Pith. sign in

REVIEW 4 major objections 5 minor 29 references

A Haptic Robot Finger Designed for Guqin Instrument Playing

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 ↗ Review JSON

Pith's one-line read A curved, nail-tipped tactile finger reproduces guqin tones

desk verdict Hardware is real and honestly scoped, but the paper's main design claim is contradicted by its own Table I; fix that and the rest holds up. read the letter →

arxiv 2608.07002 v1 pith:DUHBMIDQ submitted 2026-08-07 cs.RO

Tianwei Zhang , Hanming Yan , Yang Yang. Ziya Wang This is my paper · ORCID
classification cs.RO
keywords HapticSensorDexterousHandHumanoidsGuqinMusicRobotforArtBiomimeticTactileFingertipTactile-Event-TriggeredControl
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

The paper sets out to show that a robot fingertip modelled on the human fingertip—soft curved pad, a nail, a 4x4 static pressure map, and a broadband microphone—can supply the tactile feedback that fretless stringed-instrument playing requires. The authors validate this on a guqin, a seven-string Chinese zither, across open strings, stopped notes, harmonics, and a tactile-triggered bimanual coordination task. Their central result is that the curved nail-equipped fingertip in side contact ('R3-Side') produces acoustic features closest to a live human reference, and that the tactile contact-onset signal can trigger a right-hand pluck with a repeatable delay of 481±56 ms. They are careful to frame the work as validation of the fingertip on selected contact tasks, not as a full guqin performance.

What carries the argument

The load-bearing object is the biomimetic multimodal tactile fingertip itself: a three-layer assembly (silicone encapsulation, porous piezoresistive layer, rigid PCB with interdigitated electrodes and a backside MEMS microphone) shaped with a curved contact surface and a fingernail. The 4x4 static array plays the role of slow-adapting Merkel receptors, and the microphone plays the role of fast-adapting Pacinian corpuscles; the two channels are shown to respond complementarily over the validated 1-100 Hz range. The event-triggered coordination mechanism is a simple threshold on the band-limited rate of change of the summed 16-taxel signal, which detects contact onset and commands the right hand to pluck. The half-nail, half-fingerpad geometry is the design choice that the authors argue lowers friction against the lacquered guqin board while keeping stable string contact, and it is the configuration that yields the closest acoustic match to the human reference.

What would settle it

Mount the fingertip on a shaker alongside a flat-response reference microphone and record the same string press and release; if the contact-onset transient or the harmonic-release transient shows dominant spectral energy above 100 Hz that the fingertip microphone does not reproduce, then the claimed sufficiency of the 1-100 Hz band is falsified and the event-trigger delay would change. Alternatively, an expert guqin-player listening test comparing 120-200 ms release delays against a professional reference would test whether 160 ms is indeed the best parameter.

Watch

Extended reading notes

Core claim

The paper's claim, stated for a fair reader, is that cutaneous tactile sensing at the fingertip is sufficient to close the loop for three representative guqin string-contact skills. The proposed fingertip embeds a 4x4 piezoresistive array for static contact force and a MEMS microphone for dynamic vibration, covered by a curved silicone shell with a biomimetic nail. In controlled comparisons against a live human press, the curved 'half-nail, half-fingerpad' side-contact configuration (R3-Side) was consistently closest to the human reference on log-mel spectral cosine, chroma DTW cost, and envelope correlation, while flat or nail-less designs were more distant. In the bimanual experiment, the aggregated 16-taxel signal's rate of change above a fixed threshold triggered a pre-planned right-hand pluck, giving a mean tactile-onset-to-audio delay of 481±56 ms across 10 trials. For harmonics, a release-delay sweep plus a 10-person listening panel selected 160 ms as the delay most similar to a professional reference, and the authors present this as preliminary parameter selection rather than perceptual validation. The paper's scope is explicitly limited to the tactile fingertip and selected contact tasks, not complete guqin performance.

Load-bearing premise

The load-bearing premise is that the 1-100 Hz calibrated dynamic range captures the contact-onset transient and the harmonic-release cue; if the decisive tactile information lives above 100 Hz, the event-trigger timing and the 160 ms release delay would need to be re-measured with a wider-band sensor.

Editorial extensions

If this is right

  • Robots could move beyond open-loop position control for fretless stringed instruments, using tactile contact-onset events to time the other hand's plucking action.
  • The curved-surface-plus-nail design principle—reducing sliding friction while preserving a soft contact patch—can guide fingertip design for other instruments and for tactile-rich manipulation.
  • The 160 ms harmonic release delay provides a concrete, if preliminary, parameter for robotic harmonics on the guqin.
  • The calibrated static array's pressure resolution (approximately 13.82 kPa over 0-800 kPa and 2.63 kPa over 800-1600 kPa) supports string-pressing without damage, enabling stopped-note and harmonic contact.
  • The sensing architecture could generalize to tactile-rich medical and assistive tasks such as palpation and massage, as the authors note in their limitations.

Reading between the lines

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

  • If the 1-100 Hz band is genuinely sufficient, the same event-triggered contact-onset architecture could transfer to other fretless instruments (guitar, violin, oud) with minimal re-tuning—but this is my inference, since the paper only validates on the guqin.
  • The reported 481 ms delay almost certainly includes manipulator and planning latency, not just sensor latency; separating the two would require a direct impact test of the fingertip against a high-speed reference, which the paper does not report.
  • The harmonic release delay that sounds best may depend on string pitch and pluck position; a fixed 160 ms may be an artifact of the single tested configuration. A pitch-swept listening test would test this.
  • The choice of 160 ms rests on a 10-person non-expert panel; an expert guqin player panel could overturn the parameter, and the authors themselves flag this as preliminary.
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 manuscript describes a biomimetic multimodal haptic fingertip comprising a 4×4 piezoresistive static pressure array and a MEMS microphone dynamic channel, mounted on a UR5 arm and validated on selected guqin string-contact tasks. The authors compare three fingerpad designs (R1 flat, R2 flat-with-nail, R3 curved-with-nail) against a human reference using acoustic similarity metrics; demonstrate tactile-event-triggered bimanual coordination with a mean tactile-to-audio delay of 481±56 ms; and tune harmonic production by press depth and left-hand release delay, reporting a harmonic peak at 8 mm and a preferred 160 ms release delay. The paper explicitly scopes the work as validation of the fingertip on selected tasks rather than a complete guqin performance system, and it lists hardware, sensing, and evaluation limitations in Section V.

Significance. If the results hold, the paper contributes a useful empirical demonstration that cutaneous tactile sensing can support event-triggered bimanual coordination on a fretless stringed instrument, and the static calibration and durability data (Section III-C) provide a solid hardware characterization. The honest statement of limitations and the explicit distinction between event-triggered coordination and continuous tactile servo control are strengths. However, the central design conclusion is not supported by the reported table, and the harmonic results rest on single trials and a small non-expert panel; these gaps must be addressed before the main claims can be accepted.

major comments (4)
  1. [Section IV-A, Table I] The sentence "R3-Side is consistently closest to the human reference across all reported metrics" is not supported by Table I. For the metrics where a human reference can be inferred, R3-Front is closer in |ΔLUFS| (4.014 vs 4.659) and in envelope correlation (0.937 vs 0.869), while R3-Side is closer only in DTW cost (0.087 vs 0.096) and |ΔCentroid| (41.724 vs 49.969), with log-mel tied at 0.988. As written, the design conclusion favoring R3-Side is internally inconsistent with the paper's own table and needs either a corrected statement, a different aggregation that justifies the ordering, or additional evidence.
  2. [Section IV-A, Table I] All entries in Table I are means over 10 trials with no variance or significance testing, so the apparent R3-versus-R2 ordering is unquantified. Please report per-trial standard deviations or confidence intervals and, where appropriate, paired tests across the 10 trials for the key metrics (DTW, envelope correlation, |ΔLUFS|, |ΔCentroid|); without this, the conclusion that R3 is closest to the human reference cannot be distinguished from trial-to-trial noise.
  3. [Section IV-C, Fig. 11] The harmonic "peak at 8 mm" is based on a single trial at each depth (described as 11 trials at different depths), with no repetition or error bars; the text claims a peak and a gradual decline that the data as presented cannot statistically support. The 160 ms release delay was selected by 10 non-expert listeners in a preliminary check, which the paper acknowledges, but the conclusion that the fingertip "reliably executes" harmonics (Section VI) goes beyond this evidence. Please either add repeated trials and a quantitative criterion for the peak, or soften the central claim to match the preliminary evidence.
  4. [Section III-A] The dynamic sensing channel is calibrated only over 1–100 Hz, and the statement that this range is "sufficient for the contact-detection and event-triggering role used in this paper" is an assumption rather than a demonstrated property. Because string-contact transients and harmonic-quality cues may contain energy above 100 Hz, please provide a concrete test (for example, compare event-trigger timing against a wideband reference, or show spectra of the relevant contact transients) or explicitly weaken the claim in the conclusions to "validated up to 100 Hz."
minor comments (5)
  1. [Section IV-A, Fig. 8] The caption says "higher indicates greater spectral similarity" but the heatmap's color scale is not defined; please add a colorbar and specify the metric and value range.
  2. [References] References [13] and [19] are the same paper ("Pluck and play: self-supervised exploration of chordophones for robotic playing"); the duplicate entries should be merged.
  3. [Section IV-B] The fixed threshold for the tactile event trigger is described but its value is not reported; please specify the threshold or the normalization used so that the coordination experiment is reproducible.
  4. [Section IV-C] The phrase "11 trials at different depths" is ambiguous; please clarify whether each depth was tested once or multiple times.
  5. [Section III-A] The term "Azure Kinect V4" appears to be an error; the Azure Kinect DK does not have a V4 model, so the camera model should be corrected.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the guqin experiments are independent empirical measurements; the few self-citations are background and not load-bearing.

full rationale

The paper's central claims are empirical validations rather than derived predictions: Table I reports measured acoustic similarity metrics for three finger structures against a human reference, the harmonic release delay is selected via a listener panel and explicitly labeled as preliminary evidence supporting parameter selection, and the tactile-to-audio delay of 481±56 ms is measured from time-synchronized tactile and audio streams. None of these results is obtained by fitting a parameter and then re-predicting the same data. The fingertip hardware is carried over from the authors' prior work [14]–[15], but that prior work is cited as background on the sensor design, not as proof of the guqin-specific results, and the guqin experiments are new and externally compared to human audio references. There is no equation in the paper that defines a predicted quantity in terms of a fitted input, no imported uniqueness theorem, and no ansatz smuggled in via citation. The only notable textual issue is an internal inconsistency in Section IV-A: the sentence 'R3-Side is consistently closest to the human reference across all reported metrics' is not supported by Table I, where R3-Front is closer on envelope correlation (0.937 vs 0.869) and on |delta LUFS| (4.014 vs 4.659). That is a reporting and supportability flaw, not a circularity. The self-citations present are minor and non-load-bearing, so the circularity score is 1.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no theoretical entities; the central parameters are experimental settings and thresholds. The key assumptions are about sensor bandwidth sufficiency, metric validity, and cross-instrument generalizability.

free parameters (3)
  • Tactile event-trigger threshold = not specified
    The bimanual coordination experiment triggers when the band-limited rate of change of summed taxel intensity exceeds a fixed threshold; the threshold value is not reported, so the timing result (481±56 ms) depends on an unstated hand-set parameter.
  • Left-hand release delay = 160 ms
    Selected via a 10-person perceptual check for the harmonic test; this delay is a tuned task parameter, not a model prediction.
  • Press depth for harmonic peak = 8 mm
    Observed as the depth where harmonic tone peaks in 11 single trials; used as evidence for the pressure-sensitive harmonic behavior.
assumptions (4)
  • domain assumption The 1-100 Hz calibration range of the dynamic channel is sufficient for contact-onset detection and event-triggering in string interactions.
    Stated in Section III-A; string contact transients and vibrations likely extend beyond 100 Hz, but the paper argues the transient onset is the relevant cue. If energy above 100 Hz matters, the tactile event timing may be compromised.
  • domain assumption Acoustic similarity metrics (log-mel cosine, DTW chroma cost, envelope correlation) are valid proxies for perceptual similarity to a human reference.
    Used in Section IV-A to rank finger designs; without validation against listening tests, the ranking may not reflect perceived quality.
  • domain assumption The piezoresistive and microphone channels are sufficiently decoupled to treat them as independent static and dynamic modalities.
    The paper demonstrates minimal coupling on a single-axis platform (Fig. 5e-f), but assumes this holds during real string contact with sliding and vibrato.
  • domain assumption Results obtained on the single guqin instrument used in this study generalize to other handcrafted guqins.
    Section V notes each guqin is handcrafted with unique tonal characteristics, yet the experiments use a single instrument; the results may not transfer to other guqins.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Haptic Robot Finger Designed for Guqin Instrument Playing." pith.science (2026). https://pith.science/paper/DUHBMIDQ

@misc{pith2026260807002,
  author       = {Pith},
  title        = {Pith review of: A Haptic Robot Finger Designed for Guqin Instrument Playing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DUHBMIDQ}},
  note         = {Machine review of arXiv:2608.07002}
}
read the original abstract

With the rapid advancement of humanoid robotics and embodied intelligence technologies, numerous musical instrument-playing robots have emerged in recent years, such as pianos, chime bells, and taiko drums. These robots primarily employ open-loop positional control, rendering them incapable of operating instruments requiring dexterous hands and precise tactile perception, such as a violin, guitar, and guqin. This paper describes the design and validation of a high-precision tactile-sensing finger. By mimicking the shape of the fingertip and fingernail found on a human finger, we develop a biomimetic multimodal haptic fingertip and validate it on selected guqin string-contact tasks, including open-string and stopped-note comparisons, harmonic-tuning, and tactile-triggered bimanual coordination, using the guqin, a traditional Chinese musical instrument, as a challenging validation scenario rather than as a fully demonstrated robotic performance system. This research integrates tactile sensing with robotics technology, thereby contributing to applications in world heritage conservation and cultural dissemination.

Figures

Figures reproduced from arXiv: 2608.07002 by the authors.

Figure 1
Figure 1. The haptic robot finger is designed for guqin (a). The guqin playing [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The Jianzipu is a textbook documenting playing techniques, including the positions, sequence, and hand gestures for plucking the strings. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Schematic diagram of human skin illustrating mechanoreceptors: [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: (a) Static force calibration platform. (b) Dynamic vibration testing [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: (a) Normalised average force–electrical response calibration curve for the 16 static sensing units. (b) Signal response of a single piezoresistive sensing [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: (a) Schematic of the robotic arm pressing a string. (b) Timing diagram [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Waveform–spectrogram comparison of human open strings versus three robotic fingertips under front/side contact configurations (onset-aligned; fixed [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 9
Figure 9. Figure 9: The Haptic sensing feedback in guqin playing. Here are three types [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: (a) is the system simulation, (b) shows the real robot system. (c) A Tactile-Event-Triggered Bimanual Coordination Experiment. [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]
Figure 11
Figure 11. Figure 11: Harmonic playing parameter tuning experiments. (a) and (b) show 11 trials of different sound waveforms and haptic curves corresponding to the [PITH_FULL_IMAGE:figures/full_fig_p010_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 27 canonical work pages

  1. [14]

    Robot embodied dynamic tactile perception of liquid in containers,

    Y . Xu, Y . Ma, W. Lin, Z. Sun, T. Zhang, and Z. Wang, “Robot embodied dynamic tactile perception of liquid in containers,” in2024 IEEE-RAS 23rd International Conference on Humanoid Robots (Humanoids), 2024, pp. 1034–1039. 2

    work page 2024

  2. [15]

    A bioinspired finger for super-resolution dynamic tactile sensing,

    Y . Ma, L. Liang, J. Zhong, Z. Zhu, T. Zhang, and Z. Wang, “A bioinspired finger for super-resolution dynamic tactile sensing,” in 2025 IEEE-RAS 24th International Conference on Humanoid Robots (Humanoids), 2025, pp. 1219–1224. 2

    work page 2025

  3. [1]

    Li,Guqin zongyi [古琴综议: A survey of the guqin]

    X. Li,Guqin zongyi [古琴综议: A survey of the guqin]. Beijing: China Renmin University Press, 2014. 1, 4

    work page 2014

  4. [2]

    Unesco culture sector - intangible heritage - 2003

    UNESCO. Unesco culture sector - intangible heritage - 2003. [Online]. Available: https://web.archive.org/web/20131012192421/http: //www.unesco.org/culture/ich/en/RL/00061 1

    work page arXiv 2003

  5. [3]

    The origin of chinese guqin development,

    H. Li, “The origin of chinese guqin development,”Journal of Social Science and Humanities, vol. 6, no. 9, pp. 5–10, 2024. [Online]. Available: https://doi.org/10.53469/jssh.2024.6(09).02 1

    work page doi:10.53469/jssh.2024.6(09).02 2024

  6. [4]

    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,” inConference on Robot Learning (CoRL), 2023. 2

    work page 2023

  7. [5]

    Dexterous robotic piano playing at scale,

    L. Chen, Y . Zhao, J. Schneider, Q. Gao, S. Guist, C. Qian, J. Kannala, B. Sch ¨olkopf, J. Pajarinen, and D. B ¨uchler, “Dexterous robotic piano playing at scale,”arXiv preprint arXiv:2511.02504, 2025. 2

    arXiv 2025

  8. [6]

    Biomimetic rigid-soft finger design for highly dexterous and adaptive robotic hands,

    N. Zhang, P. Zhou, X. Yang, F. Shen, J. Ren, T. Hou, L. Dong, R. Bian, D. Wang, G. Guet al., “Biomimetic rigid-soft finger design for highly dexterous and adaptive robotic hands,”Science Advances, vol. 11, no. 17, p. eadu2018, 2025. 2

    work page 2025

Show all 29 references
  1. [7]

    Human–robot cooperative piano playing with learning-based real-time music accompaniment,

    H. Wang, X. Zhang, and F. Iida, “Human–robot cooperative piano playing with learning-based real-time music accompaniment,”IEEE Transactions on Robotics, vol. 40, pp. 4650–4669, 2024. 2

    2024

  2. [8]

    Learning to play piano in the real world,

    Y .-S. Zeulner, S. Selvaraj, and R. Calandra, “Learning to play piano in the real world,”arXiv preprint arXiv:2503.15481, 2025. 2

    2025 arXiv

  3. [9]

    Development of anthropomorphic robot finger for violin fingering,

    H. Park, B. Lee, and D. Kim, “Development of anthropomorphic robot finger for violin fingering,”ETRI Journal, vol. 38, no. 6, pp. 1218–1228,

  4. [10]

    Harp plucking robotic finger,

    D. Chadefaux, J.-L. Le Carrou, M.-A. Vitrani, S. Billout, and L. Quartier, “Harp plucking robotic finger,” in2012 IEEE/RSJ international confer- ence on intelligent robots and systems. IEEE, 2012, pp. 4886–4891. 2

    2012

  5. [11]

    Design of an expressive robotic guitarist,

    N. Yang, A. Rogel, and G. Weinberg, “Design of an expressive robotic guitarist,”IEEE Robotics and Automation Letters, vol. 8, no. 11, pp. 7066–7073, 2023. 2

    2023

  6. [12]

    Learning visuotactile skills with two multifingered hands,

    T. Lin, Y . Zhang, Q. Li, H. Qi, B. Yi, S. Levine, and J. Malik, “Learning visuotactile skills with two multifingered hands,” in2025 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2025, pp. 5637–5643. 2

    2025

  7. [13]

    Pluck and play: Self-supervised exploration of chordophones for robotic playing,

    M. G ¨orner, N. Hendrich, and J. Zhang, “Pluck and play: Self-supervised exploration of chordophones for robotic playing,” in2024 IEEE Inter- national Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 18 286–18 293. 2

    2024

  8. [16]

    An artificial neural tactile sensing system,

    S. Chun, J.-S. Kim, Y . Yoo, Y . Choi, S. J. Jung, D. Jang, G. Lee, K.-I. Song, K. S. Nam, I. Younet al., “An artificial neural tactile sensing system,”Nature Electronics, vol. 4, no. 6, pp. 429–438, 2021. 2

    2021

  9. [17]

    Robot manipulation based on embodied visual perception: A survey,

    S. Wang, M. N. Nikoli ´c, T. L. Lam, Q. Gao, R. Ding, and T. Zhang, “Robot manipulation based on embodied visual perception: A survey,” CAAI Transactions on Intelligence Technology, vol. 10, no. 4, pp. 945– 958, 2025. 3

    2025

  10. [18]

    Incremental polyphonic audio to score alignment using beat tracking for singer robots,

    T. Otsuka, T. Takahashi, H. G. Okuno, K. Komatani, T. Ogata, K. Mu- rata, and K. Nakadai, “Incremental polyphonic audio to score alignment using beat tracking for singer robots,” in2009 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2009, pp. 2289–2296. 3

    2009

  11. [19]

    Pluck and play: Self-supervised exploration of chordophones for robotic playing,

    M. G ¨orner, N. Hendrich, and J. Zhang, “Pluck and play: Self-supervised exploration of chordophones for robotic playing,” in2024 IEEE In- ternational Conference on Robotics and Automation (ICRA), 2024, pp. 18 286–18 293. 3

    2024

  12. [20]

    Six degree-of-freedom haptic simulation of a stringed musical instrument for triggering sounds,

    D. Wang, X. Zhao, Y . Shi, Y . Zhang, and J. Xiao, “Six degree-of-freedom haptic simulation of a stringed musical instrument for triggering sounds,” IEEE Transactions on Haptics, vol. 10, no. 2, pp. 265–275, 2017. 3

    2017

  13. [21]

    We can do more to save guqin: Design and evaluate interactive systems to make guqin more accessible to the general public,

    M. Yu, M. Zhang, C. Yu, X. Ma, X.-D. Yang, and J. Zhang, “We can do more to save guqin: Design and evaluate interactive systems to make guqin more accessible to the general public,” inProceedings of the 2021 CHI Conference on Human Factors in Computing Systems, 2021, pp. 1–12. 3

    2021

  14. [22]

    How fast can a robotic drummer beat using dielectric elastomer actuators?

    S. Wakle, T.-H. Lin, S. Huang, S. Basu, and G.-K. Lau, “How fast can a robotic drummer beat using dielectric elastomer actuators?”IEEE Robotics and Automation Letters, vol. 9, no. 3, pp. 2638–2645, 2024. 3

    2024

  15. [23]

    Robot drummer: Learning rhythmic skills for humanoid drumming,

    A. A. Shahid, F. Braghin, and L. Roveda, “Robot drummer: Learning rhythmic skills for humanoid drumming,”arXiv preprint arXiv:2507.11498, 2025. 3

    2025 arXiv

  16. [24]

    Recognition of radicals of guqin music notation by yolos,

    M. Hayami, S. Kuremoto, M. Koshiba, T. Kuremoto, and S. Mabu, “Recognition of radicals of guqin music notation by yolos,” inIn- ternational Conference on Multimedia Information Technology and Applications. Springer, 2025, pp. 118–124. 4

    2025

  17. [25]

    Cnn-based optical music recognition of handwritten guqin scores with advanced dripping algorithm,

    L. Zhao, T. Shen, H. Luo, M. Liu, and W. Xu, “Cnn-based optical music recognition of handwritten guqin scores with advanced dripping algorithm,”Available at SSRN 5235395. 4

  18. [26]

    Abenics: Active ball joint mechanism with three-dof based on spherical gear meshings,

    K. Abe, K. Tadakuma, and R. Tadakuma, “Abenics: Active ball joint mechanism with three-dof based on spherical gear meshings,”IEEE Transactions on Robotics, vol. 37, no. 5, pp. 1806–1825, 2021. 9

    2021

  19. [27]

    From audio encoders to piano judges: Benchmarking performance understanding for solo piano,

    H. Zhang, J. Liang, and S. Dixon, “From audio encoders to piano judges: Benchmarking performance understanding for solo piano,” in Proceedings of the International Society for Music Information Retrieval Conference (ISMIR), San Francisco, USA, 2024. 10

    2024

  20. [28]

    Audio-based piano performance evaluation for beginners with convolutional neural network and attention mechanism,

    W. Wang, J. Pan, H. Yi, Z. Song, and M. Li, “Audio-based piano performance evaluation for beginners with convolutional neural network and attention mechanism,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 29, pp. 1119–1133, 2021. 10

    2021

  21. [29]

    Jiang and C

    Y . Jiang and C. Cannam. Piano precision. [Online]. Available: https://github.com/yucongj/piano-precision 10 Tianwei Zhangis an Associate Research Scientist at Shenzhen Institute of Artificial Intelligence and Robotics for Society, The Chinese University of Hong Kong, Shenzhen...

    2013

Pith tools

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