Pith. sign in

REVIEW 3 major objections 4 minor 29 references

SKY-Piano is an 11-hour multimodal piano dataset pairing directly measured hand and body motion capture with audio, MIDI, multi-view video, scores, and pseudo-fingering labels for 19 pianists, structured across technique, difficulty, and ex

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 10:01 UTC pith:DFBLMOEY

load-bearing objection A genuinely new and useful dataset—first to combine real hand and body mocap with amateurs, structured repertoire, and synchronized audio/MIDI/video—published as-is it would be a solid resource, but the fingering Tier C validation is circular and the imputed motion stream needs end-to-end validation. the 3 major comments →

arxiv 2607.27296 v1 pith:DFBLMOEY submitted 2026-07-29 cs.SD cs.MM

SKY-Piano: A Multimodal Piano Performance Dataset

Joonhyung Bae , Dawon Park , Taegyun Kwon , Yoon-Seok Choi , Hyeon Hur , Satoshi Obata , Shigeru Kai , Yohei Wada
show 5 more authors
Yu Takahashi Akira Maezawa Jaebum Park Jonghwa Park Juhan Nam
This is my paper
classification cs.SD cs.MM
keywords multimodal piano datasetmotion capturehand mocappiano fingeringMIDI-to-motion generationperformance analysisoptical motion captureamateur vs professional
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper introduces SKY-Piano, an 11-hour multimodal piano performance dataset that synchronizes directly measured hand and body motion capture with audio, MIDI, multi-view video, MusicXML scores, and body-segment kinematics for 19 pianists, 7 professional and 12 amateur. The central claim is that no prior corpus combined real hand and body motion with a repertoire structured to allow controlled comparisons across playing technique, difficulty, and performer expertise. To make the motion useful, the authors include a fingering-labeling pipeline that uses measured fingertip depth to assign each MIDI note to the finger that pressed it, with explicit ambiguity flags and an imputation tier for unresolved cases. They also show the dataset works with an existing MIDI-to-motion generation model by fine-tuning it on their data. If the release holds up, it offers a reference point for piano research that needs true motion rather than motion estimated from video.

Core claim

The paper claims to be the first public dataset to pair real optical motion capture of both hands and body with frame-synchronized audio, MIDI, multi-view video, and score data, across a shared core repertoire that crosses two technique categories, three difficulty levels, and two expertise tiers. The measured fingertip trajectories, rather than video-estimated landmarks, are the enabling element: they let a geometry-based fingering labeler assign each note to a finger by detecting which fingertip sits inside the key and lowest in depth at the onset, reaching 94.5% strict precision on a hand-audited subset while covering all notes via a learned imputation tier. A fine-tuning experiment confi

What carries the argument

The central mechanism is a hardware-synchronized optical motion-capture setup: reflective markers on each hand plus a full-body marker suit, driven by a common timecode and cross-checked with audio-to-MIDI alignment. On top of that, the fingering pipeline operates on measured fingertip z-depth: for each MIDI note it scores every fingertip by whether it falls within the key's 3D rectangle and below a press-depth threshold, resolves ties with z-onset evidence, and imputes leftovers with a bidirectional LSTM, releasing per-note source tags. The dual flagged/imputed release of the motion streams is part of the machinery: users can choose between raw markers with validity flags and self-attention

Load-bearing premise

The load-bearing premise is that the released hand-marker trajectories, after imputation of occluded frames and post-hoc alignment of older sessions, place fingertips accurately enough at the moment of keystrokes that the pseudo-fingering labels and the claim of 'directly measured' motion hold; the paper itself reports median per-trial landmark residuals at the centimeter scale and a 14% median missing-marker rate, so that premise is not guaranteed.

What would settle it

Compute the per-note fingering agreement on the three hand-audited pieces using only the flagged (non-imputed) motion, then compare with the released audit labels: if the subset of notes whose labels came from the imputation tier has an agreement far below the reported 94.5%, or if median fingertip z at MIDI onsets shifts by more than a few millimeters when occlusion flags are removed, the central precision claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Piano motion research can train and evaluate generation models on measured hand and body trajectories instead of video-derived pseudo motion, avoiding centimeter-scale per-joint noise.
  • The shared scale and technique exercises across professionals and amateurs make it possible to isolate expertise effects on keystroke and movement kinematics under matched musical conditions.
  • The fingering labels, released per note with algorithm/imputed/null source tags and explicit ambiguity flags, provide training signal for automatic fingering estimation with known confidence.
  • The synchronized audio, MIDI, video, and score streams enable joint audio-visual transcription and score-following experiments within one corpus.
  • The fine-tuning result indicates that models trained on estimated-motion corpora can adapt to measured motion, but that fingertip-definition mismatches require convention-aware evaluation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The audio-MIDI and motion alignment precision could make SKY-Piano a benchmark for evaluating marker-occlusion imputation: because flagged and imputed versions ship together, anyone can measure how much downstream tasks degrade when synthetic samples are removed.
  • If the centimeter-scale coordinate-alignment residuals for pre-timecode sessions are representative, users working on finger-level analyses may want to restrict to the hardware-synchronized subset; the paper does not yet quantify this subset's share of the 11 hours.
  • The tiered confidence of fingering labels suggests a natural curriculum for fingering-estimation models: train on Tier A/B labels, then fine-tune on Tier C, exploiting the source tags rather than treating all labels as equally reliable.
  • Because the repertoire avoids extended techniques and recent music, a natural extension would add contemporary or jazz material to test whether the measured-motion advantage persists in non-classical idioms.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper introduces SKY-Piano, an 11-hour multimodal piano performance dataset recorded from 19 pianists (7 professional, 12 amateur). The dataset combines optical hand and body motion capture with audio, MIDI, multi-view video, MusicXML scores, and Visual3D kinematics, and is structured so that technique, difficulty, and performer expertise can be compared on shared repertoire. The authors also contribute a geometry-based pseudo-fingering pipeline with an imputation tier, an interactive web explorer, and a fine-tuning experiment with the Tipiano MIDI-to-motion model as a use case. The central claim is that this is the first public corpus combining directly measured hand and body motion capture with synchronized MIDI, score, and a controlled amateur/professional design.

Significance. If the validation gaps are closed, this would be a substantial community resource. The combination of real hand and body optical mocap with MIDI, multi-view video, and structured repertoire is genuinely novel in the MIR dataset landscape. The release of both flagged and imputed motion, per-note source tags for fingering, an interactive explorer, and reproducible pipeline code are concrete strengths. The 2,269-note human audit of fingering labels on the exact released trials is a good reproducibility practice, and the CQT-based audio-to-MIDI alignment plus per-stage manual review add credibility. The fine-tuning case study usefully demonstrates interoperability with an existing MIDI-to-motion model, although the reported key-contact F1 drop needs to be addressed more convincingly. Overall, the dataset has high potential value, but the paper's load-bearing accuracy claims — particularly 'directly measured' sub-centimeter motion and the fingering validation — are not yet fully supported.

major comments (3)
  1. [§3.3 (Coordinate alignment)] The Introduction motivates the dataset by requiring 'sub-centimeter motion data,' and the paper repeatedly describes the motion as 'directly measured.' However, §3.3 states that earlier sessions are temporally synchronized via wrist-velocity cross-correlation and spatially aligned via weighted Procrustes with 'median per-trial landmark residuals at the centimeter scale.' This is not sub-centimeter, and it is larger than the 5 mm z-gap used by the Tier B fingering rule in §5. The manuscript does not report how many sessions used the post-hoc alignment, nor the distribution of alignment errors at the fingertips and at key-press instants. Please report these quantities separately for hardware-synced and post-hoc sessions, and either restrict the sub-centimeter claim to the subset that supports it or provide evidence that key-press fingertip z accuracy is sufficient despite centimeter-scale
  2. [§4 and §3.3 (Imputation)] The released motion has a median per-marker missing-cell rate of 14%, rising to 44% on thumb-carpal pairs, and these gaps are filled by SAITS. No held-out validation of the imputed trajectories is reported. Since the imputed form is the full-coverage version that users will consume and likely the target of the fine-tuning experiment in §6.2, the absence of a measured imputation error — especially for fingertip z at key-press instants — leaves the downstream claims unquantified. Please report SAITS accuracy on held-out occlusions, ideally stratified by marker and by proximity to key-press events, and clarify explicitly whether the fine-tuning MPJPE is evaluated against imputed or flagged trajectories.
  3. [§5 (Fingering validation)] The fingering evaluation has two gaps. First, the 84.3% Tier C accuracy is a self-consistency statistic: the BiLSTM is trained and evaluated on labels produced by the same algorithmic pipeline. The external 2,269-note human audit evaluates only the geometry-based committed subset (94.0% coverage), so the imputed 6% of notes are not externally validated. Second, the keystroke-detection validation uses tolerances of ±50 ms and 15 mm above the key surface, which are much looser than the Tier B assignment thresholds of ±25 ms and ≥5 mm z-gap; therefore the 91.4% figure does not validate the finger-assignment stage at the thresholds actually used. Please report accuracy separately by source tag (algorithm/imputed) on the audited notes, provide confidence intervals for the 94.5% precision, and either audit the imputed notes or explicitly state that their accuracy is unknown.
minor comments (4)
  1. [Table 1] The Video column shows a checkmark without noting that video is only available for the professional cohort. Table 3 indicates amateur and mixed cohorts have no video; please qualify the Table 1 entry (e.g., 'up to 4-view, professional only').
  2. [§6.2 / Table 6] The key-contact F1 drops from 0.93 to 0.66 after fine-tuning while MPJPE improves. The fingertip-versus-MANO-joint explanation is plausible but unverified. Consider adding an evaluation that maps the released markers to the MANO convention, or reporting both conventions explicitly, so users can interpret the fine-tuning result as a compatibility demonstration rather than a regression.
  3. [§5 / Table 5] The text uses both 'Set1 graded' and 'Set 1 graded' (also 'Set 1' in §3.1). Please unify the terminology.
  4. [§8 Acknowledgments] Minor typo: 'Y AMAHA' should be 'YAMAHA'.

Circularity Check

1 steps flagged

Tier C fingering imputation is a self-consistency metric; core dataset and A/B fingering audit are independent.

specific steps
  1. fitted input called prediction [Section 5 'Annotation of Fingering Labels', Tier C paragraph (Figure 4)]
    "Tier C is trained in the masked-fill spirit of the SAITS [26] marker imputation (Section 3.3). Per note it consumes pitch, inter-onset interval, duration, and either the geometry-derived finger label or a mask token. For multi-candidate notes, the softmax is masked to the algorithm’s candidate set, ensuring a geometrically plausible prediction. We train on the modality-complete professional subset (excluding the Mixed cohort), augmented with PIG [15] as a symbolic prior, and recover the algorithm labels at 84.3% on a 10% trial-level holdout (random-chance 10% over 10 classes)."

    Tier C's training labels ('geometry-derived finger label') and its evaluation target ('algorithm labels') are both outputs of the same Tiers A/B pipeline. The reported 84.3% therefore measures Tier C's ability to reconstruct the pipeline's own labels, not agreement with an independent ground truth. Similarly, the claim that Tier C closes coverage to 100% is true by construction because it assigns a label to every residual note. The paper does disclose that these are 'algorithm labels' and ships source tags, and the 2,269-note human audit independently checks the A/B committed subset; still, the 84.3% number is a self-consistency statistic rather than external validation.

full rationale

The core dataset contribution is externally anchored: hand/body motion comes from optical mocap hardware (OptiTrack/Qualisys), audio/MIDI/video are hardware-synchronized via SMPTE timecode where available, and the release includes both flagged (raw) and imputed forms with disclosed missing-cell rates and centimeter-scale alignment residuals. These are accuracy/robustness limitations, not circularity. The pseudo-fingering pipeline is openly adapted from the authors' prior PianoV AM work, and its geometry-based Tiers A/B are checked against an independent 2,269-note human audit; the paper also labels the output 'pseudo' and ships ambiguity flags and source tags, so the central fingering claim does not reduce to a self-citation. Self-citations to PianoV AM and Tipiano are used as method sources and baselines, not as uniqueness theorems or forced-answer justifications. The main circularity-adjacent element is Tier C's 84.3% recovery rate, which is self-consistency on the pipeline's own labels, and the paper's own limitations section acknowledges the labels are pseudo annotations. Overall, the dataset and the A/B human-audited portion are independent; the score reflects the one disclosed but potentially misread self-consistency metric and overlapping self-citations, not a collapse of the central claim into its inputs.

Axiom & Free-Parameter Ledger

4 free parameters · 3 axioms · 0 invented entities

The dataset contribution rests on measurement infrastructure and annotation thresholds rather than fitted scientific constants. The three listed assumptions are the main unpaid premises: imputation fidelity, marker-to-key geometry, and the validity of expertise/difficulty labels. No new physical or conceptual entities are introduced.

free parameters (4)
  • Tier A candidate thresholds = 50%/80% score thresholds
    Inherited from PianoVAM; set by hand to decide single/multi/no-candidate tiers (Section 5).
  • z-onset refinement window = ±25 ms around MIDI onset
    Hand-chosen window for Tier B disambiguation (Section 5).
  • z-gap threshold for multi-candidate resolution = ≥5 mm
    Hand-chosen separation required to assign lowest fingertip as press finger (Section 5).
  • keystroke validation tolerances = ±50 ms and 15 mm above key surface
    Thresholds used to report 91.4% keystroke-detection coverage (Section 5).
axioms (3)
  • domain assumption Optical marker data after SAITS imputation preserves true finger kinematics closely enough for per-note fingering and sub-centimeter analysis.
    The paper reports 14-44% missing marker rates and ships imputed data without independent validation of imputation fidelity (Sections 3.3, 4).
  • domain assumption Marker-to-key 3D geometry and calibrated press-depth thresholds correctly map fingertips to piano keys.
    Keystroke detection relies on the Yamaha key geometry rectangle mapping and z below press threshold (Section 5).
  • domain assumption Professional/amateur skill labels and RCM difficulty levels are valid operationalizations of expertise and difficulty.
    Pianist tiers are assigned by the authors/collaborators; no standardized proficiency test is used (Sections 3.1, 4).

pith-pipeline@v1.3.0-daily-deepseek · 10634 in / 10803 out tokens · 90111 ms · 2026-08-01T10:01:17.207457+00:00 · methodology

0 comments
read the original abstract

Music information retrieval research on piano performance increasingly involves diverse modalities of data and annotations beyond audio and MIDI. We present SKY-Piano, a multimodal piano performance dataset that includes 11 hours of performance recordings of motion, multi-view video, audio, MIDI from 7 professional and 12 amateur pianists along with MusicXML scores. The performance pieces were selected considering playing technique, difficulty, and performer expertise on a shared core repertoire. The motion data include both hand and body motion, released in both flagged form, where samples lost to marker occlusion are marked as unreliable, and imputed form, where those gaps are reconstructed, together with Visual3D body-segment kinematics and other time-synchronized modalities. To easily browse different modalities of data at a glance, we provide an interactive web browser. In addition, we developed a fingering annotation model and tool for deriving pseudo fingering annotations from the MIDI and motion data. Lastly, we present MIDI-to-motion generation through a fine-tuning experiment as a use case of the dataset.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

29 extracted references · 3 linked inside Pith

  1. [1]

    Datasets for the analysis of expressive musical ges- tures,

    A. Sarasúa, B. Caramiaux, A. Tanaka, and M. Ortiz, “Datasets for the analysis of expressive musical ges- tures,” inProc. of the 4th International Conference on Movement Computing (MOCO), ser. MOCO ’17. New York, NY , USA: Association for Computing Machin- ery, 2017, pp. 13:1–13:4

    2017

  2. [2]

    Solos: A dataset for audio-visual music analysis,

    J. F. Montesinos, O. Slizovskaia, and G. Haro, “Solos: A dataset for audio-visual music analysis,” inProc. of the 22nd IEEE International Workshop on Multimedia Signal Processing (MMSP), 2020, pp. 1–6

    2020

  3. [3]

    MOSA: Music motion with semantic annotation dataset for cross-modal music processing,

    Y .-F. Huang, N. Moran, S. Coleman, J. Kelly, S.-H. Wei, P.-Y . Chen, Y .-H. Huang, T.-P. Chen, Y .-C. Kuo, Y .-C. Wei, C.-H. Li, D.-Y . Huang, H.-K. Kao, T.-W. Lin, and L. Su, “MOSA: Music motion with semantic annotation dataset for cross-modal music processing,” IEEE/ACM Transactions on Audio, Speech, and Lan- guage Processing, vol. 32, pp. 4157–4170, 2024

    2024

  4. [4]

    FürElise: Capturing and physically synthesizing hand motion of piano performance,

    R. Wang, P. Xu, H. Shi, E. Schumann, and C. K. Liu, “FürElise: Capturing and physically synthesizing hand motion of piano performance,” inSIGGRAPH Asia 2024 Conference Papers, ser. SA ’24. New York, NY , USA: Association for Computing Machinery, 2024

    2024

  5. [5]

    PianoMo- tion10M: Dataset and benchmark for hand motion generation in piano performance,

    Q. Gan, S. Wang, S. Wu, and J. Zhu, “PianoMo- tion10M: Dataset and benchmark for hand motion generation in piano performance,” inProc. of the 13th International Conference on Learning Represen- tations (ICLR), 2025, spotlight. [Online]. Available: https://openreview.net/forum?id=rxVvRBgqmS

    2025

  6. [6]

    From pose to muscle: Multimodal learning for piano hand muscle elec- tromyography,

    R. Liu, Y . Peng, T. Oku, C.-C. Liao, E. Wu, S. Furuya, and H. Koike, “From pose to muscle: Multimodal learning for piano hand muscle elec- tromyography,” inAdvances in Neural Information Processing Systems (NeurIPS), 2025. [Online]. Avail- able: https://openreview.net/forum?id=ftZEltGArK

    2025

  7. [7]

    PianoV AM: A multimodal piano perfor- mance dataset,

    Y . Kim, J. Park, J. Bae, K. Kim, T. Kwon, A. Lerch, and J. Nam, “PianoV AM: A multimodal piano perfor- mance dataset,” inProc. of the 26th International So- ciety for Music Information Retrieval Conference (IS- MIR), Daejeon, Korea, 2025

    2025

  8. [8]

    Tipiano: Cascaded piano hand motion synthesis via fingertip priors,

    J. Bae, K. Kim, H. Cho, S. Lee, Y .-S. Choi, H. Hur, G. Lee, A. Maezawa, S. Obata, J. Park, J. Park, and J. Nam, “Tipiano: Cascaded piano hand motion synthesis via fingertip priors,” 2026. [Online]. Available: https://arxiv.org/abs/2604.09692

    Pith/arXiv arXiv 2026

  9. [9]

    Separate to collaborate: Dual-stream diffusion model for coordinated piano hand motion synthesis,

    Z. Liu, M. Ou, Z. Xu, J. Huang, H. Han, R. Li, and X. Li, “Separate to collaborate: Dual-stream diffusion model for coordinated piano hand motion synthesis,” inProc. of the 33rd ACM International Conference on Multimedia, 2025, pp. 9743–9752

    2025

  10. [10]

    Sight to sound: An end-to-end approach for visual pi- ano transcription,

    A. S. Koepke, O. Wiles, Y . Moses, and A. Zisserman, “Sight to sound: An end-to-end approach for visual pi- ano transcription,” inProc. of the IEEE International Conference on Acoustics, Speech and Signal Process- ing (ICASSP), 2020, pp. 1838–1842

    2020

  11. [11]

    A CRNN-GCN piano transcription model based on au- dio and skeleton features,

    Y . Li, X. Wang, R. Wu, W. Xu, and W. Chen, “A CRNN-GCN piano transcription model based on au- dio and skeleton features,” inProc. of the IEEE Inter- national Conference on Acoustics, Speech and Signal Processing Workshops (ICASSPW), 2023, pp. 1–5

    2023

  12. [12]

    A two- stage audio-visual fusion piano transcription model based on the attention mechanism,

    Y . Li, X. Wang, R. Wu, W. Xu, and W. Cheng, “A two- stage audio-visual fusion piano transcription model based on the attention mechanism,”IEEE/ACM Trans- actions on Audio, Speech, and Language Processing, vol. 32, pp. 3618–3630, 2024

    2024

  13. [13]

    Multivariate kinematic analysis of piano per- formance: A case study of a professional pianist,

    D. Park, T. Kwon, J. Bae, J. Park, J. Nam, and J. Park, “Multivariate kinematic analysis of piano per- formance: A case study of a professional pianist,” in Proc. of the XXIX Congress of the International Soci- ety of Biomechanics (ISB), Fukuoka, Japan, 2023

    2023

  14. [14]

    Temporal control and hand movement efficiency in skilled music performance,

    W. Goebl and C. Palmer, “Temporal control and hand movement efficiency in skilled music performance,” PLOS ONE, vol. 8, no. 1, p. e50901, 2013

    2013

  15. [15]

    Statistical learning and estimation of piano fingering,

    E. Nakamura, Y . Saito, and K. Yoshii, “Statistical learning and estimation of piano fingering,”Informa- tion Sciences, vol. 517, pp. 68–85, 2020

    2020

  16. [16]

    Automatic piano fingering from par- tially annotated scores using autoregressive neural net- works,

    P. Ramoneda, D. Jeong, E. Nakamura, X. Serra, and M. Miron, “Automatic piano fingering from par- tially annotated scores using autoregressive neural net- works,” inProc. of the 30th ACM International Confer- ence on Multimedia (MM ’22), 2022, pp. 6502–6510

    2022

  17. [17]

    RoboPianist: Dexterous piano playing with deep reinforcement learning,

    K. Zakka, P. Wu, L. Smith, N. Gileadi, T. Howell, X. B. Peng, S. Singh, Y . Tassa, P. Florence, A. Zeng, and P. Abbeel, “RoboPianist: Dexterous piano playing with deep reinforcement learning,” inProc. of The 7th Con- ference on Robot Learning (CoRL), ser. Proceedings of Machine Learning Research, vol. 229. PMLR, 2023, pp. 2975–2994

    2023

  18. [18]

    PianoMime: Learning a generalist, dexterous piano player from internet demonstrations,

    C. Qian, J. Urain, K. Zakka, and J. Peters, “PianoMime: Learning a generalist, dexterous piano player from internet demonstrations,” 2024. [Online]. Available: https://arxiv.org/abs/2407.18178

    Pith/arXiv arXiv 2024

  19. [19]

    PANDORA: Diffusion policy learning for dexterous robotic piano playing,

    Y . Huang, R. Li, and Z. Tu, “PANDORA: Diffusion policy learning for dexterous robotic piano playing,”

  20. [20]

    Embodied hands: Modeling and capturing hands and bodies to- gether,

    J. Romero, D. Tzionas, and M. J. Black, “Embodied hands: Modeling and capturing hands and bodies to- gether,”ACM Transactions on Graphics, vol. 36, no. 6, pp. 245:1–245:17, 2017

    2017

  21. [21]

    Fink,Mastering Piano Technique: A Guide for Stu- dents, Teachers, and Performers

    S. Fink,Mastering Piano Technique: A Guide for Stu- dents, Teachers, and Performers. Pompton Plains, NJ: Amadeus Press, 1992

    1992

  22. [22]

    Toronto, Canada: The Frederick Harris Music Company, 2015

    The Royal Conservatory of Music,Piano Syl- labus, 2015 Edition. Toronto, Canada: The Frederick Harris Music Company, 2015. [On- line]. Available: https://www.rcmusic.com/learning/ examinations/academic-resources-and-policies/ syllabi-and-syllabi-errata

    2015

  23. [23]

    Enabling factorized piano music modeling and generation with the MAESTRO dataset,

    C. Hawthorne, A. Stasyuk, A. Roberts, I. Simon, C.-Z. A. Huang, S. Dieleman, E. Elsen, J. Engel, and D. Eck, “Enabling factorized piano music modeling and generation with the MAESTRO dataset,” inProc. of the 7th International Conference on Learning Representations (ICLR), 2019. [Online]. Available: https://openreview.net/forum?id=r1lYRjC9F7

    2019

  24. [24]

    An industrial-strength audio search al- gorithm,

    A. L.-C. Wang, “An industrial-strength audio search al- gorithm,” inProc. of the 4th International Conference on Music Information Retrieval (ISMIR), 2003, pp. 7– 13

    2003

  25. [25]

    EgoBlur: Responsible innovation in Aria,

    N. Raina, G. Somasundaram, K. Zheng, S. Miglani, S. Saarinen, J. Meissner, M. Schwesinger, L. Pesqueira, I. Prasad, E. Miller, P. Gupta, M. Yan, R. Newcombe, C. Ren, and O. M. Parkhi, “EgoBlur: Responsible innovation in Aria,”arXiv preprint arXiv:2308.13093, 2023

    Pith/arXiv arXiv 2023

  26. [26]

    SAITS: Self-attention- based imputation for time series,

    W. Du, D. Cote, and Y . Liu, “SAITS: Self-attention- based imputation for time series,”Expert Systems with Applications, vol. 219, p. 119619, 2023

    2023

  27. [27]

    Automatic note-level score-to- performance alignments in the ASAP dataset,

    S. D. Peter, C. E. Cancino-Chacón, F. Fos- carin, A. McLeod, F. Henkel, E. Karystinaios, and G. Widmer, “Automatic note-level score-to- performance alignments in the ASAP dataset,”Trans- actions of the International Society for Music Informa- tion Retrieval, vol. 6, no. 1, pp. 27–42, 2023

    2023

  28. [28]

    Automated gap-filling algorithm for kinetic data of finger movement: Pianist hand motion cleaning us- ing spatio-temporal imputation,

    T. Kwon, D. Park, J. Bae, J. Park, J. Park, and J. Nam, “Automated gap-filling algorithm for kinetic data of finger movement: Pianist hand motion cleaning us- ing spatio-temporal imputation,” inProc. of the XXIX Congress of the International Society of Biomechanics (ISB), Fukuoka, Japan, 2023

    2023

  29. [2025]

    Available: https://arxiv.org/abs/2503

    [Online]. Available: https://arxiv.org/abs/2503. 14545