Artificial Intelligence-Assisted Music Education: A Critical Synthesis of Challenges and Opportunities

Abstract

Artificial intelligence (AI) is a hot topic that presents new challenges and opportunities for the improvement of educational processes. The disruptive and transformative force of this new technological development implies the adaptation of educational ecosystems for its use and integration as a didactic and pedagogical resource. From this perspective, a systematic literature review has been conducted to analyze the didactic potential of generative AI tools in the field of promoting artistic creativity in music education. The research results confirm that the incorporation of AI in music education is paving the way for a more personalized, interactive and efficient learning experience. In addition, the analysis suggests nine fundamental fields of IA implementation in music education: virtual and augmented reality (VR; VA); learning personalization, intelligent tutoring systems; composition assistants; improved historical and contextual learning; assessment systems; interactive ear training and music theory systems; tools for music collaboration and performance; and assistive technologies. Furthermore, the challenges presented by the intersection of AI and digital didactics in the field of music education are discussed.

1. Introduction

Artificial intelligence (hereinafter AI) is revolutionizing various facets of each and every one of the areas of knowledge and disciplinary fields that make up human knowledge. These developments are exponential in nature and are transforming in a decisive way the way in which information is produced, transmitted, communicated and received, both in the academic sphere and in people’s ordinary lives. In the field of music, these transformations have been taking place silently in almost any of the dimensions that constitute the access, production, edition and consumption of music in the digital sphere, causing unimaginable changes that are already surreptitiously part of the everyday life of any citizen [1].

The integration and exploitation of this new technological reality in the specific environment of music education have favored the proliferation and development of novel and disruptive tools and methodologies that improve the teaching–learning process, highlighting, among other innovative practices, the potential to personalize learning [2], the possibility of providing real-time feedback on musical practice and auditory training [3,4], the development of virtual assistants for composition [5] or even the creation of automated systems aimed at the evaluation of performance practice, both in expressive and technical aspects [6,7].

This paper aims to identify and analyze, from a critical perspective, the situation of AI in the music education environment, highlighting the challenges and opportunities, as well as the most paradigmatic areas of application, which project the development of AI in the organization and development of teaching–learning processes in music.

2. Methodology

The aim of this article is to provide the reader with an overview of the development of technological advances in educational ecosystems, in particular, the implementation of AI in the field of music education. From this approach, a systematic literature review (SLR) has been proposed to critically observe the state of implementation and development of AI, both from a pedagogical and technological point of view, in the digital didactics of music. As this is a field of research in which multiple disciplines such as music, art education and computer science converge, forming a complex object of study with multiple edges and derivations, a transversal and interdisciplinary approach is proposed that allows for a holistic view of the object of study. This review has been carried out following the updated PRISMA 2020 protocol [8] and has been divided into different phases: (i) formulation of the research questions; (ii) formulation of the search equations; (iii) design of the inclusion and exclusion criteria; (iv) qualitative selection of the analyzed corpus. Likewise, it was decided to limit the search to the Web of Science (WoS) and Scopus databases, as they contain the largest sample of scientific literature of a multidisciplinary nature at present.

2.1. Defining Research Questions

• What is the state of the art regarding the involvement of AI in the field of music education?
• What are the most relevant application areas in the implementation of AI in music education ecosystems?
• What are the new possibilities offered by the integration of AI in schools from a pedagogical, organizational and curriculum development point of view?
• What are the challenges of the intersection of AI and digital didactics in the field of music education?

2.2. Formulation of the Search Equation

The design of the search equations began with several exploratory tests to optimize the trade-off between relevance and completeness of the search results in the selected databases. After various tests and refinement processes, the search equation used was as follows:

TOPIC: (“Artificial Intelligence” OR AI OR “machine learning” OR “deep learning”) AND TOPIC: music AND TOPIC: (education* OR teach* OR pedagog* OR didactics* OR “music curriculum”)

2.3. Inclusion and Exclusion Criteria

After entering the search equation in the databases, with the aim of filtering the results and achieving a representative and relevant sample that condenses the most relevant and significant works, the following inclusion criteria were applied: documents such as articles, monographs and chapters in academic publishers and conference proceedings, all in English, were selected; the search was limited to the period 2018–2024 in order to consider the most up-to-date and cutting-edge scientific advances. On the other hand, we excluded research that did not focus on AI in the field of education, as well as studies focused exclusively on the development of algorithms and web applications because they were excessively technical in nature, and we discarded research that made reference to technological advances in arts education but did not include content on music education. Studies that were in retraction status were also excluded. In addition, duplicates of literature found in the WoS and Scopus databases were eliminated, and a study corpus of n= 328 results was obtained.

2.4. Qualitative Selection of the Final Sample

The qualitative selection phase of the final corpus consisted of a qualitative review by those responsible for the research, which was agreed upon by means of content analysis and the selection of a sample of n = 25, which included the works agreed upon by coincidence between the researchers responsible for the research.

3. Results

The critical reading of the selected bibliography has allowed us to present a panoramic and global vision of the main research currents on artificial intelligence and its implementation in music education, identifying a total of eight areas of intervention where the most relevant and significant advances have been made with respect to the impact of AI in music education. Figure 1 shows these eight areas: virtual and augmented reality (VR; VA); learning personalization, intelligent tutoring systems; composition assistants; improved historical and contextual learning; assessment systems; interactive ear training and music theory systems; and tools for music collaboration and performance.

However, a ninth area—called assistive technologies—of an auxiliary and instrumental nature, related to improving accessibility to resources in the contexts of inclusive education for people with and without disabilities, has been incorporated in a differentiated manner.

Within these broad areas of application identified, two major fields stand out: intelligent tutoring systems (16%) and composition assistants (16%). These were the most represented in the scientific literature, present in 32% of the sample obtained, with the field of improved historical and contextual learning being the least represented, with 4% of the total.

The following

Table 1. Selected articles distributed by field, educational level and didactic potential.

Area	Authors	Educational Level	Didactic Potential of Generative AI
Virtual and augmented reality (VR; VA) 12%	Wang (2021) [9]	Early childhood education	Implementation of virtual reality in early childhood classrooms, as it maximizes children’s mental and thinking capacity by lowering the abstraction level of concepts, processes and ideas
	Cui (2023) [10]	Basic musical skills	Analyzes the possibilities of augmented reality for first steps music performance instruction. Specifically, the acquisition of basic piano skills on mobile devices by means of augmented reality
	Chen (2022) [11]	All educational levels	Develops a music learning system based on virtual reality technology
Learning personalization 12%	Zhang, L. (2023) [12]	All educational levels	Development of the innovative EFDfO (entropy features data fusion optimized) framework, a model that offers personalized musical education to students and improves their academic performance
	Qian (2023) [2]	Higher music education	Personalization and multimodal methods will help increase engagement among college music students
	Xu (2024) [13]	Elementary school	AI-based education system that offers personalized guidance to enhance the interest of music students
Intelligent tutoring systems 16%	Vinze et al., (2021) [14]	Basic musical skills	Assisted vocal music training
	Zheng (2024) [15]	All educational levels	Interactive vocal music education
	Agarwal & Greer (2023) [3]	Conservatories	Flute performance improvement
	Chin & Xia (2022) [16]	Basic musical skills	AI-empowered music tutor with a systematic curriculum design and multimodal feedback
Composition assistance 16%	Ventura (2022) [17]	Conservatories	Harmonization of music basslines
	Huang et al. (2019) [18]	Elementary school	Approachable music composition with machine learning at scale
	Choi (2023) [5]	Elementary school	(Doodle Bach) music creation classes based on creative teaching design
	Knapp et al. (2023) [19]	All educational levels	(Soundtrap) web-based digital audio workstation
Improved historical and contextual learning 4%	Wan (2024) [20]	Higher music education	Improved the effectiveness of music course resources
Automated assessment systems 12%	Hou (2024) [21]	Higher music education	Evaluation of the vocal music teaching process
	Cao (2022) [22]	Basic musical skills	Evaluation of the vocal music teaching process
	Burrows et al. (2018) [6]	All educational levels	Objective evaluation of musical progress in musical performances
Interactive ear training and music theory systems 8%	Cruz et al. (2018) [4]	Basic musical skills	Improved perception of musical structures
	Solanskyi et al. (2024) [23]	Conservatories	Technological adaptation of piano pedagogy; individualized approach to musical learning with platforms such as MusicFlow
Tools for music collaboration and performance 12%	Zhang, W. (2023) [24]	Higher music education	Improved performance of vocal music
	Konecki (2023) [25]	Self-paced learning	Adaptive drum learning system
	Burns and Traube (2020) [26]	All educational levels	Improved the learning of musical instruments, specifically the guitar, and presented the Novaxe online learning platform (OLP)
Assistive technologies 8%	Della Ventura (2019) [27]	Higher music education	Improved the learning of dyslexic students in music education
	Niediek et al. (2018) [28]	Secondary education	Challenges of digital musical instruments and applications in inclusive music education environments

shows the distribution of the corpus of articles analyzed according to their ascription to the areas of intervention mentioned above. Similarly, the most disruptive and innovative didactic and pedagogical principles are identified from the point of view of the development of the musical teaching–learning process that favors new developments and areas of application.

4. Discussion

The results of this research show that, for several years now, there has been a paradigm shift in educational ecosystems thanks to the exponential increase in technological resources that, in parallel, have promoted research and the renewal of pedagogical praxis. In this context, the gradual implementation of artificial intelligence in the classroom has been a major upheaval in the way of understanding education and, particularly, music teaching [29] since, as [13] points out, AI is pushing music education towards a more efficient and intelligent era where resources are personalized, commitment to the discipline is strengthened and new possibilities are proposed for the development of artistic and creative skills.

In this scenario, AI-powered adaptive learning platforms allow self-learners, for example, to assess their level of musical competence development and personalize their learning pace, adapting routines and study habits to their needs and abilities. In this way, platforms such as SmartMusic and Yousician use AI algorithms to create a personalized study plan that evolves as the student progresses, adapting to the student’s particularities, interests and competence level. This personalized approach enhances the educational experience, especially in blended learning environments, and favors the integration of technological resources that adapt methodologies from vocational education to informal learning processes and vice versa. As [2] points out, “because AI-based solutions are flexible, students with different backgrounds and skill levels can benefit from a personalized learning experience” (p. 9) that optimizes study time and the teaching process while fostering the development of artistic and musical personality and individuality.

The personalization of educational processes in the music teaching–learning environment is, to a large extent, fostered by the proliferation and development of digital platforms that integrate resources and tools for intelligent tutoring that allow for the analysis of the student’s performance (and learning performance in general) in real time, offering instant feedback and assessments on crucial aspects such as rhythm, tuning or expressive issues such as dynamics or technical performance. Authors such as [3] highlight, in turn, the double benefit of these systems, since, on the one hand, they reduce the teaching load of music instructors and, on the other hand, they reduce educational costs. Thus, very popular and proven software such as Tonara Version 4.3.98 have, as their most outstanding functionality, the possibility of identifying errors and suggesting corrections, allowing students to self-evaluate and correct themselves in daily practice without the need for the presence of a tutoring agent. In addition, Tonara allows teachers to create very concrete practice objectives, optimizing the students’ home rehearsal process [30], especially in the early stages of instrumental practice, as most children do not yet have the critical spirit that this process requires.

In the field of music composition, the application of AI resources allows music to be composed, generating ideas and motifs or entire pieces from the user’s personal input. These tools not only contribute to the development of the individual’s creativity but, more significantly, favor the contextualized understanding of different compositional techniques, methodologies and styles. In this sense, the ability to emulate different musical styles offers students a scenario unimaginable years ago for the promotion of creative experimentation, the hybridization of genres and styles, and, in short, the cultivation and enrichment of compositional skills. Examples include the Doodle Bach platform designed by Google that allows users to compose baroque chorales [5] or Soundtrap, a digital audio workstation in which musicians compose songs intuitively and collaborate with other users synchronously and asynchronously [19].

Similarly, collaboration and performance are also aided by AI. Virtual accompanists, such as those found in programs like Cadenza and MyPianist, simulate the accompaniment for solo practice, adapting to the student’s tempo and style. This allows users to practice with a “live” accompaniment, making their practice sessions more engaging and realistic. In addition, AI-based collaboration platforms enable remote cooperation between students, facilitating joint projects and group work despite geographical barriers.

Likewise, music theory and ear training, fundamental aspects of music education, benefit significantly from AI. AI-powered interactive learning tools provide engaging exercises for music theory and ear training, offering personalized feedback and dynamically adjusting the difficulty levels to match student progress, like EarMaster 7.0 software for PCs and mobiles. Real-time analysis tools help students understand and analyze music theory concepts while playing or listening to music, reinforcing their theoretical knowledge through practical application. Furthermore, the proposal by [4] is interesting, as they maintain that it is necessary to take advantage of the power of visual representation provided by new AI-based systems to improve the perception of musical structures and musicians’ aural skills, even pointing out that this type of system to improve the understanding of a composition could be extended to other performing arts.

In turn, the role of AI in assessment is particularly noteworthy. Automated marking systems help teachers assess technical accuracy and expression, saving time and ensuring consistency. This is particularly significant in the field of music, as the interpretation of pieces or creativity are highly subjective elements. AI can also analyze large amounts of performance data to provide information on student progress, helping educators to identify areas that need more attention and adapt their teaching strategies accordingly. From this perspective, PosyMus [31] has been developed as an e-assessment system specifically designed for the music classroom that measures the users’ level of skill in rhythmic and melodic practice developed through tablets.

In addition, AI enhances historical and contextual learning in music education. Tools that analyze the history of music can provide students with a deeper understanding of the development of different genres and styles, using visualization and interactive elements to make learning more engaging, such as Folk-RNN, a recurrent neural network trained with traditional music songs [32]. AI-based recommender systems suggest materials and learning pieces based on the learner’s interests and progress [20], fostering a more complete and enjoyable learning experience.

The integration of AI with virtual reality (VR) and augmented reality (AR) technologies creates immersive learning environments [9,10,11]. In this way, these technologies can simulate historical concert experiences, interactive instrument tutorials and virtual classrooms, providing an enriching and engaging learning experience. In this sense, it is worth highlighting the PianoVision 0.1.32 app that integrates virtual reality to offer students a different visual experience in their piano rehearsals or the MuseLab Conductor project through which students will be able to conduct an orchestra by detecting their hand gestures.

Finally, AI-based assistive technologies are making music education more accessible. Through gesture recognition, eye tracking and other adaptive technologies, AI ensures that students with disabilities can participate in music education in a meaningful way. Furthermore, as the researchers who developed the software CAMA (Computer Added Musical Analysis) [27] argue, these systems can help to minimize the excessive fatigue and psychological discomfort experienced by students in the early stages of instrumental practice.

However, despite all the benefits described above, AI also presents some troubling challenges. Some of these challenges are data collection, privacy, transparency in the programming of algorithms, the superficiality of AI responses, the lack of resources at present or other issues derived from an ethical use of these technologies [33,34] that, today, have not yet been addressed in depth by current legislations [35,36]. Moreover, as [37] points out, AI requires huge amounts of data to be trained, and this question becomes complicated when we introduce minors into the equation: is it ethical to capture hundreds of data from the learning process of each of our children and young people, is there the possibility of filtering them, will we create minors who are too dependent on the technologies, what impact do the inevitable biases of algorithms have, will we create minors who are too dependent on the technologies, and what impact do the inevitable biases of algorithms have? These are some of the questions on which the bulk of the literature reviewed agrees, although it is true that opinions are divided in this regard. For example, [37] states that “in the field of education it might be difficult to gather enough data to train an AI” (p. 7), while [19] states that the limitation is not so much the availability of data, but “the creation of analytical methods that adequately use this wealth of data to draw valid and useful conclusions for music learning” (p. 11). For their part, [36] points to the alienation of students’ personalities as a risk factor, which is enhanced by the recommendations of algorithms. However, ref. [12] emphasizes the advantages of working on teaching strategies according to the unique learning profiles of each student. In this sense, in line with [38], the reflection on how, why and for what purpose new technologies are used in educational ecosystems becomes even more relevant in the era of artificial intelligence.

5. Conclusions

The most disruptive potential of generative AI in the field of music education lies in its most essential aspects, which are its ability to create immersive and adaptive learning environments that simulate real-world-like intervention contexts. This enhances experiential and creative learning and is an unprecedented advance over the capabilities and resources of traditional methods. The new capabilities of developing contextual and interactive educational experiences enable more attractive teaching–learning processes adapted to individual needs, capabilities and interests by virtue of the possibility of generating content dynamically in augmented spaces, expanding the limits of interactive education.

Generative AI also opens up new and transformative ways of personalizing learning, allowing the creation of tailor-made educational experiences that adapt to the pace of each learner’s competence development and the possibilities of commitment, dedication or study capacity. AI-assisted educational ecosystems can adjust the content, adapt training plans and develop work routines based on the processing of the information collected in real time and in relation to the systematic practice of students and users, guaranteeing the optimization of time, resources, methodologies and work routines for each student. These new possibilities of personalization of learning highlight the importance of addressing individual differences in education, especially in an area as idiosyncratic as music education, subject to so many psychological, cognitive and/or behavioral particularities and conjunctures. The generation of personalized content in an automated way, unheard of to date, constitutes a very relevant resource to help solve problems such as the lack of engagement of students, offering them tailor-made learning experiences. Similarly, the intelligent tutoring systems associated with these systems represent a particularly significant functionality for the transformation of educational methods and processes in the music learning environment. These systems provide personalized and detailed feedback, simulating individual tutoring experiences with experts in music pedagogy. These systems offer considerable potential for improving student support and informal learning processes outside the classroom, which is paradigmatic of the knowledge society in the digital age [39]. In the promotion of artistic experimentation and development, AI is also a transformative element with enormous didactic potential. In the field of composition support, generative AI offers real-time support to students during the compositional process, providing inputs for inspiration and guidance on the creation and development of formal structures or characterization of genres and styles. These tools reduce the cognitive load and learning curve for students, allowing them to focus on content generation, creative development and critical thinking.

However, future research needs to consider the risks and impact of this technology on student motivation, as this can potentially be undermined by AI such as software like Suno AI Version 3.0, which generates music through written guidelines and results in professional productions of instrumentation and voice with complex harmonies and well-resolved lyrics—compositions that, for a musician, require years of specialization and all kinds of skills and abilities, not only musical but also technological skills related to the field of production. Likewise, from the perspective of the challenges in the field of AI-driven intelligent tutoring systems, ethical and, by extension, legal considerations will play an increasingly important role in the consolidation of a reference framework that clearly establishes the guidelines for action. The generalization and implementation of these systems necessarily require the analysis and impact of the possible biases of tutoring and assessment mechanisms and, more decisively, of the conditions under which these systems are symbiotically integrated into the music classroom.

On the other hand, researchers interested in digital music didactics should also pay attention to the study and analysis for the development and implementation of optimized hybrid models that combine AI with face-to-face teaching activity under the guidance of the music teacher. These new environments should be constituted as spaces in which AI complements the teacher’s task, alleviating the teaching load, optimizing processes and resources and enabling new possibilities for the establishment of curricula and methodologies adapted to the individual as opposed to the traditional system in which the evolution of the classroom establishes the guidelines and patterns of evolution of the teaching activity.

At the same time, it should be stressed that the development of AI-assisted educational systems and platforms, in relation to music education, must peremptorily contemplate the need for the digitization and mass encoding of musical information (specifically the contents written in notated music) as an indispensable requirement to enhance the generation and inference of contents and resources in the system of communication of musical information itself: the musical text or score. If, to date, AI systems take advantage of the enormous amount of knowledge and information that try to analyze the musical question from the epistemological principles of reasoning and narrative discourse in natural language, the future must be dominated by technologies especially focused on generating content and inferring knowledge from the immense amount of knowledge and possibilities that reside in the musical texts (scores) themselves as a paradigmatic form of recording and communicating musical expression.

In conclusion, new generative AI technologies that are applied in the music education environment represent a transformative agent in relation to practices, resources and methodologies that concern various fields of action, from immersive learning environments to the development of highly specialized training plans, assisted by intelligent tutoring systems. The use and optimization of these resources should aim to enable music education that is more oriented towards artistic experimentation and the promotion of creativity while creating more accessible and inclusive environments, continuing and extending the paths of democratization of access to music teaching and learning that digital technologies have been tracing for decades.