Pith. sign in

REVIEW 4 major objections 5 minor 42 references

Frame-Level Pansori Mode Classification with Complementary Audio Representations

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Frame-level pansori mode classification generalizes to unseen works, with F1 dropping only 2.1–3.6 points when entire batang are held out.

desk verdict A genuinely useful new corpus and a sensible multi-representation benchmark for pansori mode, but the headline generalization claim rests on an unverified no-leakage condition in the work-level split. read the letter →

arxiv 2608.06633 v1 pith:YLUVY3MD submitted 2026-08-06 cs.SD

Sangheon Park , Seonguk Ju , Suin Chung , Danbinaerin Han , Dasaem Jeong This is my paper · ORCID
classification cs.SD
keywords pansorijomodeclassificationframe-levelannotationmulti-representationaudiomusicinformationretrievalmicrotonalornamentwork-levelsplitsourceseparation
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

This paper tries to establish that a machine can label pansori mode (jo) frame by frame, and that it does so from the musical properties that define the mode rather than from memorized repertoire. Across the three well-represented classes—Ujo, Gyemyeonjo, and Aniri—performance drops only 2.1–3.6 points of F1 when entire batang are held out of training, which the authors read as evidence of genuine mode learning. The paper further claims that different input representations each capture a different theoretical dimension of mode: MIDI carries scale, F0 carries microtonal ornament, and the mel spectrogram carries vocal timbre. A reader should care because it offers a computational handle on a performance tradition where mode is intentionally multidimensional, and where scale alone cannot explain classification.

What carries the argument

The machinery is a multi-representation frame classifier plus a two-way split diagnostic. Four inputs encode different facets of mode—mel spectrogram for timbre and envelope, F0 contour for vibrato and bent notes, MIDI piano roll for the scale skeleton, and a multi-culturally pretrained encoder for generic acoustic priors—all feeding convolutional-recurrent heads that emit per-frame logits over four classes. The diagnostic is the comparison between a Daemok-Shared Split, where the same passage appears in train and test with different singers, and a Work-level Split, where entire daemok are held out; the small gap between the two is the paper's operational test for whether the model learned mode properties instead of memorizing repertoire.

What would settle it

Re-run the Work-level Split with folds that are explicitly singer-disjoint and studio-disjoint, or probe whether the trained models can name the held-out batang via a trivial auxiliary classifier; if held-out performance drops by substantially more than the reported 2.1–3.6 F1 points, the small split gap was not evidence of mode generalization.

Watch

Extended reading notes

Core claim

On a new expert-annotated 46-hour corpus spanning all five surviving batang, four audio representations trained as frame-level classifiers stay within 2.1–3.6 F1 points of their performance on known daemok when entire works are held out, for Ujo, Gyemyeonjo, and Aniri. The paper argues this small split gap shows the models rely on mode-relevant acoustic features rather than song-specific shortcuts, with symbolic pitch and F0 representations generalizing slightly better than spectral ones. It also reports that source separation removes exactly the percussion cue that distinguishes Changjo, and that a multi-culturally pretrained SSL model fails specifically on the Ujo–Gyemyeonjo distinction, suggesting that generic pre-training captures less of the mode's performative dimension. The central discovery is thus not a single architecture but evidence that frame-level mode classification can be learned from complementary representations while remaining interpretable: cross-modal disagreement recovers musicologically documented cases where scale, ornament, and contour pull in opposite directions.

Load-bearing premise

The paper's case for genuine mode learning rests on the assumption that the Work-level Split prevents repertoire memorization: that held-out batang segments share no melodic passages, singers, or recording conditions with the training folds, so a small train–test gap cannot come from leakage.

Editorial extensions

If this is right

  • If the split gap reflects genuine mode learning, frame-level pansori classification can be applied to unseen works, including modern changjak pansori, with the reported F1 levels.
  • Source separation is not harmless for this task: removing percussion erases the cue that defines Changjo, so practical annotation tools should retain the full mix.
  • MIDI and F0 representations are the most portable across repertoire, while spectral and SSL representations are more sensitive to the performative surface; any deployed system should combine symbolic and continuous inputs.
  • Cross-modal disagreement can be read musicologically: when an ensemble of representations contradicts the ground-truth label, the disagreement identifies passages where scale, ornament, and timbre conflict.
  • An ensemble is useful as an analytical instrument rather than as a top scorer; the paper's reported ensemble never beats the best single representation on overall masked macro-F1.

Reading between the lines

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

  • A natural test of the paper's split claim would be to build singer-disjoint and studio-disjoint folds; if F1 gaps grow beyond the reported 2.1–3.6 points, the generalization conclusion would need restating as singer familiarity rather than mode learning.
  • Because each representation specializes in a different dimension of mode, the disagreement patterns could be turned into a weak supervision signal for musicological annotation, flagging frames where human experts should re-listen.
  • The SSL model's failure on Ujo–Gyemyeonjo suggests that multi-cultural pre-training may encode scale statistics but miss culture-specific ornament; fine-tuning on microtonal contours rather than only the final adapter layer might close that gap.
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

4 major / 5 minor

Summary. The paper introduces a 46-hour, frame-level pansori mode classification dataset spanning all five canonical batang, with seven expert-provided mode labels collapsed into four classes. It evaluates four input representations (mel spectrogram, F0 contour, MIDI piano roll, and a CultureMERT SSL embedding) plus a theory-grounded ensemble under two split strategies: Daemok-Shared, which tests generalization across singers on known passages, and Work-level, which holds out entire batang segments. The central claim is that the small F1 gap between the two splits for the three well-represented modes (2.1–3.6 points) shows the models learn mode-relevant features rather than memorizing repertoire, and that cross-modal disagreement patterns recover documented musicological distinctions, including Changjo's dependence on percussion and the failure of generic SSL pretraining on the Ujo–Gyemyeonjo distinction.

Significance. If the main generalization claim holds, this is a substantial contribution: the released frame-level annotation corpus is the first of its scale for pansori, and the multi-representation design is theoretically motivated by the multidimensional definition of jo. The paper also makes concrete, falsifiable claims about modality-specific cues (e.g., Changjo is identified primarily by drum accompaniment, and MIDI/F0 capture scale-theoretic versus micro-pitch dimensions) and provides publicly available code and annotations, which are clear strengths. The qualitative comparison with published score-based analyses of modern changjak pansori is a valuable interpretability check. However, the central quantitative conclusion depends on verification of the no-leakage property of the Work-level Split and on statistical stability of the reported F1 values, neither of which is currently demonstrated.

major comments (4)
  1. [§3.3, Work-level Split] The Work-level Split is intended to rule out repertoire memorization, but the no-leakage condition is not verified. The corpus of 396 tracks spans only five batang, and the same sorikkun and gosu repeatedly appear across batang; recording sessions can also contain multiple batang. Section 3.3 describes partitioning each batang into two segments and assigning danga exclusively to training, but it does not report whether singer identity, accompanist identity, or recording-session/studio conditions are disjoint across folds. If a held-out segment shares a singer or session with training tracks, a model could exploit vocal-timbre or recording-environment shortcuts that transfer across batang, and the 2.1–3.6 point F1 gap reported in Section 5.2 would no longer establish mode-relevant generalization. Please add a metadata-based analysis of overlap across folds or otherwise demonstrate identity-disjoint splits.
  2. [§5.2, Table 2] The central quantitative claim is the small gap between Daemok-Shared and Work-level F1 scores, but every number in Table 2 appears to come from a single training run. No multiple seeds, confidence intervals, or significance tests are reported. Given typical CRNN variance on imbalanced frame-level tasks, a 2–4 point F1 gap can easily fall within run-to-run noise. Please report means and standard deviations over several seeds, or paired bootstrap intervals, for at least the main representations and both split conditions.
  3. [§5.2, Table 2] The results are not contextualized against simple baselines. The class distribution is heavily imbalanced (Gyemyeonjo 58.2%, Ujo 19.0%, Aniri 15.2%, Changjo 1.8% of labeled frames), and the reported masked macro-F1 values (roughly 0.54–0.87) are difficult to interpret without a majority-class, class-prior, or frequency-based baseline. Without such baselines, the absolute accuracy levels and the relative ranking of representations cannot be properly assessed. Please add at least a majority-class and a class-prior baseline to Table 2.
  4. [§3.1 and §6] All frame-level labels were produced by a single expert annotator, and the qualitative validation in Section 6 appears to involve the same co-author with formal pansori training. This raises a circularity concern for the claim that cross-modal disagreement 'recovers musicologically documented phenomena': the annotation and its interpretation share the same expert judgment. Please state explicitly whether the Section 6 reviewer is the annotator, and provide inter-annotator agreement on a subset of frames or a documented independent second-opinion protocol.
minor comments (5)
  1. [Table 2 caption] The abbreviation 'Gmj' is not defined in the caption or text; please spell out 'Gyemyeonjo' in the table header for readability.
  2. [§3.3 and §5.1] Section 3.3 says '18 specific daemok' while Section 5.1 refers to a 'shared test set of 18 pieces'; please clarify whether these are the same items and describe how validation and test pieces are drawn in the Daemok-Shared split.
  3. [Table 1] The percentages in the left panel do not sum to exactly 100.0 because of rounding; please either adjust the values or note that percentages are rounded.
  4. [§5.2] The sentence 'the split gap averages only 2.1–3.6 points of F1 for all four signal-derived representations' is ambiguous because Table 2 shows different gaps per mode; please specify that this is the range across the three well-represented modes.
  5. [§6.2] Patterns A and B are described qualitatively; adding counts of how often each pattern occurs among the reviewed test frames would strengthen the claim of bidirectionality.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: empirical benchmark with no equation-level reduction; the main generalization claim rests on an unverified no-leakage condition, which is a correctness risk, not circularity.

full rationale

This is an empirical supervised classification paper, not a derivation chain; no equation defines a prediction in terms of the target, and no fitted parameter is renamed as a prediction. The four representations are selected a priori from the musicological definition of jo (Section 2) and combined by a fixed uniform average (Section 4.5); the only data-dependent choices are standard model-selection decisions such as checkpointing and adapter-layer selection via validation macro-F1. The principal interpretive step—Section 5.2's claim that the 2.1–3.6 F1 split gap shows mode-relevant generalization—depends on the Work-level Split's no-leakage condition (no shared singers, accompanists, studios, or stock melodic passages across folds). Section 3.3 describes the fold construction but does not report singer/session metadata, so this condition is unverified; that is a correctness risk and a possible cause of the small gap, not a circular reduction, because nothing in the paper forces the gap by construction. Reference [32], by co-author Suin Chung, is used in Section 6.3 as one of two 'independently analyzed' published score-based validations of modern pansori; since the central quantitative claims do not depend on this validation and reference [33] is external, this is at most a minor non-load-bearing self-citation. Under the hard rules, I find no significant circularity; score 2 reflects only that minor self-citation, not a circular derivation.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The framework rests on musicological premises about jo being multidimensional, on the merger of minority modes into Ujo, and on changjo being drum-dependent; these are domain assumptions sourced from the literature and the annotating co-author. No new physical or theoretical entities are introduced. The main fitted quantities are validation-selected hyperparameters, especially the CultureMERT adaptation layer.

free parameters (2)
  • CultureMERT adaptation layer = 8
    Layer 8 was selected empirically by probing all Transformer layers on validation macro-F1. This choice is tuned to the validation set and could inflate reported performance.
  • Per-modality batch size and scheduler patience = not reported
    Selected via validation performance per modality; values are not given, so comparisons across representations are not fully reproducible or controlled.
assumptions (5)
  • domain assumption Pansori mode is defined by the entanglement of pitch collection, sigimsae, vocal timbre, and affect, not by scale alone.
    This musicological premise motivates the multi-representation design and the interpretation of representation disagreement; introduced in Section 2.2.
  • domain assumption Minor modes Pyeongjo, Seollongje, and Gyeongdeureum can be merged into Ujo for classification.
    The merger is justified by musicological convention and class imbalance; Section 3.1.
  • domain assumption Changjo identity is primarily determined by the presence of drum accompaniment.
    Used to interpret the source-separation result in Section 5.2 via reference [31].
  • domain assumption F0 extracted with PESTO and MIDI transcription with the model of Li et al. sufficiently capture mode-relevant pitch and ornament cues.
    Conclusions about sigimsae and scale rely on these pretrained tools being accurate on pansori vocal production; Sections 4.2 and 4.3.
  • domain assumption Frames labeled Unknown can be excluded from training and evaluation without biasing results.
    The paper excludes Unknown frames but does not report their prevalence or distribution; Section 5.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Frame-Level Pansori Mode Classification with Complementary Audio Representations." pith.science (2026). https://pith.science/paper/YLUVY3MD

@misc{pith2026260806633,
  author       = {Pith},
  title        = {Pith review of: Frame-Level Pansori Mode Classification with Complementary Audio Representations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YLUVY3MD}},
  note         = {Machine review of arXiv:2608.06633}
}
read the original abstract

Pansori is a traditional Korean vocal genre whose mode system (jo) is defined not by scale alone but by the entanglement of pitch collection, microtonal ornament (sigimsae), and vocal timbre. In this study, we introduce a 46-hour frame-level pansori mode annotation, expert-labeled across all five canonical batang, and evaluate four complementary input representations (mel spectrogram, F0 contour, MIDI piano roll, and a multi-cultural SSL encoder) under two split strategies designed to detect shortcut learning. Across the three well-represented modes, performance degrades by only 2.1--3.6 points of F1 when entire works are held out, indicating that the models learn mode-relevant features rather than memorizing repertoire. Per-class results further show that source separation removes the percussion cue on which changjo depends, and that generic multi-cultural pre-training fails specifically on the Ujo--Gyemyeonjo distinction. Qualitative analysis of cross-modal disagreement recovers musicologically documented phenomena and agrees with published score-based analyses of modern changjak pansori.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 39 canonical work pages

  1. [32]

    Computational Analysis of Yaredawi YeZema Silt in Ethiopian Orthodox Tewahedo Church Chants

    M. A. Muluneh, Y .-T. Peng, and L. Su, “Computa- tional analysis of yaredawi yezema silt in ethiopian orthodox tewahedo church chants,”arXiv preprint arXiv:2412.18788, 2024

    work page Pith review arXiv 2024

  2. [1]

    Frame-Level Pansori Mode Classification with Complementary Audio Representations

    INTRODUCTION The computational analysis of non-Western music tradi- tions has become an increasingly active area within music information retrieval (MIR), with substantial progress on traditions such as Turkish, Hindustani, and Arabic art music, particularly in the study of their respective mode systems (makam,raga, andmaqam) [1 –3]. In each of these case...

    work page 2026

  3. [2]

    BACKGROUND ON PANSORI SINGING 2.1 Performance Tradition and Vocal Characteristics Pansori is a traditional Korean epic storytelling art, rec- ognized in 2003 as a UNESCO Intangible Cultural Her- itage of Humanity. A performance features a single vocalist (sorikkun) who delivers a long-form narrative through a combination of singing (sori), stylized speech...

    work page Pith review arXiv 2003

  4. [3]

    in Pansori Performance and Theory

    DATASET 3.1 Data Collection and Annotation The corpus comprises 396 tracks (46 h 02 m) spanning all five canonical batang together with danga and miscella- neous pieces, annotated at the frame level by a co-author holding a Ph.D. in Pansori Performance and Theory. Each piece was segmented by mode transitions into seven stan- dard pansori categories:Gyemye...

    work page

  5. [4]

    This formulation re- flects the musicological reality that mode can shift within a singledaemok

    PANSORI MODE CLASSIFICATION We formulate pansori mode classification as a frame-level sequence labeling task: given an audio recording, the model outputs a mode label for each frame. This formulation re- flects the musicological reality that mode can shift within a singledaemok. At training and inference time, recordings are processed in fixed 30-second s...

    work page

  6. [5]

    Death of Lady Gwak

    EXPERIMENTS 5.1 Experimental Setup Training.All models are trained for 10,000 iterations with the Adam optimizer, using aReduceLROnPlateau scheduler (factor = 0.5, lrmin = 1E-6) driven by validation macro-F1. We evaluate on the validation set every 200 it- erations and retain the checkpoint with the highest valida- tion macro-F1 for test evaluation. All m...

    work page arXiv

  7. [6]

    ANALYSIS AND DISCUSSION To examine in concrete musical terms what each modal- ity has learned, a co-author with formal pansori training qualitatively reviewed the held-out test passages, identify- ing cases where specific modalities succeeded while others failed, as well as cases where all models converged on a label that contradicts the ground truth. Tog...

    work page

  8. [7]

    CONCLUSION We presented a frame-level pansori mode classification framework on a 46-hour expert-annotated corpus, using complementary representations each tied to a different di- mension of pansori mode. Performance is stable under a strict work-level split for the well-represented modes, and the disagreements among representations align with expert music...

    work page

Show all 42 references
  1. [8]

    ACKNOWLEDGMENTS This work was supported by the National Research Founda- tion of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2025-00560548)

    2025

  2. [9]

    A classification study for turkish folk music makam recognition using machine learning with data augmentation techniques,

    A. Börekci and O. Sevli, “A classification study for turkish folk music makam recognition using machine learning with data augmentation techniques,”Neural Computing and Applications, vol. 36, no. 4, pp. 1621– 1639, 2024

    2024

  3. [10]

    Raga classification from vocal performances using multimodal analysis

    M. Clayton, P. Rao, N. N. Shikarpur, S. Roychowdhury, and J. Li, “Raga classification from vocal performances using multimodal analysis.” inISMIR, 2022, pp. 283– 290

    2022

  4. [11]

    Analysis of variation of feature extraction methods in the classification of al-qur’an maqam using machine learning,

    M. A. A. Alaydrus and A. Zahra, “Analysis of variation of feature extraction methods in the classification of al-qur’an maqam using machine learning,”Journal of Theoretical and Applied Information Technology, vol. 101, no. 21, pp. 7085–7093, 2023

    2023

  5. [12]

    Two theories of Ujo and Pyeongjo in Pan- sori: Comparison of theories of musical mode in Pansori between Baek Daewoong and Lee Bohyeong,

    E. J. Shin, “Two theories of Ujo and Pyeongjo in Pan- sori: Comparison of theories of musical mode in Pansori between Baek Daewoong and Lee Bohyeong,”Studies in Korean Music, vol. 63, pp. 233–268, 2018

    2018

  6. [13]

    Musical modes and their functions in

    H.-j. Kim, “Musical modes and their functions in "Simcheong-ga" (『심청가』의악조와그기능 ),”Stud- ies in Pansori ( 판소리연구), vol. 4, pp. 181–209, 12 1993

    1993

  7. [14]

    A study on vocal education of pansori singers (판소리창자의발성교육에대하여 ),

    K.-h. Lee, “A study on vocal education of pansori singers (판소리창자의발성교육에대하여 ),”Studies in Pansori (판소리연구), vol. 9, pp. 77–94, 11 1998

    1998

  8. [15]

    Musical characteristics of mode( 선법) and tori(토리) (조 (調)가지시하는선법과토리의개념 ),

    B. Lee, “Musical characteristics of mode( 선법) and tori(토리) (조 (調)가지시하는선법과토리의개념 ),” Studies in Korean Music, pp. 245–271, 2012

    2012

  9. [16]

    Seoul, Korea: National Center for Korean Traditional Performing Arts, 2008

    National Center for Korean Traditional Performing Arts, Korean Musicology Series 2: Pansori. Seoul, Korea: National Center for Korean Traditional Performing Arts, 2008

    2008

  10. [17]

    Study on a few basic terms for building a ‘pansori-science’: Focusing on a comparative study of similar terms (『판소리학』정립을위한몇가지기본 용어고찰:유사용어의비교연구를중심으로 ),

    J.-t. Kim, “Study on a few basic terms for building a ‘pansori-science’: Focusing on a comparative study of similar terms (『판소리학』정립을위한몇가지기본 용어고찰:유사용어의비교연구를중심으로 ),”Stud- ies in Pansori ( 판소리연구), vol. 25, pp. 177–206, 4 2008

    2008

  11. [18]

    Towards leitmotif activity detection in opera recordings,

    M. Krause, M. Müller, and C. Weiß, “Towards leitmotif activity detection in opera recordings,”Transactions of the International Society for Music Information Re- trieval, Nov 2021

    2021

  12. [19]

    Local key esti- mation in classical music recordings: A cross-version study on schubert’s winterreise,

    H. Schreiber, C. Weiß, and M. Müller, “Local key esti- mation in classical music recordings: A cross-version study on schubert’s winterreise,” inICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2020, pp. 501–505

    2020

  13. [20]

    Convo- lutional recurrent neural networks for music classifica- tion,

    K. Choi, G. Fazekas, M. Sandler, and K. Cho, “Convo- lutional recurrent neural networks for music classifica- tion,” in2017 IEEE International conference on acous- tics, speech and signal processing (ICASSP). IEEE, 2017, pp. 2392–2396

    2017

  14. [21]

    Music artist classification with convolutional recurrent neural networks,

    Z. Nasrullah and Y . Zhao, “Music artist classification with convolutional recurrent neural networks,” in2019 International Joint Conference on Neural Networks (IJCNN). IEEE, 2019, pp. 1–8

    2019

  15. [22]

    Explainable deep learning anal- ysis for raga identification in indian art music,

    P. Singh and V . Arora, “Explainable deep learning anal- ysis for raga identification in indian art music,”IEEE Transactions on Audio, Speech and Language Process- ing, vol. 33, pp. 2302–2311, 2025

    2025

  16. [23]

    Deepsrgm- sequence classification and ranking in indian classical music with deep learning,

    S. T. Madhusudhan and G. Chowdhary, “Deepsrgm- sequence classification and ranking in indian classical music with deep learning,” in20th International Society for Music Information Retrieval Conference, ISMIR

  17. [24]

    Pesto: Real-time pitch es- timation with self-supervised transposition-equivariant objective,

    A. Riou, B. Torres, B. Hayes, S. Lattner, G. Hadjeres, G. Richard, and G. Peeters, “Pesto: Real-time pitch es- timation with self-supervised transposition-equivariant objective,”Proceedings of the 26rd International So- ciety for Music Information Retrieval Conference (IS- MIR), 2025

    2025

  18. [25]

    D. Shah, N. Jagtap, P. Talekar, and K. Gawande,Raga Recognition in Indian Classical Music Using Deep Learning, 04 2021, pp. 248–263

    2021

  19. [26]

    Hybrid trans- formers for music source separation,

    S. Rouard, F. Massa, and A. Défossez, “Hybrid trans- formers for music source separation,” inICASSP 2023- 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–5

    2023

  20. [27]

    Specaugment: A simple data augmentation method for automatic speech recog- nition,

    D. S. Park, W. Chan, Y . Zhang, C.-C. Chiu, B. Zoph, E. D. Cubuk, and Q. V . Le, “Specaugment: A simple data augmentation method for automatic speech recog- nition,”Interspeech 2019, p. 2613, 2019

    2019

  21. [28]

    Large-scale midi-based composer classification,

    Q. Kong, K. Choi, and Y . Wang, “Large-scale midi-based composer classification,” 2020. [Online]. Available: https://arxiv.org/abs/2010.14805

    2020 arXiv

  22. [29]

    Composer classification with cross-modal transfer learning and musically-informed augmentation

    D. Yang and T. Tsai, “Composer classification with cross-modal transfer learning and musically-informed augmentation.” inProceedings of the 22rd International Society for Music Information Retrieval Conference (ISMIR), 2021, pp. 802–809

    2021

  23. [30]

    Robust singing voice transcription serves synthesis,

    R. Li, Y . Zhang, Y . Wang, Z. Hong, R. Huang, and Z. Zhao, “Robust singing voice transcription serves synthesis,” 2024. [Online]. Available: https: //arxiv.org/abs/2405.09940

    2024 arXiv

  24. [31]

    Finding tori: Self-supervised learning for analyzing korean folk song,

    D. Han, R. C. Repetto, and D. Jeong, “Finding tori: Self-supervised learning for analyzing korean folk song,” Proceedings of the 24rd International Society for Music Information Retrieval Conference (ISMIR), 2023

    2023

  25. [33]

    A comparative study of pak tong-sil’s and jeong cheol-ho’s <yi jun yeolsa-ga> (박동실과정철호 의 <이준열사가> 비교연구),

    D.-H. Lee, “A comparative study of pak tong-sil’s and jeong cheol-ho’s <yi jun yeolsa-ga> (박동실과정철호 의 <이준열사가> 비교연구),” Master’s Thesis (음악학 석사학위논문), Seoul National University (서울대학교 대학원), 8 2017, department of Music, Korean Music Theory Major (음악과국악이론전공)

    2017

  26. [34]

    Joint detection and classification of singing voice melody using convolutional recurrent neural networks,

    S. Kum and J. Nam, “Joint detection and classification of singing voice melody using convolutional recurrent neural networks,”Applied Sciences, vol. 9, no. 7, p. 1324, 2019

    2019

  27. [35]

    Mert: Acoustic music understanding model with large-scale self-supervised training,

    Y . Li, R. Yuan, G. Zhang, Y . Ma, X. Chen, H. Yin, C. Xiao, C. Lin, A. Ragni, and E. e. a. Benetos, “Mert: Acoustic music understanding model with large-scale self-supervised training,”The Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austri...

    2024

  28. [36]

    Marble: Mu- sic audio representation benchmark for universal eval- uation,

    R. Yuan, Y . Ma, Y . Li, G. Zhang, X. Chen, H. Yin, Y . Liu, J. Huang, Z. Tian, B. Denget al., “Marble: Mu- sic audio representation benchmark for universal eval- uation,”Advances in Neural Information Processing Systems, vol. 36, pp. 39 626–39 647, 2023

    2023

  29. [37]

    Culturemert: Continual pre-training for cross-cultural music representation learning,

    A.-N. Kanatas, C. Papaioannou, and A. Potamianos, “Culturemert: Continual pre-training for cross-cultural music representation learning,” inProceedings of the 26rd International Society for Music Information Re- trieval Conference (ISMIR), 2025

    2025

  30. [38]

    Universal language model fine-tuning for text classification,

    J. Howard and S. Ruder, “Universal language model fine-tuning for text classification,” inProceedings of the 56th Annual Meeting of the Association for Computa- tional Linguistics (V olume 1: Long Papers), 2018, pp. 328–339

    2018

  31. [39]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, and P. Dollár, “Focal loss for dense object detection,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 2980–2988

    2017

  32. [40]

    Musical characteristics and func- tions of p’ansori ch’angjo -focusing on song u-hyang’s pansori

    S. J. Jo and조순자, “Musical characteristics and func- tions of p’ansori ch’angjo -focusing on song u-hyang’s pansori "ch’unhyangga"- (판소리 "창조"의음악적특 성과기능고찰 -성우향판소리「춘향가」를중심으 로-),”Journal of the Society for Korean Historico- Musicology ( 한국음악사학보), vol. 52, pp. 241–276, 2014

    2014

  33. [41]

    A study on jeong cheol-ho’s creative pansori <yi sun-sin yeolsa-ga> (정철호의창작판소리 <이순 신열사가 > 연구),

    S. Chung, “A study on jeong cheol-ho’s creative pansori <yi sun-sin yeolsa-ga> (정철호의창작판소리 <이순 신열사가 > 연구),”Studies in Pansori ( 판소리연구), 2026

    2026

  34. [2019]

    International Society for Music Information Retrieval, 2019, pp. 533–540

    2019

Pith tools

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