Adaptive Music Education Platform for Low-Resource Settings Using AI Feedback

Thesis Abstract

This study addresses the persistent inequities in music education access by developing an adaptive digital learning platform that leverages artificial intelligence to provide personalized feedback for students in low-resource settings. The core problem is the mismatch between available instructional time, qualified music teachers, and diverse learner needs, which traditionally results in suboptimal skill acquisition and uneven progression in music literacy and performance. The aim is to design, implement, and evaluate an AI-enhanced platform that personalizes practice sequences, provides real-time feedback on playing accuracy and rhythm, and adapts resources to individual learner profiles. Specific objectives include (i) implementing a multimodal AI feedback engine capable of audio, rhythm, and pitch analysis; (ii) integrating an intelligent tutoring system—grounded in Vygotsky’s zone of proximal development and constructivist learning principles—to scaffold practice and progression; (iii) evaluating the platform’s impact on musical competency, motivation, and practice adherence; (iv) examining equity outcomes across gender, age, and baseline proficiency; and (v) assessing scalability and acceptability among learners, teachers, and community stakeholders. Methodologically, the study adopts a mixed-methods, quasi-experimental design conducted over eight months in three rural and peri-urban schools serving an estimated 420 students aged 9–16. A sample of 300 students meeting inclusion criteria will be selected through stratified random sampling, with 150 assigned to the intervention group and 150 to a conventional instruction control group. The platform’s AI feedback engine will utilize signal processing (pitch detection with autocorrelation, tempo and rhythm decoding via dynamic time warping), machine learning for learner modeling (adaptive difficulty, progress forecasting, and intervention selection), and a rule-based pedagogical layer aligned to national music education standards. Data collection instruments include standardized pre- and post-tests of rhythmic accuracy, pitch accuracy, and musical literacy; validated motivation and self-regulation scales; system logs capturing practice frequency and duration; and semi-structured interviews with a purposive subsample of 40 participants, 10 teachers, and 10 guardians. Validity and reliability will be established through pilot testing, test-retest reliability for the assessments (Cronbach’s alpha > .80), and triangulation across quantitative and qualitative data. Data analysis will combine quantitative and qualitative techniques multivariate regression and ANCOVA to assess post-intervention differences while controlling for baseline scores; hierarchical linear modeling to account for nested data (students within classes within schools); time-series analysis of practice adherence; and thematic analysis of interview transcripts following Braun and Clarke’s framework. A design-based research approach will guide iterative refinement across three cycles, informing platform adaptations in response to field feedback. Expected findings include statistically significant improvements in rhythmic accuracy, pitch accuracy, and musicianship scores for the intervention group relative to controls, with effect sizes in the small-to-moderate range (Cohen’s d ? 0.35–0.60). Enhanced practice engagement and intrinsic motivation are anticipated, mediated by timely AI feedback and socio-cognitive scaffolding. Qualitative insights are expected to reveal increased learner autonomy, perceived relevance of material, and higher teacher workload efficiency due to standardized feedback. The study contributes to knowledge by evidencing the viability of AI-driven adaptive pedagogy in resource-constrained settings, integrating ecological validity with scalable ICT interventions, and enriching theoretical discourse on technology-enhanced music education in under-resourced communities. It advances practical understanding of how learning analytics, pedagogy-driven AI, and contextually appropriate interfaces can narrow educational disparities and inform national policy on ICT-enabled music instruction. Conclusions will articulate the platform’s potential as a scalable model for low-resource contexts, with recommendations for (i) policy alignment to subsidize access to devices and offline functionality, (ii) teacher professional development focused on interpreting AI-generated feedback and orchestrating blended instruction, (iii) iterative content expansion to cover additional instruments and genres, and (iv) longitudinal studies to examine long-term pedagogical and psychosocial outcomes.