Pith. sign in

REVIEW 3 major objections 6 minor 33 references

Separate-and-Detect: Unified Drum Transcription and Stem Generation via Latent Diffusion

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A separate-and-detect drum transcription pipeline, built on five-stem latent diffusion separation, outperforms a U-Net separation baseline in overall F1 and beats an end-to-end model on kick and snare, while also generating editable audio…

desk verdict A genuinely new generative separate-and-detect pipeline, but the headline variant is selected on the test sets and the Demucs front-end shift is unquantified, so the main comparisons are weaker than they look. read the letter →

arxiv 2608.01093 v1 pith:G3563W6L submitted 2026-08-02 cs.SD

Wei-Han Hsu , Chih-Cheng Chang , Bo-Yu Chen , Li Su , Yi-Hsuan Yang This is my paper · ORCID
classification cs.SD
keywords automaticdrumtranscriptionsourceseparationlatentdiffusionstemsonsetdetectiongenerativeaudioseparate-and-detectstemgeneration
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 is trying to establish that automatic drum transcription does not have to be a direct mixture-to-events mapping: separating the audio into five drum stems first, then detecting onsets per stem, can match or beat end-to-end models on the core kick and snare while also returning editable audio. On two public benchmarks the proposed latent-diffusion separator with an onset-training branch raises overall F1 over a U-Net separation baseline and beats an end-to-end ADT system on kick and snare F1. The design matters because it turns transcription into an interpretable pipeline whose intermediate products—five drum stems—are useful for editing, remixing, and production. The paper also shows that training-only auxiliary heads can steer the separator without adding inference cost.

What carries the argument

The central object is a five-stem latent diffusion separator adapted from the MSG-LD framework: a VAE encodes mel spectrograms of the drum mixture and target stems into a compact latent space, a U-Net denoiser conditioned on the mixture latent and five learnable track tokens predicts the target stem latents, and a HiFi-GAN vocoder renders the decoded mel spectrograms into waveforms. Two training-only auxiliary branches attach to the penultimate decoder feature map: an onset branch (OB) predicts frame-level onset logits to encourage sparse percussive structure, and a timbre branch (TB) predicts seven DrumGAN timbral descriptors. Both branches shape the separator during training and are discarded at inference, while a fixed CNN-based onset detector converts each separated waveform into symbolic events.

What would settle it

Separate the same MDB and ENST mixtures with the fixed music-separation front end, feed both its output and the original drum-only mixes to the frozen MSG-LD separator, and compare overall transcription F1; a large drop on the front-end side would show the reported gains depend on that unexamined distribution shift.

Watch

Extended reading notes

Core claim

The paper claims that a separate-and-detect pipeline built on a five-stem latent diffusion separator is a practical alternative to direct end-to-end drum transcription. Trained on synthetic and real drum multitracks, the separator jointly generates kick, snare, toms, hi-hats, and cymbals in a VAE latent space; a fixed onset detector converts each stem to events. On MDB Drums and ENST-Drums, the +OB variant achieves higher overall F1 than the U-Net-based LarsNet baseline (0.707 vs 0.613 on MDB; 0.640 vs 0.493 on ENST) and beats the end-to-end ADTOF baseline on kick and snare F1 on both sets (kick 0.931 vs 0.851 on MDB and 0.821 vs 0.795 on ENST; snare 0.760 vs 0.752 on MDB and 0.642 vs 0.599 on ENST), while also outputting separated audio stems.

Load-bearing premise

The separator is trained only on clean drum mixtures paired with isolated stems, but full songs reach it through a fixed music-separation front end whose artifacts are never measured, so the end-to-end transcription numbers rest on the unexamined assumption that that front end's output resembles clean drum mixtures.

Editorial extensions

If this is right

  • With the same frozen onset detector, the latent diffusion separator yields higher overall transcription F1 than the U-Net separation baseline on both MDB Drums and ENST-Drums.
  • The pipeline beats an end-to-end ADT system on kick and snare F1 under the evaluation protocol, while additionally providing editable audio stems that direct-prediction systems cannot produce.
  • Enabling the onset branch gives the most stable transcription gains among the auxiliary branches, and the timbre branch changes the trade-off between reconstruction quality, acoustic stem quality, and onset detection.
  • Transcribing dense broadband textures such as hi-hats and cymbals remains the main difficulty for separation-based methods, so future gains likely require higher-resolution latent representations or better vocoders.
  • Because the auxiliary branches add less than 0.06% of trainable parameters and are removed at inference, the transcription benefit comes with no added inference overhead.

Reading between the lines

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

  • If the fixed music-separation front end used at inference were trained or fine-tuned to match the separator's input distribution, full-mixture gains could be larger than reported; the paper only flags this as future work.
  • The observed divergence between FAD-based audio quality and transcription F1 suggests that distributional audio metrics alone would mislead practitioners choosing a front end for transcription; task-based evaluation is necessary.
  • The 16 kHz sampling rate and about 25 seconds of diffusion sampling per 10-second clip are practical barriers for hi-hat/cymbal fidelity and real-time use; faster sampling or distillation could change the comparison with single-pass end-to-end models.
  • Because the onset branch specifically shapes sparse percussive structure, similar training-only auxiliary supervision could transfer to other percussive instruments or multi-instrument separation tasks where onset detectability matters.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes a separate-and-detect pipeline for automatic drum transcription and stem generation. A Demucs front end extracts a drum-dominant signal, a five-stem latent diffusion separator adapted from MSG-LD generates kick, snare, toms, hi-hats, and cymbals in a VAE latent space, and a fixed onset detector converts each stem into symbolic events. Two training-only auxiliary branches (onset branch OB and timbre branch TB) are introduced to shape the separator without adding inference-time parameters. The system is trained on StemGMD and IDMT-SMT-Drums and evaluated on MDB Drums and ENST-Drums against LarsNet and ADTOF. The main claims are consistent improvement over LarsNet in overall transcription F1, superior kick/snare F1 relative to ADTOF, and the additional availability of editable audio stems.

Significance. If the results hold, the paper offers a useful demonstration that a generative source separator can serve as a practical front end for interpretable drum transcription, while also producing editable stems. The manuscript has several strengths: it uses two standard evaluation datasets, a unified onset-detection protocol for all separation-based baselines, publicly available code and demos, and a Wilcoxon significance test for the main comparison. The training-only auxiliary branches are a clean and inexpensive way to inject task-specific inductive bias. However, the confirmatory value of the headline claims is weakened by model selection on the test sets and by an unexamined distribution shift between training inputs and full-mixture inference inputs.

major comments (3)
  1. [Sec. 5.2, Table 4] The proposed configuration MSG-LD (+OB) is selected because it achieves the highest mean overall F1-score on the MDB Drums and ENST-Drums test sets, but no validation split is used for model selection. Table 4 shows that the OB gain over Vanilla is inconsistent: overall F1 improves by +0.018 on MDB (0.707 vs. 0.689) but decreases by -0.005 on ENST (0.640 vs. 0.645), with no significance test or confidence interval. The Wilcoxon p<0.01 reported in Table 3 for +OB versus LarsNet is therefore a post-selection comparison, and the claim that OB provides the most stable transcription gains is not supported by the reported numbers. Please select the auxiliary-branch configuration on a held-out validation split or, if the ablation is intended as exploratory, state this explicitly and avoid confirmatory language and post-selection p-values. The same issue affects the headline comparison to ADTOF in Table 3, which is reported only for the selected +OB variant.
  2. [Sec. 3.1 and Sec. 5.3] The separator is trained only on drum mixtures paired with isolated stems, yet during full-mixture inference it receives a Demucs-extracted drum-dominant signal, which is a different input distribution. The paper does not measure how Demucs artifacts degrade the separator; the separation-quality metrics in Tables 1 and 2 do not appear to be computed on Demucs outputs, and the issue is acknowledged only as a future-work item. Because the end-to-end transcription comparison in Table 3 rests on this unexamined distribution shift, please add an experiment that quantifies its effect: either report mel-MSE/FAD for the separator with clean and Demucs-processed inputs, or evaluate transcription with and without the Demucs front end. Without this, the reader cannot tell how much of the reported advantage comes from the latent diffusion separator itself rather than from the upstream front end.
  3. [Table 3 caption] The Wilcoxon signed-rank test is reported as p<0.01 for overall F1 on 'evaluation segments,' but the number and definition of the segments, the pairing across systems, and whether the test accounts for the multiple MSG-LD variants are not given. Please report these details. In addition, because the test compares only the selected +OB variant after inspecting the test sets, the p-value should be adjusted for model selection or computed on a held-out split.
minor comments (6)
  1. [Abstract] The abstract states 'Trained on synthetic drum multitracks,' but Sec. 4.1 describes training on both StemGMD (synthesized) and IDMT-SMT-Drums (real-world acoustic); please adjust the wording to reflect both subsets.
  2. [Sec. 3.2] The phrase 'a 1×1×13D convolution' is ambiguous; please clarify the intended kernel shape and dimension ordering.
  3. [Table 3] Per-class F1 values are reported without any variance measure; with only 11 test tracks on MDB, reporting segment-level confidence intervals or per-track ranges would improve interpretability.
  4. [Table 4] The header 'MSG-LD MSG-LD (+OB) MSG-LD (+TB) MSG-LD (+OB+TB)' is repetitive; consider using a 'Variant' column for clarity.
  5. [Sec. 5.2] The sentence 'Our approach also outperforms ... in transcribing kick and snare across both datasets' should note that the overall F1 is below ADTOF on both datasets, so the claim is class-specific rather than a general superiority claim.
  6. [Sec. 5.1] There is an extra space in 'Fréchet Audio Distance (F AD)'; it should read 'Fréchet Audio Distance (FAD)'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the transcription claims are assessed against external benchmarks and baselines with a frozen detector, and the acknowledged test-set model selection is a methodological bias, not a circular reduction.

full rationale

The central claims are empirical comparisons against external baselines (LarsNet, ADTOF) on unseen test sets (MDB Drums, ENST-Drums). The onset detector's hyperparameters are tuned only on the MDB training split and then frozen for all test evaluations (Sec. 4.5). The separator is trained on a disjoint corpus (StemGMD plus IDMT-SMT-Drums), not on the evaluation data, and no fitted parameter is used to define the reported F1 values. The +OB variant is selected after inspecting test-set results in Table 4, which is a real methodological weakness in model selection, but it does not make the reported numbers equivalent to the inputs by construction. The only self-citation ([15]) is a related-work mention of large-scale audio-to-MIDI data and is not load-bearing. The acknowledged limitation that the Demucs front end produces signals different from the clean drum mixtures seen in training (Sec. 5.3) is an unexamined distribution shift and a correctness risk, not a circular inference. Therefore no circular step meeting the required quote-and-reduction standard is present.

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

No new physical or conceptual entities are introduced. All components are drawn from prior work; the auxiliary branches are architectural extensions, not new entities with external falsifiable handles.

free parameters (3)
  • Auxiliary loss weights lambda_on and lambda_tim = lambda_on = 0.05, lambda_tim = 0.5, with a 2,000-step linear warm-up
    Hand-chosen to balance loss magnitudes; they control how strongly the onset and timbre branches shape the separator and are central to the ablation claims.
  • Focal BCE hyperparameters alpha and gamma = alpha = 0.25, gamma = 2.0
    Fixed for the onset branch; standard focal loss settings chosen without reported tuning of these values.
  • Onset detector peak-picking thresholds and minimum inter-onset intervals = thresholds 0.35 to 0.60; min inter-onset intervals 30 to 50 ms, class-specific
    Fitted once on the MDB Drums training split with Optuna and then frozen for all test evaluations, so these values directly shape the reported F1 scores.
assumptions (4)
  • domain assumption Demucs drum-dominant estimates at inference resemble the clean drum-mixture training distribution closely enough for the MSG-LD separator to work unmodified.
    The separator is trained on drum-only mixtures; full-mixture inference passes Demucs output through it, and the paper lists robustness to upstream separation artifacts only as future work (Sec. 5.3).
  • domain assumption Synthetic StemGMD data plus 2.1 hours of IDMT-SMT-Drums transfers to real MDB Drums and ENST-Drums recordings.
    Training data is mostly synthetic; evaluation is on real-world recordings, and the transfer is validated only by the final metrics.
  • domain assumption Time alignment of the VAE mel-latent and HiFi-GAN vocoder is preserved within the 50 ms tolerance needed for onset matching.
    The diffusion model reconstructs mel spectrograms then waveforms, and no timing-error analysis is given to show that the 10 ms frame hop and neural vocoder preserve onset positions.
  • domain assumption FAD and mel-MSE are adequate substitutes for waveform-domain separation metrics like SDR and SI-SDR.
    The paper declines to report SDR and SI-SDR because of vocoder effects; this choice affects how separation quality is interpreted.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Separate-and-Detect: Unified Drum Transcription and Stem Generation via Latent Diffusion." pith.science (2026). https://pith.science/paper/G3563W6L

@misc{pith2026260801093,
  author       = {Pith},
  title        = {Pith review of: Separate-and-Detect: Unified Drum Transcription and Stem Generation via Latent Diffusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G3563W6L}},
  note         = {Machine review of arXiv:2608.01093}
}
read the original abstract

Automatic Drum Transcription (ADT) is commonly formulated as a direct mapping from a music mixture to symbolic drum events. While effective for transcription, this formulation discards the acoustic stems that are useful for editing, remixing, and production. We revisit an alternative separate-and-detect formulation, where a drum source separation front end first produces five editable drum stems, and a fixed onset detector then converts each stem into symbolic events. The separator is built on a five-stem latent diffusion model that jointly generates kick, snare, toms, hi-hats, and cymbals in a compact VAE latent space. We further study two training-only auxiliary branches--an onset branch (OB) and a timbre branch (TB)--which shape the separator during learning but are discarded at inference. Trained on synthetic drum multitracks and evaluated on MDB Drums and ENST-Drums, the proposed pipeline consistently improves over a strong U-Net-based drum separation baseline in overall transcription F1. It also outperforms a representative end-to-end ADT system on kick and snare F1 under our evaluation protocol, while additionally providing separated audio stems. The ablation results show that OB gives the most stable transcription gains, whereas TB changes the trade-off between reconstruction, acoustic stem quality, and onset detection. These results suggest that generative drum demixing can serve not only as a source separation model, but also as a practical front end for interpretable drum transcription.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 31 canonical work pages

  1. [1]

    Separate-and-Detect: Unified Drum Transcription and Stem Generation via Latent Diffusion

    INTRODUCTION Automatic Drum Transcription (ADT) aims to convert au- dio into symbolic drum events such as kick, snare, tom, hi-hat, and cymbal hits. The task is especially difficult in full-mixture music, where drum events overlap with vo- cals, harmonic instruments, and other sources. Recent deep learning systems have improved ADT performance, © W.-H. Hs...

    work page Pith review arXiv 2026

  2. [2]

    Prior diffusion work typically targets generic 4-stem or instrument-agnostic settings

    RELATED WORK 2.1 Diffusion Models for Music Source Separation Diffusion models have been explored for music source sep- aration (MSS) by learning joint multitrack distributions that support both separation and generation [5].Latent variants denoise in a compact V AE latent space and ren- der audio with a neural vocoder [2, 3]; user/text-guided demixing fu...

    work page

  3. [3]

    Given an input drum mixture, the sys- tem first isolates it into five drum-piece stems correspond- ing to kick, snare, toms, hi-hats, and cymbals

    METHODS We formulate drum transcription as a sequential separate- and-detect pipeline. Given an input drum mixture, the sys- tem first isolates it into five drum-piece stems correspond- ing to kick, snare, toms, hi-hats, and cymbals. Each sepa- rated audio stem is then processed by an independent, fixed onset detector to obtain symbolic events, which are ...

    work page

  4. [4]

    We evaluate on two unseen datasets: MDB Drums [22] and ENST-Drums [23]

    EXPERIMENTAL SETUP 4.1 Datasets and Preprocessing The MSG-LD models are trained on a combined corpus of StemGMD[11] (>1200 hours, synthesized) andIDMT- SMT-Drums[21] (2.1 hours, real-world acoustic). We evaluate on two unseen datasets: MDB Drums [22] and ENST-Drums [23]. Following the MIREX 2017 50/50 split for MDB Drums, the training split is utilized so...

    work page arXiv 2017

  5. [5]

    In terms of physical reconstruction, all MSG-LD variants significantly outperform the LarsNet baseline [11] in mel-spectrogram MSE across all drum classes

    RESULTS & DISCUSSION 5.1 Evaluation of Drum Source Separation The reconstruction and distributional audio-quality results are summarized in Table 1 and Table 2. In terms of physical reconstruction, all MSG-LD variants significantly outperform the LarsNet baseline [11] in mel-spectrogram MSE across all drum classes. While theV anillamodel achieves the lowe...

    work page

  6. [6]

    CONCLUSION & FUTURE WORK In this paper, we presented a generative separate-and- detect pipeline for automatic drum transcription. Evalu- ated on the MDB and ENST datasets, our multi-track latent diffusion separator consistently outperforms the strong U- Net-based baseline (LarsNet) in overall F1-score while re- maining competitive with strong end-to-end m...

    work page

  7. [7]

    ACKNOWLEDGMENTS The work is supported by grants from Google Asia Pacific, the National Science and Technology Council of Taiwan (NSTC 114-2628-E-002-013-MY3), and the Ministry of Education (MOE) of Taiwan (for Taiwan Centers of Ex- cellence in Artificial Intelligence)

    work page

  8. [8]

    AI USAGE STATEMENT During the preparation of this work, generative AI tech- nologies were utilized solely as writing and development assistance tools. Specifically, they were used for grammat- ical editing and text polishing of the manuscript, as well as for modifying minor utility scripts and debugging code during the experimental setup. No generative AI...

    work page

Show all 33 references
  1. [9]

    Spleeter: a fast and efficient music source sepa- ration tool with pre-trained models,

    R. Hennequin, A. Khlif, F. V oituret, and M. Moussal- lam, “Spleeter: a fast and efficient music source sepa- ration tool with pre-trained models,” inProc. Interna- tional Society for Music Information Retrieval Confer- ence (ISMIR), 2019

    2019

  2. [10]

    Adtof: A large dataset of non-synthetic music for automatic drum transcription,

    M. Zehren, M. Alunno, and P. Bientinesi, “Adtof: A large dataset of non-synthetic music for automatic drum transcription,” inProc. International Society for Music Information Retrieval Conference (ISMIR), 2021

    2021

  3. [11]

    Simul- taneous music separation and generation using multi- track latent diffusion models,

    T. Karchkhadze, M. R. Izadi, and S. Dubnov, “Simul- taneous music separation and generation using multi- track latent diffusion models,” inProc. IEEE Inter- national Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2025

    2025

  4. [12]

    Musicldm: Enhanc- ing novelty in text-to-music generation using beat- synchronous mixup strategies,

    K. Chen, Y . Wu, H. Liu, M. Nezhurina, T. Berg- Kirkpatrick, and S. Dubnov, “Musicldm: Enhanc- ing novelty in text-to-music generation using beat- synchronous mixup strategies,” inProc. IEEE Inter- national Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2024

    2024

  5. [13]

    Drumgan: Synthesis of drum sounds with timbral feature con- ditioning using generative adversarial networks,

    J. N. Hurle, S. Lattner, and G. Richard, “Drumgan: Synthesis of drum sounds with timbral feature con- ditioning using generative adversarial networks,” in Proc. International Society for Music Information Re- trieval Conference (ISMIR), 2020

    2020

  6. [14]

    Multi-source diffusion models for simultaneous music generation and sepa- ration,

    G. Mariani, I. Tallini, E. Postolache, M. Mancusi, L. Cosmo, and E. Rodolà, “Multi-source diffusion models for simultaneous music generation and sepa- ration,” inarXiv preprint arXiv:2302.02257, 2023

    2023 arXiv

  7. [15]

    User-guided generative source separation,

    Y . Wen, M. Kim, and P. Smaragdis, “User-guided generative source separation,” inarXiv preprint arXiv:2507.01339, 2025

    2025 arXiv

  8. [16]

    Hybrid trans- formers for music source separation,

    S. Rouard, F. Massa, and A. Défossez, “Hybrid trans- formers for music source separation,” inProc. IEEE In- ternational Conference on Acoustics, Speech, and Sig- nal Processing (ICASSP), 2023

    2023

  9. [17]

    Open-unmix-a reference implementation for music source separation,

    F.-R. Stöter, S. Uhlich, A. Liutkus, and Y . Mitsufuji, “Open-unmix-a reference implementation for music source separation,”Journal of Open Source Software, vol. 4, no. 41, p. 1667, 2019

    2019

  10. [18]

    separate-and-detect

    augmented ADTOF with drum stem separation to ex- pand the output taxonomy and estimate MIDI velocities, aiming for more realistic symbolic drum tracks. Similarly, theInverse Drum Machine[19] leverages an analysis-by- synthesis framework to jointly optimize drum transcription 1...

  11. [19]

    Music source separation with band- split rnn,

    Y . Luo and J. Yu, “Music source separation with band- split rnn,” inIEEE/ACM Trans. Audio, Speech, Lang. Process., 2023

    2023

  12. [20]

    Toward deep drum source separation,

    A. I. Mezza, R. Giampiccolo, A. Bernardini, and A. Sarti, “Toward deep drum source separation,”Pat- tern Recognition Letters, 2024

    2024

  13. [21]

    Benchmarking music demixing models for deep drum source separation,

    ——, “Benchmarking music demixing models for deep drum source separation,” inProc. IEEE Inter- national Symposium on the Internet of Sounds (IS2), 2024

    2024

  14. [22]

    Maximum filter vibrato sup- pression for onset detection,

    S. Böck and G. Widmer, “Maximum filter vibrato sup- pression for onset detection,” inConf. Digital Audio Effects (DAFx-14), Maynooth, Ireland, 2013

    2013

  15. [23]

    On- line real-time onset detection with recurrent neural networks,

    S. Böck, A. Arzt, F. Krebs, and M. Schedl, “On- line real-time onset detection with recurrent neural networks,” inConf. Digital Audio Effects (DAFx-14), York, UK, 2012

    2012

  16. [24]

    Improving auto- matic drum transcription using large-scale audio-to- midi aligned data,

    I.-C. Wei, C.-W. Wu, and L. Su, “Improving auto- matic drum transcription using large-scale audio-to- midi aligned data,” inProc. IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP), 2021, pp. 246–250

    2021

  17. [25]

    Drum transcription via joint beat and drum modeling using convolutional recurrent neural networks

    R. V ogl, M. Dorfer, G. Widmer, and P. Knees, “Drum transcription via joint beat and drum modeling using convolutional recurrent neural networks.” inProc. In- ternational Society for Music Information Retrieval Conference (ISMIR), 2017, pp. 150–157

    2017

  18. [26]

    Noise-to-notes: Diffusion-based genera- tion and refinement for automatic drum transcription,

    M. Yeung, K. Toyama, T. Teramoto, S. Takahashi, and T. Kojima, “Noise-to-notes: Diffusion-based genera- tion and refinement for automatic drum transcription,” arXiv preprint arXiv:2509.21739, 2025

    2025

  19. [27]

    Enhanced automatic drum transcription via drum stem source separation,

    X. Riley and S. Dixon, “Enhanced automatic drum transcription via drum stem source separation,”arXiv preprint arXiv:2509.24853, 2025

    2025

  20. [28]

    The inverse drum machine: Source separation through joint tran- scription and analysis-by-synthesis,

    B. Torres, G. Peeters, and G. Richard, “The inverse drum machine: Source separation through joint tran- scription and analysis-by-synthesis,”IEEE Transac- tions on Audio, Speech and Language Processing, 2025

    2025

  21. [29]

    Hifi-gan: Generative ad- versarial networks for efficient and high fidelity speech synthesis,

    J. Kong, J. Kim, and J. Bae, “Hifi-gan: Generative ad- versarial networks for efficient and high fidelity speech synthesis,” inAdvances in Neural Information Process- ing Systems, vol. 33, 2020, pp. 17 022–17 033

    2020

  22. [30]

    Real-time transcription and separation of drum recordings based on nmf de- composition

    C. Dittmar and D. Gärtner, “Real-time transcription and separation of drum recordings based on nmf de- composition.” inDAFx, 2014, pp. 187–194

    2014

  23. [31]

    Mdb drums: An annotated subset of medleydb for au- tomatic drum transcription,

    C. Southall, C.-W. Wu, A. Lerch, and J. Hockman, “Mdb drums: An annotated subset of medleydb for au- tomatic drum transcription,” inProc. International So- ciety for Music Information Retrieval Conference (IS- MIR), 2017

    2017

  24. [32]

    Enst-drums: an extensive audio-visual database for drum signals processing,

    O. Gillet and G. Richard, “Enst-drums: an extensive audio-visual database for drum signals processing,” in Proc. International Society for Music Information Re- trieval Conference (ISMIR), 2006

    2006

  25. [33]

    madmom: a new Python Audio and Mu- sic Signal Processing Library,

    S. Böck, F. Korzeniowski, J. Schlüter, F. Krebs, and G. Widmer, “madmom: a new Python Audio and Mu- sic Signal Processing Library,” inProceedings of the 24th ACM International Conference on Multimedia, 2016

    2016

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.