Pith. sign in

REVIEW 3 major objections 4 minor 35 references

Where Does AI Innovation Go? Measuring Research Attention Imbalance in AI Music

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read AI music research attention is systematically skewed toward scalable content tasks, leaving education, health, and governance under-supported.

desk verdict A transparent, useful bibliometric framework for AI music, but the headline lag numbers rest on sparse cells and need sensitivity analysis before they carry weight. read the letter →

arxiv 2608.06903 v1 pith:JGMETRSY submitted 2026-08-07 cs.CY

classification cs.CY
keywords AImusicresearchattentionimbalancescientometricsfrontier-methodadoptionlagtechnicalinvestmentresidualeducationhealthmethodologicaldiversity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

AI music research has grown rapidly, but this paper argues its attention is lopsided: technical investment and frontier-method adoption concentrate in tasks with scalable data and standardized benchmarks, particularly music generation, information retrieval, and audio processing, while education, health, and governance receive less technical support and adopt new methods years later. To make this imbalance measurable, the paper introduces the Research Attention Profile, four indicators computed over a corpus of 6,839 publications from 2015 to April 2026, labeled with a joint taxonomy of application tasks and technical method families. The headline numbers are clear: generation adopted the Transformer, diffusion, and foundation-model families with an average lag of 0.33 years, versus 4.33 years for education and 5.00 years for health. A sympathetic reader would take the paper as establishing that the field's methodological progress is channeled not by societal relevance but by task-level capacity to absorb emerging methods.

What carries the argument

The carrying mechanism is the Research Attention Profile, a set of four indicators computed from a jointly labeled corpus of AI music publications. Technical Investment Residual (TIR) compares a task's observed number of technical papers to the number expected from its publication volume and the overall annual share of technical papers. Task-Method Investment Residual (TMIR) performs the same comparison for a specific method family within a task. Normalized Methodological Diversity (NMD) is a normalized Shannon entropy of method proportions within a task, ranging from 0 (one method dominates) to 1 (methods evenly spread). Frontier Method Adoption Lag (FMAL) measures the years between a frontier method's first appearance in the AI music corpus and its first appearance within a given task. Together, the four indicators separate the amount of technical support, its allocation across methods, the structural diversity of that support, and the timing of adoption.

What would settle it

Re-run the four indicators on an expanded corpus that adds education, health, and governance venues beyond the three sources used in the paper—for example, music-education journals, clinical music-therapy journals, and arts-policy venues—and check whether education, health, and governance still show negative TIR and multi-year FMAL values; if their technical support becomes comparable once those venues are counted, the imbalance would not reproduce.

Watch

Extended reading notes

Core claim

The paper's central discovery is that research attention in AI music is systematically uneven, and the unevenness has a structure: content- and model-intensive tasks receive stronger and earlier technical support than socially embedded tasks. Music generation and creation (1,522 papers) and music information retrieval (1,496) dominate the corpus, while governance (163) and health (114) remain limited. Technical Investment Residuals are positive for audio processing, generation, and singing-voice technologies, and negative for education, governance, health, and datasets and benchmarks. Frontier methods are allocated most heavily to generation (diffusion TMIR of 1.317) and MIR (foundation-model TMIR of 0.973), while education, health, and governance show below-expected residuals across all three frontier families. The paper concludes that the diffusion of frontier methods is shaped more strongly by compatibility with scalable datasets and benchmark pipelines than by the relative societal importance of the applications.

Load-bearing premise

The paper assumes that the retrieved corpus of 6,839 publications, the single-primary-label mapping validated on samples of 403–500 papers, and the indicators derived from those labels faithfully represent the field's true research attention; if relevant practice-oriented venues were missed, or if labeling errors correlate with task type, the reported under-support of education, health, and governance could be exaggerated.

Editorial extensions

If this is right

  • The gap between content-oriented and socially embedded applications is structural, not just a matter of volume: even after controlling for publication count, generation and audio processing receive more technical investment than expected while education, governance, and health receive less.
  • Frontier methods diffuse unevenly across tasks, with AI music generation adopting the Transformer, diffusion, and foundation-model families within months on average, while education and health lag by over four years and no diffusion-based health study appears in the study period.
  • Methodological diversity does not compensate for limited investment, as governance shows high diversity (NMD of 0.864) but on only 39 method-mapped papers, reflecting dispersion within a small literature rather than broad support.
  • Closing the gap requires not only model development but task-appropriate datasets, context-sensitive evaluation, longitudinal validation, and interdisciplinary collaboration in education, health, and governance.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The adoption-lag metric, as defined, is sensitive to a method's very first year of appearance, so a single early paper can dominate the number; measuring the year a method reaches a meaningful share of a task's technical papers would test whether generation's 0.33-year lag is a robust early-adoption pattern or an artifact of one paper.
  • The same four-indicator profile could be applied to other AI domains, such as educational technology, clinical AI, or computational law, to test whether a systematic under-support of socially embedded, non-benchmark-driven tasks is a general feature of AI research rather than specific to music.
  • The corpus cutoff in April 2026 may undercount very recent diffusion-based studies in health and governance; re-running the analysis after additional years would clarify whether those areas have a lasting lag or simply a longer publication cycle.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper analyzes 6,839 AI music publications from 2015 to April 2026 and introduces a Research Attention Profile with four indicators: Technical Investment Residual (TIR), Task-Method Investment Residual (TMIR), Normalized Methodological Diversity (NMD), and Frontier Method Adoption Lag (FMAL). Using a joint taxonomy of application tasks and technical method families, with LLM-assisted labeling validated by human annotation, the authors report that technical and frontier-method support is concentrated in content-oriented tasks such as generation, MIR, and audio processing, while education, health, and governance remain under-supported. The headline quantitative finding is that generation adopts frontier methods with an average lag of 0.33 years, versus 4.33 years for education and 5.00 years for health. The paper concludes with a call for a more socially responsive AI music research agenda.

Significance. If the measurement framework is robust, the Research Attention Profile is a useful diagnostic tool for science-of-science studies of applied AI fields, and the paper makes a constructive contribution by tying bibliometric attention to methodological support, diversity, and adoption timing. The strengths are the explicit proportional-allocation baselines, the multi-source corpus, the validation protocol with high reported inter-annotator agreement (Cohen's kappa 0.93–0.98, accuracy above 93%), and the clear statement of the four indicators. However, the quantitative edge of the paper, especially the adoption-lag contrast, is not yet established because the headline statistics are computed from sparse cells using extreme-order statistics and because no uncertainty or sensitivity analysis is reported. The qualitative direction—content-heavy tasks receive more technical attention—is plausible, but the precise residual magnitudes and lag differences require additional support before the central claims can be fully accepted.

major comments (3)
  1. [Section 5.3 / Eqs. (7)–(8)] The headline quantitative result—adoption lags of 0.33 years for generation versus 4.33 and 5.00 years for education and health—rests on first-adoption years in cells with very few papers (health has 114 papers in the full corpus and an even smaller method-mapped subset; governance has 39 method-mapped papers). The first publication year in a cell is an extreme order statistic: a single missing early paper from an unindexed practice-oriented venue, or one LLM mislabel, shifts t_{a,m} by years. Table 1 reports only aggregate Cohen's kappa, accuracy, and macro-F1 over validation samples of 403–500 papers; there are no per-task or per-method error rates, and no confidence intervals or bootstrap estimates are reported for TIR, TMIR, NMD, or FMAL. Section 7 acknowledges that first-adoption years are sensitive to a few early papers, but no sensitivity analysis is provided. This gap is load-bearing because the abstract and discussion state the lag contrast as a main finding; without per-cell error rates and a leave-one-out or bootstrap analysis, the quantitative strength of that finding is not yet established.
  2. [Section 4 / Figure 2] The frontier-method set F is introduced in Eq. (7) only as 'F⊆M', and the three frontier families used in Section 5.3 are never named in the methods; the reader must infer Transformer, diffusion, and foundation models from context. Because FMAL_a is averaged only over the methods in F_a that were adopted, the set's membership directly changes the reported average lags, and unadopted methods are excluded rather than penalized. The manuscript also gives inconsistent taxonomy counts: the abstract says 12 application categories and 11 method families, Section 3.2 says 12 primary categories for each dimension, Figure 2 says 13 application categories and 12 method categories, Table S1 lists A0–A12, and Table S2 lists M0–M11. For NMD's denominator log|M| to be interpretable and for FMAL to be reproducible, the paper must define F and M explicitly and reconcile these counts.
  3. [Section 3.1 / Algorithm 1] The claim that education, health, and governance are 'under-supported' depends on the corpus covering those literatures and on the LLM-assigned primary labels being accurate in the small cells. The retrieval is keyword-based from Semantic Scholar, ISMIR, and arXiv, which may under-represent practice-oriented venues where music education, music therapy, and policy/governance research is published; the limitations paragraph concedes that the corpus 'does not cover all relevant publications' but does not quantify this risk. In addition, the stratified validation in Algorithm 1 validates labels over all tasks pooled, so it cannot rule out that label errors are concentrated in exactly the small categories that drive TIR and TMIR. I would like to see per-category validation statistics and a sensitivity analysis that re-computes the main indicators under a multi-label assignment or under alternative corpus-retrieval rules; the current evidence supports a qualitative direction but not the precise residual magnitudes.
minor comments (4)
  1. [Figure 1] The petal lengths are described qualitatively; please add a legend or a quantitative mapping so the reader can relate petal length to TIR, TMIR, NMD, and FMAL values.
  2. [Algorithm 1] The text sets the acceptance threshold tau_Acc = 0.90, but the pseudocode only checks tau_kappa and tau_F1; either include tau_Acc in the algorithm or remove it from the prose.
  3. [Section 3.3] The LLM used for taxonomy mapping is not identified; Gemini-3.1-Flash-Lite is named only for taxonomy generation. Please state the model, prompt version, and inference settings used for the final label assignment to support reproducibility.
  4. [Figure 5 caption] The caption says 'negative values indicate low expected investment'; this should read 'below-expected investment' to match the definition of TIR in Eq. (2).

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the attention indicators are descriptive residuals and first-adoption lags computed directly from labeled counts, with no fitted parameter or self-citation carrying the argument.

full rationale

The paper's central measurements are not derived from their conclusions. TIR (Eq. 2) is a standardized residual comparing observed technical-paper counts q_{a,y} to an explicit proportional baseline E^tech_{a,y} = n_{a,y} Q_y / N_y; TMIR (Eq. 4) uses the analogous independence baseline; NMD (Eq. 6) is a normalized entropy; FMAL (Eqs. 7-8) is the difference between two first-adoption years read from the labeled corpus. None of these equations contain the headline quantities (e.g., the 0.33 vs 4.33/5.00 year lags) as free parameters or fitted values. The taxonomy was built partly from corpus-based keyword clustering and LLM-assisted generation, but the labels are validated against human annotations (Kappa 0.93-0.98, accuracy 93.55-96.60%, macro-F1 93.91-94.18%), and the indicator formulas are applied to the resulting counts; this is a data-construction and labeling step, not an equation-identity circularity. The only author self-citation ([20], a spiking-neural-network example) appears in the related-work enumeration and is not load-bearing for any indicator. Section 7's admission that 'first-adoption years may be sensitive to a few early papers' is an uncertainty and robustness limitation, not a circularity. The directional imbalance claim is a direct descriptive summary of the computed residuals, so the conclusion is not equivalent to its inputs by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 1 invented entities

The paper adds no physical entities. Its quantitative contribution hinges on hand-set choices (frontier method set, validation thresholds) and on the domain assumption that the corpus and labels are representative. The four indicators are new constructs validated internally on a labeled sample but not against external benchmarks.

free parameters (2)
  • Frontier method set F = {Transformer/Attention, Diffusion, Foundation/Agentic}
    FMAL and the frontier-component of TMIR depend directly on which method families are labeled 'frontier'; the selection appears hand-made, with no data-driven criterion for choosing these three families over others such as graph or neuromorphic methods.
  • Validation thresholds = kappa=0.80, accuracy=0.90, macro-F1=0.90
    Acceptance thresholds in Algorithm 1 determine when LLM labels are used for the full corpus; the thresholds are chosen by hand and are not sensitivity-tested.
assumptions (4)
  • domain assumption Keyword-based retrieval from Semantic Scholar, ISMIR, and arXiv yields a representative sample of AI music research
    Section 3.1; the paper acknowledges in Section 7 that the corpus does not cover all relevant publications, and non-English or practice-oriented venues may be under-represented.
  • domain assumption Single-primary-label assignment adequately represents multi-task and multi-method papers
    Section 3.3 and Section 7: 'Assigning one primary task and method simplifies cross-task and multi-method studies.' This simplification is load-bearing for the residual and diversity indicators.
  • domain assumption Volume-proportional allocation is the correct null baseline for 'expected' technical investment
    Equations (1)-(4); imbalance is measured relative to this baseline, so a task whose appropriate methodology is naturally small would be flagged as under-supported by construction.
  • standard math Normalized Shannon entropy divided by log|M| measures methodological diversity
    Equation (6); the entropy normalization is a standard instrument in diversity and inequality measurement.
invented entities (1)
  • Research Attention Profile with indicators TIR, TMIR, NMD, FMAL
    purpose: Quantify distribution, allocation, diversity, and timing of technical support across application tasks
    The framework is introduced and demonstrated on the same corpus that generated the taxonomy; no external benchmark or independent validation establishes that these indicators capture a ground-truth 'attention imbalance', so the constructs are only internally validated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Where Does AI Innovation Go? Measuring Research Attention Imbalance in AI Music." pith.science (2026). https://pith.science/paper/JGMETRSY

@misc{pith2026260806903,
  author       = {Pith},
  title        = {Pith review of: Where Does AI Innovation Go? Measuring Research Attention Imbalance in AI Music},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JGMETRSY}},
  note         = {Machine review of arXiv:2608.06903}
}
read the original abstract

The rapid growth of artificial intelligence (AI) in music has expanded research from generation and information retrieval to education, health, and governance. Yet this growth does not necessarily imply balanced research attention. Where is research attention directed across diverse music tasks, and how can such imbalance be systematically measured? Existing studies examine AI music from separate technical, application-specific, or bibliometric perspectives, but lack a systematic framework for measuring field-level imbalance. To address this gap, we analyze 6,839 AI music publications from 2015 to April 2026 using a joint taxonomy of 12 application categories and 11 technical method families. We propose the Research Attention Profile, comprising four indicators of technical investment, method allocation, methodological diversity, and frontier-method adoption lag. Results show that technical support is concentrated in scalable, content-oriented tasks, while education, health, and governance remain under-supported. Generation adopts frontier methods after only 0.33 years on average, compared with 4.33 years for education and 5.00 years for health. These findings reveal uneven methodological development and support a more socially responsive AI music research agenda.

Figures

Figures reproduced from arXiv: 2608.06903 by the authors.

Figure 1
Figure 1. Research attention imbalance across representative AI music applications. Petal length indicates research volume, technical and frontier-method investment, adoption speed, and methodological diversity. method diffusion across AI music tasks. • We introduce the Research Attention Profile comprising four key indicators, Technical Investment Residual, Task￾Method Investment Residual, Normalized Methodological Diversity… view at source ↗
Figure 2
Figure 2. Joint taxonomy of AI music research. The upper panel presents the 12 primary technical-method categories, while the lower panel presents the 13 primary application categories. 77 subcategories, covering music information retrieval, music generation, music recommendation, music education, music health, among others. The subcategories further specify the scope of each primary category, with details provided in the sup… view at source ↗
Figure 3
Figure 3. Overview of the AI music research landscape. (A) Annual publication trends from 2015 to 2026, showing the overall growth of research attention. (B) Distribution of application tasks across the AI music literature. (C) Annual publication trends of the application categories. (D) Corpus profile, including publication volume, temporal coverage, taxonomy size, source composition, and technical-annotation status. . (E) D… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Technical investment and method allocation across AI music applications. The left bar panel indicates Technical Investment Residuals (TIR) across application categories. The right heatmap draws Task-Method Investment Residuals (TMIR) across all model families. Positive…
Figure 5
Figure 5. Figure 5: Temporal evolution of Technical Investment Residuals (TIR) across application categories in 2015–2017, 2018–2022, and 2023– 2025. Positive values indicate above-expected technical investment, while negative values indicate low expected investment. whereas education and…
Figure 7
Figure 7. Figure 7: Average Frontier Method Adoption Lag (FMAL) across AI music application categories. Lower values indicate earlier adoption of frontier methods, whereas higher values indicate delayed adoption. guards. The observed concentration therefore appears to reflect not only dif…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 29 canonical work pages

  1. [1]

    The ai index 2025 annual report

    Nestor Maslej, Loredana Fattorini, Raymond Perrault, Yolanda Gil, Vanessa Parli, Njenga Kariuki, Emily Cap- stick, Anka Reuel, Erik Brynjolfsson, John Etchemendy, Katrina Ligett, Terah Lyons, James Manyika, Juan Carlos Niebles,YoavShoham,RussellWald,TobyWalsh,Armin Hamrah, Lapo Santarlasci, Julia Betts Lotufo, Alexan- dra Rome, Andrew Shi, and Sukrut Oak....

  2. [2]

    Cambridgeuniversity press, 2000

    TiaDeNora.Musicineverydaylife. Cambridgeuniversity press, 2000

  3. [3]

    A survey on deep learning for symbolic music generation: Representations, algorithms, evaluations, and challenges.ACM Computing Surveys, May 2023

    Shulei Ji, Xinyu Yang, and Jing Luo. A survey on deep learning for symbolic music generation: Representations, algorithms, evaluations, and challenges.ACM Computing Surveys, May 2023. doi: 10.1145/3597493

  4. [4]

    Natural language processing methods for symbolic music generation and information retrieval: A survey.ACM Computing Surveys, 57(7):1–40, 2025

    Dinh-Viet-Toan Le, Louis Bigo, Dorien Herremans, and Mikaela Keller. Natural language processing methods for symbolic music generation and information retrieval: A survey.ACM Computing Surveys, 57(7):1–40, 2025

  5. [5]

    Deep learning in music recommendation systems.Frontiers in Applied Mathematics and Statistics, 5:44, 2019

    Markus Schedl. Deep learning in music recommendation systems.Frontiers in Applied Mathematics and Statistics, 5:44, 2019

  6. [6]

    The usage of artificial intelli- gence technology in music education system under deep learning.Ieee Access, 12:130546–130556, 2024

    Yinchi Chen and Yan Sun. The usage of artificial intelli- gence technology in music education system under deep learning.Ieee Access, 12:130546–130556, 2024

  7. [7]

    Afirstlookatgenerativeartifi- cial intelligence based music therapy for mental disorders

    Lin Shen, Haojie Zhang, Cuiping Zhu, Ruobing Li, Kun Qian, Wei Meng, Fuze Tian, Bin Hu, Björn W Schuller, andYoshiharuYamamoto. Afirstlookatgenerativeartifi- cial intelligence based music therapy for mental disorders. IEEE Transactions on Consumer Electronics, 2024

  8. [8]

    Bahar Mahmud, Guan Hong, and Bernard Fong. A study of human–ai symbiosis for creative work: Recent developmentsandfuturedirectionsindeeplearning.ACM TransactionsonMultimediaComputing,Communications and Applications, 20(2):1–21, 2023

Show all 35 references
  1. [9]

    Springer, 2020

    Jean-Pierre Briot, Gaëtan Hadjeres, and François-David Pachet.Deep learning techniques for music generation, volume 1. Springer, 2020

  2. [10]

    Solomon Gwerevende and Zama M Mthombeni. Safe- guarding intangible cultural heritage: exploring the syner- gies in the transmission of indigenous languages, dance and music practices in southern africa.International Journal of Heritage Studies, 29(5):398–412, 2023

  3. [11]

    Juan Sebastián Gómez-Cañón, Erick Siavichay, Doga Buse Cavdir, Blair Kaneshiro, Lorenzo Porcaro, et al. Beyond a western center of music information re- trieval: A bibliometric analysis of the first 25 years of ismirauthorship.TransactionsoftheInternationalSociety for Music In...

  4. [12]

    instrumental

    Max V Mathews. The digital computer as a musical instrument: A computer can be programmed to play" instrumental" music, to aid the composer, or to compose unaided.Science, 142(3592):553–557, 1963

  5. [13]

    An expert system for harmonizing four- part chorales.Computer Music Journal, 12(3):43–51, 1988

    Kemal Ebcioğlu. An expert system for harmonizing four- part chorales.Computer Music Journal, 12(3):43–51, 1988

  6. [14]

    Multiple viewpoint systems for music prediction.Journal of New Music Research, 24(1):51–73, 1995

    Darrell Conklin and Ian H Witten. Multiple viewpoint systems for music prediction.Journal of New Music Research, 24(1):51–73, 1995

  7. [15]

    Convolutionalrecurrentneuralnetworks for music classification

    Keunwoo Choi, György Fazekas, Mark Sandler, and KyunghyunCho. Convolutionalrecurrentneuralnetworks for music classification. In2017 IEEE International conference on acoustics, speech and signal processing (ICASSP), pages 2392–2396. IEEE, 2017

  8. [16]

    Findingtemporal structure in music: Blues improvisation with lstm recur- rent networks

    DouglasEckandJuergenSchmidhuber. Findingtemporal structure in music: Blues improvisation with lstm recur- rent networks. InProceedings of the 12th IEEE workshop on neural networks for signal processing, pages 747–756. IEEE, 2002

  9. [17]

    Musegan: Multi-track sequential generative adversarial networks for symbolic music generation and accompaniment

    Hao-Wen Dong, Wen-Yi Hsiao, Li-Chia Yang, and Yi- Hsuan Yang. Musegan: Multi-track sequential generative adversarial networks for symbolic music generation and accompaniment. InProceedings of the AAAI Conference on Artificial Intelligence, volume 32, 2018

  10. [18]

    Music transformer.arXiv preprint arXiv:1809.04281, 2018

    Cheng-Zhi Anna Huang, Ashish Vaswani, Jakob Uszko- reit, Noam Shazeer, Ian Simon, Curtis Hawthorne, An- drew M Dai, Matthew D Hoffman, Monica Dinculescu, and Douglas Eck. Music transformer.arXiv preprint arXiv:1809.04281, 2018

  11. [19]

    Noise2music: Text-conditioned music generation with diffusion models

    Qingqing Huang, Daniel S Park, Tao Wang, Timo I Denk, Andy Ly, Nanxin Chen, Zhengdong Zhang, Zhishuai Zhang, Jiahui Yu, Christian Frank, et al. Noise2music: Text-conditioned music generation with diffusion models. arXiv preprint arXiv:2302.03917, 2023

  12. [20]

    A spiking neural network inspired by neuroscience and psychology forwesternmode-andkey-conditionedmusiclearningand composition.Scientific Reports, 16(1):12956, 2026

    Qian Liang, Yi Zeng, and Menghaoran Tang. A spiking neural network inspired by neuroscience and psychology forwesternmode-andkey-conditionedmusiclearningand composition.Scientific Reports, 16(1):12956, 2026

  13. [21]

    Xia, Huan Zhang, Ilaria Manco, 8 Jiawen Huang, Julien Guinot, Liwei Lin, Luca Marinelli, MaxW.Y.Lam,MeghaSharma,QiuqiangKong,RogerB

    Ying-Chao Ma, Anders Oland, Anton Ragni, Bleiz MacSen Del Sette, Charalampos Saitis, Chris Donahue, Chenghua Lin, Christos Plachouras, Emmanouil Benetos, Elio Quinton, Elona Shatri, Fabio Morreale, Ge Zhang, Gyorgy Fazekas, Gus G. Xia, Huan Zhang, Ilaria Manco, 8 Jiawen Huang,...

  14. [22]

    Notagen: advancing musicality in symbolic music generation with large language model training paradigms

    Yashan Wang, Shangda Wu, Jianhuai Hu, Xingjian Du, YueqiPeng,YongxinHuang,ShuaiFan,XiaobingLi,Feng Yu, and Maosong Sun. Notagen: advancing musicality in symbolic music generation with large language model training paradigms. InProceedings of the Thirty-Fourth International Joi...

  15. [23]

    John Wiley & Sons, 2022

    Alexander Lerch.An introduction to audio content analy- sis: Music Information Retrieval tasks and applications. John Wiley & Sons, 2022

  16. [24]

    Asurveyonmultimodalmusicemotionrecognition.arXiv preprint arXiv:2504.18799, 2025

    Rashini Liyanarachchi, Aditya Joshi, and Erik Meijering. Asurveyonmultimodalmusicemotionrecognition.arXiv preprint arXiv:2504.18799, 2025

  17. [25]

    The psychological basis of musicappreciation: Structure,self,source.Psychological Review, 130(1):260, 2023

    William Forde Thompson, Nicolas J Bullot, and Eliz- abeth Hellmuth Margulis. The psychological basis of musicappreciation: Structure,self,source.Psychological Review, 130(1):260, 2023

  18. [26]

    Psychology Press, 2006

    Irène Deliège and Geraint A Wiggins.Musical creativ- ity: Multidisciplinary research in theory and practice. Psychology Press, 2006

  19. [27]

    Music, health, and well-being: A review.International journal of qualitative studies on health and well-being, 8(1):20635, 2013

    Raymond AR MacDonald. Music, health, and well-being: A review.International journal of qualitative studies on health and well-being, 8(1):20635, 2013

  20. [28]

    Popular music, cul- tural memory, and heritage, 2016

    Andy Bennett and Susanne Janssen. Popular music, cul- tural memory, and heritage, 2016

  21. [29]

    Compu- tational copyright: Towards a royalty model for music generative ai.arXiv preprint arXiv:2312.06646, 2023

    Junwei Deng, Shiyuan Zhang, and Jiaqi Ma. Compu- tational copyright: Towards a royalty model for music generative ai.arXiv preprint arXiv:2312.06646, 2023

  22. [30]

    Kluwer Law International BV, 2025

    Anton Ylikallio.Musical Works, Copyright, and Genera- tive AI: Legal Perspectives on Originality and Authorship. Kluwer Law International BV, 2025

  23. [31]

    Artificial intelligence, post-work and musiclabor.INSAMJournalofContemporaryMusic, Art and Technology, (12):32–45, 2024

    Srđan Atanasovski. Artificial intelligence, post-work and musiclabor.INSAMJournalofContemporaryMusic, Art and Technology, (12):32–45, 2024

  24. [32]

    Fairness of large music models: From a culturally diverse perspective

    Qinyuan Wang, Bruce Gu, He Zhang, and Yunfeng Li. Fairness of large music models: From a culturally diverse perspective. In2024 IEEE 9th International Conference on Data Science in Cyberspace (DSC), pages 706–712. IEEE, 2024

  25. [33]

    Musicians’ ethical concerns about ai: an interview study.AI & SOCIETY, 41(2): 1075–1088, 2026

    Jonathan Herington, Raffaella Borasi, Benjamin J Guer- rero,DavidEMiller,BlaireKoerner,YuJungHan,Zenon Borys, and Rachel Roberts. Musicians’ ethical concerns about ai: an interview study.AI & SOCIETY, 41(2): 1075–1088, 2026

  26. [34]

    Convention on the protection and promotion of the diversity of cultural expressions.https://www

    UNESCO. Convention on the protection and promotion of the diversity of cultural expressions.https://www. unesco.org/en/legal-affairs/convention-pro tection-and-promotion-diversity-cultural-e xpressions, 2005. Adopted in Paris, 20 October 2005. Accessed: 2026-07-07

  27. [35]

    Non-technical / Not model-focused

    Fei Tong, Dongjing Jiang, Qingchong Jiao, Albina Isufi, and Flynnwell Jianfei Zhang. Artificial intelligence in music: a bibliometric and systematic review of creation, performance, and education.Journal of Artificial Intel- ligence and Soft Computing Research, 16(2):185–214, ...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.