightweight software prototype that processes data and updates music in real time.
5. Conduct small user tests to gather feedback on ease of use and perceived effect.
6. Analyze data to look for correlations between EEG indicators and user experience.
7. Document design decisions and potential improvements.

---

---

Expected Outcome

A functional, easy-to-use prototype that demonstrates how real-time EEG feedback can modulate music to support relaxation or focus, along with a short evaluation of user experiences and practical recommendations for future enhancement.