Pith. sign in

REVIEW 4 major objections 4 minor 25 references

Planing It by Ear: Convolutional Neural Networks for Acoustic Anomaly Detection in Industrial Wood Planers

T0 review · 4 major / 4 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 Two neural nets spot wood planer faults by ear, top AUC 0.875

desk verdict A useful new dataset and a solid but statistically unquantified application of known autoencoder variants to wood planer acoustic anomaly detection. read the letter →

arxiv 2501.04819 v1 pith:INVHYHNG submitted 2025-01-08 cs.SD cs.AIeess.AS

Anthony Deschênes , Rémi Georges , Cem Subakan , Bruna Ugulino , Antoine Henry , Michael Morin This is my paper · ORCID
classification cs.SDcs.AIeess.AS
keywords acousticanomalydetectionconvolutionalautoencoderskipconnectionstransformerwoodplanerindustrialmonitoringmelspectrogramunsupervisedlearning
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 aims to show that deep convolutional autoencoders can detect abnormal sounds in industrial wood planers using only the sound of normal operation for training. On a newly collected real-factory dataset of 7,562 ten-second recordings, the proposed Skip-CAE and Skip-CAE-Transformer reach areas under the ROC curve of 0.846 and 0.875, outperforming a published convolutional autoencoder, the DCASE autoencoder baseline, one-class SVM, and isolation forest. The motivation is practical: skilled planer operators are scarce, and acoustic monitoring could give novice operators an automated ear for stuck boards, broken boards, and uneven or thick wood passing through the machine.

What carries the argument

The core mechanism is the reconstruction-error anomaly score: an autoencoder is trained to reconstruct normal mel spectrograms, and a high reconstruction error marks an anomaly. The Skip-CAE is a convolutional autoencoder with skip connections between encoder and decoder layers, using leaky ReLU and batch normalization before pooling and upsampling to stabilize training and preserve fine spectral detail. The Skip-CAE-Transformer inserts a transformer encoder-decoder at the bottleneck (10 heads, single layer) and routes two skip connections through fully connected layers, which lets the network decide how much detail to pass; this transformer attention is what the paper credits for the further improvement in anomaly detection.

What would settle it

Re-annotating the evaluation set with exhaustive labels and recomputing the AUC would settle the claim: if many currently unlabeled recordings are anomalies, the reported 0.875 AUC and the model ranking could change.

Watch

Extended reading notes

Core claim

The central claim is that adding skip connections to a convolutional autoencoder, and further adding transformer encoder/decoder blocks, yields a practical acoustic anomaly detector for real sawmill noise. Trained only on normal planer sounds (recordings made shortly after knife jointing or head change), the models flag anomalies by reconstruction error on log-scaled mel spectrograms. On the evaluation set spanning two days of operation with 105 expert-flagged anomalies, Skip-CAE-Transformer achieves the best overall AUC of 0.875 (pAUC 0.785), and Skip-CAE achieves 0.846 (pAUC 0.787); both are the only models reaching a 20% true positive rate with zero false positives. Per anomaly type, Skip-CAE-Transformer leads in AUC or pAUC for all three categories, with its largest margin on uneven or thick wood (AUC 0.921).

Load-bearing premise

The expert anomaly labels in the evaluation set are assumed complete enough to compute a trustworthy AUC, but the paper states some anomalies may be unlabeled, so unlabeled anomalous recordings are scored as normal.

Editorial extensions

If this is right

  • If the results hold, acoustic monitoring with these models could be deployed on real planers to alert operators to stuck boards, broken boards, and uneven or thick wood in near real time.
  • The public dataset of 7,562 real-factory recordings gives other researchers a benchmark for industrial acoustic anomaly detection under realistic noise.
  • The finding that skip connections and a transformer bottleneck improve over the DCASE baseline supports further work on attention-based autoencoders for machine-condition monitoring.
  • The models' ability to reach 20% true positive rate with no false positives suggests they can support high-precision alarm systems in settings where false alarms are costly.
  • Because training uses only normal sounds, the approach fits unsupervised deployment scenarios where anomaly examples are rare or unknown.

Reading between the lines

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

  • Editorial inference: the method's dependence on the expert label set is the main risk; if the evaluation set contains unlabeled anomalies, the reported AUC is an upper bound, and the ranking between models could shift under complete labeling.
  • Editorial inference: the same Skip-CAE-Transformer recipe could plausibly transfer to other rotating wood-processing machinery, such as saws or CNC routers, but this would require re-tuning and is not demonstrated here.
  • Editorial inference: a practical extension would be to test whether the models can localize anomalies in time within a 10-second clip, which the current clip-level AUC does not address.
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

4 major / 4 minor

Summary. This paper addresses acoustic anomaly detection for industrial wood planers. The authors introduce a new dataset of 7,562 ten-second recordings from a real planing mill, with 105 expert-labeled anomaly events, and propose two neural architectures: a convolutional autoencoder with skip connections (Skip-CAE) and a variant with a transformer encoder-decoder (Skip-CAE-Transformer). The models are compared against a DCASE baseline autoencoder, the convolutional autoencoder of Duman et al., one-class SVM, and isolation forest, using AUC and pAUC on a held-out evaluation set. The reported best AUC is 0.875 for the Skip-CAE-Transformer.

Significance. If the reported results are robust, the paper contributes a useful open industrial dataset and demonstrates that skip connections and attention can improve reconstruction-based acoustic anomaly detection. The public dataset and code are a clear strength. However, the central comparative claim is currently supported only by point estimates on a single evaluation set with acknowledged incomplete labels, and the paper lacks uncertainty quantification, so the practical significance of the contribution depends on addressing these limitations.

major comments (4)
  1. [Section III and Section V-A] The evaluation-set AUCs are computed under the assumption that every unlabeled ten-second recording is normal, yet Section III states that "some anomalies might not be labeled in the evaluation set." Because unlabeled anomalous recordings would be counted as false positives when a detector assigns them high anomaly scores, the reported AUC values (e.g., 0.875 for Skip-CAE-Transformer) are not verified against complete ground truth, and the ranking of models could change if the better detector is also better at catching the unlabeled events. The authors should re-verify the evaluation-set labels with the expert, quantify the expected number of missing labels, or provide a sensitivity analysis showing that the ranking is stable under plausible label noise.
  2. [Section V-A and Table II] The paper reports single point estimates without confidence intervals, repeated training runs, or significance tests. The margin between Skip-CAE (AUC 0.846) and Skip-CAE-Transformer (0.875) is only 0.029, and the per-type results in Table II are based on very few positive events (four broken boards); without error bars or bootstrap intervals, the statement that "the Skip-CAE performs significantly better" (Section V-A) is not statistically justified. Add confidence intervals (e.g., bootstrap over evaluation recordings) and at least a few training seeds, and restrict per-type conclusions to anomaly types with sufficient positive samples.
  3. [Section V-B and Table II] The claim that "Skip-CAE-Transformer outperforms all other models in AUC or pAUC for all of the anomaly types" is contradicted by Table II: for Broken Board, Skip-CAE has AUC 0.777 versus 0.743, and for Uneven or Thick Wood, Skip-CAE has pAUC 0.820 versus 0.807. Please correct the claim or qualify it to "best in at least one of the two metrics for most types" and discuss where it is not the best.
  4. [Section V-A and Figure 3] The statement that Skip-CAE-Transformer "performs the best" is only true for AUC (0.875 versus 0.846); on pAUC, Skip-CAE reports 0.787 versus 0.785, so the overall superiority depends on the metric. The text should state which metric is being prioritized and discuss the pAUC comparison explicitly.
minor comments (4)
  1. [Section IV] There is a typo in "convolutional autocencoders" in the first sentence of Section IV; it should read "autoencoders."
  2. [Section IV] The DCASE baseline is described as an autoencoder whose input and output "lead to 32,080 neurons"; please clarify that this is the flattened mel-spectrogram dimension and state whether any dropout or other regularization is applied.
  3. [Figure 3] The ROC curves in Figure 3 are difficult to distinguish in grayscale; consider using different line styles or markers in addition to color.
  4. [Section I] The contribution bullet describing the dataset should state the license and any conditions of use, since the dataset is a central contribution.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central comparison is an empirical benchmark measured on a held-out evaluation set.

full rationale

The paper's central claim is empirical: two proposed autoencoder architectures are compared against four baselines on a new industrial audio dataset, with AUC and pAUC computed on an evaluation set that was not used for training. There is no derivation chain whose output coincides with its input, no fitted parameter that is subsequently relabeled as a prediction, and no self-citation that carries the argument. The training set contains only normal recordings, and the evaluation set contains expert-labeled anomalies; the models are trained with fixed hyperparameters (batch size 32, learning rate 1e-3, 500 epochs with early stopping) and the same protocol is applied to all models. The literature baselines (DCASE autoencoder, Duman et al. CAE, one-class SVM, isolation forest) are implemented as external comparators, not as sources of the reported result. The authors' own prior work appears only as background context on wood planer control ([6], [12]) and is not load-bearing for the anomaly detection comparison. The acknowledged limitation that some anomalies may be unlabeled in the evaluation set, and the very small number of broken-board events (4), are threats to the validity or stability of the reported AUC values; they are not circularity, because the reported numbers are measurements of held-out data rather than quantities constructed from the labels by definition. No circular step was found, so the score is 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim is an empirical performance number, not a derivation; the free parameters are training and architecture hyperparameters selected by hand. The main unverified inputs are the assumptions about label quality and feature sufficiency.

free parameters (2)
  • Neural network hyperparameters = batch size 32, learning rate 1e-3, 500 epochs, early stopping patience 30, cosine scheduler
    Chosen by hand for all models; not tuned on the evaluation set, so they do not create circularity.
  • Architecture dimensions = e.g., 64 channels at deepest layer, transformer d_model=500, 10 heads, 50-feature bottleneck
    Heuristic choices based on input size; no systematic ablation or tuning reported.
assumptions (3)
  • domain assumption Expert anomaly labels in the evaluation set are sufficiently correct and complete for AUC estimation.
    Section III states that some anomalies might not be labeled in the evaluation set; if unlabeled anomalies are frequent, the reported AUC is unreliable.
  • domain assumption Training recordings are free of anomalies, so the autoencoder learns a representation of normal operation.
    Section III claims the training set contains no anomalies, recorded shortly after jointing or head change; hidden anomalies would be modeled as normal.
  • domain assumption Log-scaled mel spectrograms are an adequate feature representation for this anomaly detection task.
    The method relies on mel spectrograms without comparing alternative features (e.g., raw waveforms or cepstral coefficients).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Planing It by Ear: Convolutional Neural Networks for Acoustic Anomaly Detection in Industrial Wood Planers." pith.science (2026). https://pith.science/paper/INVHYHNG

@misc{pith2026250104819,
  author       = {Pith},
  title        = {Pith review of: Planing It by Ear: Convolutional Neural Networks for Acoustic Anomaly Detection in Industrial Wood Planers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/INVHYHNG}},
  note         = {Machine review of arXiv:2501.04819}
}
read the original abstract

In recent years, the wood product industry has been facing a skilled labor shortage. The result is more frequent sudden failures, resulting in additional costs for these companies already operating in a very competitive market. Moreover, sawmills are challenging environments for machinery and sensors. Given that experienced machine operators may be able to diagnose defects or malfunctions, one possible way of assisting novice operators is through acoustic monitoring. As a step towards the automation of wood-processing equipment and decision support systems for machine operators, in this paper, we explore using a deep convolutional autoencoder for acoustic anomaly detection of wood planers on a new real-life dataset. Specifically, our convolutional autoencoder with skip connections (Skip-CAE) and our Skip-CAE transformer outperform the DCASE autoencoder baseline, one-class SVM, isolation forest and a published convolutional autoencoder architecture, respectively obtaining an area under the ROC curve of 0.846 and 0.875 on a dataset of real-factory planer sounds. Moreover, we show that adding skip connections and attention mechanism under the form of a transformer encoder-decoder helps to further improve the anomaly detection capabilities.

Figures

Figures reproduced from arXiv: 2501.04819 by the authors.

Figure 1
Figure 1. Architecture of the CAE with skip connections [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 4
Figure 4. Mel spectogram of an anomaly (broken board) and the [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 3
Figure 3. ROC curves, AUC and pAUC of the different models [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 21 canonical work pages

  1. [1]

    Impact of human error on lumber yield in rough mills,

    U. Buehlmann and R. E. Thomas, “Impact of human error on lumber yield in rough mills,” Robotics and Computer-Integrated Manufacturing, vol. 18, no. 3-4, pp. 197–203, 2002

    work page 2002

  2. [2]

    Acoustic anomaly detection using convolutional autoencoders in industrial processes,

    T. B. Duman, B. Bayram, and G. ˙Ince, “Acoustic anomaly detection using convolutional autoencoders in industrial processes,” in 14th In- ternational Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2019) Seville, Spain, May 13–15, 2019, Proceedings 14 . Springer, 2020, pp. 432–442

    work page 2019

  3. [3]

    Anomalous Sound Detection with Machine Learning: A Systematic Review

    E. C. Nunes, “Anomalous sound detection with machine learning: A systematic review,” arXiv preprint arXiv:2102.07820 , Feb. 2021

    work page Pith review arXiv 2021

  4. [4]

    Development of a monitoring system for guided circular saws: an experimental investiga- tion,

    A. Mohammadpanah, B. Lehmann, and J. White, “Development of a monitoring system for guided circular saws: an experimental investiga- tion,” Wood Material Science & Engineering, vol. 14, no. 2, pp. 99–106, 2019

    work page 2019

  5. [5]

    Multisensor data fusion and machine learning to classify wood products and predict workpiece characteristics during milling,

    M. Derbas, A. Jaquemod, S. Fr ¨omel-Frybort, K. G¨uzel, H.-C. Moehring, and M. Riegler, “Multisensor data fusion and machine learning to classify wood products and predict workpiece characteristics during milling,” CIRP Journal of Manufacturing Science and Technology , vol. 47, pp. 103–115, 2023

    work page 2023

  6. [6]

    Wood planer control: Predictive and prescriptive approaches via au- tomatic state matching gaussian processes,

    J.-T. Sexton, M. Morin, R. Georges, F. Abasian, and J. Gaudreault, “Wood planer control: Predictive and prescriptive approaches via au- tomatic state matching gaussian processes,” Engineering Applications of Artificial Intelligence , vol. 132, p. 107843, 2024

    work page 2024

  7. [7]

    Supervised classification of wood species during milling based on extracted cut events from ultrasonic air-borne acoustic signals,

    M. Derbas, S. Fr ¨omel-Frybort, C. Laaber, H.-C. M ¨ohring, and M. Riegler, “Supervised classification of wood species during milling based on extracted cut events from ultrasonic air-borne acoustic signals,” Wood Material Science & Engineering , vol. 18, no. 6, pp. 2040–2048, 2023

    work page 2023

  8. [8]

    Acoustic emission monitoring of sawing process: artificial intelligence approach for optimal sensory feature selection,

    V . Nasir, J. Cool, and F. Sassani, “Acoustic emission monitoring of sawing process: artificial intelligence approach for optimal sensory feature selection,” The International Journal of Advanced Manufacturing Technology, vol. 102, pp. 4179–4197, 2019

    work page 2019

Show all 25 references
  1. [9]

    Overview on devel- opment of acoustic emission monitoring technology in sawing,

    R. Zhuo, Z. Deng, B. Chen, G. Liu, and S. Bi, “Overview on devel- opment of acoustic emission monitoring technology in sawing,” The International Journal of Advanced Manufacturing Technology, vol. 116, pp. 1411–1427, 2021

    2021

  2. [10]

    Toward a process monitoring of CNC wood router. Sensor selection and surface roughness prediction,

    P. Iskra and R. E. Hern ´andez, “Toward a process monitoring of CNC wood router. Sensor selection and surface roughness prediction,” Wood Science and Technology, vol. 46, no. 1, pp. 115–128, Jan. 2012

    2012

  3. [11]

    A comparison of selected acoustic signal analysis techniques to evaluate wood surface roughness produced during routing,

    P. Iskra and C. Tanaka, “A comparison of selected acoustic signal analysis techniques to evaluate wood surface roughness produced during routing,” Wood Science and Technology , vol. 40, no. 3, pp. 247–259, Mar. 2006

    2006

  4. [12]

    Automatic state matching gaussian process ensemble for wood planer control,

    J.-T. Sexton, M. Morin, R. Georges, F. Abasian, and J. Gaudreault, “Automatic state matching gaussian process ensemble for wood planer control,” IFAC-PapersOnLine, vol. 55, no. 10, pp. 625–630, 2022

    2022

  5. [13]

    A distributed sensor management for large-scale IoT indoor acoustic surveillance,

    A. R. Hilal, A. Sayedelahl, A. Tabibiazar, M. S. Kamel, and O. A. Basir, “A distributed sensor management for large-scale IoT indoor acoustic surveillance,” Future Generation Computer Systems , vol. 86, pp. 1170– 1184, 2018

    2018

  6. [14]

    Anomalous sound event detec- tion: A survey of machine learning based methods and applications,

    Z. Mnasri, S. Rovetta, and F. Masulli, “Anomalous sound event detec- tion: A survey of machine learning based methods and applications,” Multimedia Tools and Applications, vol. 81, no. 4, pp. 5537–5586, Feb. 2022

    2022

  7. [15]

    Data-driven smart manufacturing: Tool wear monitoring with audio signals and machine learning,

    Z. Li, R. Liu, and D. Wu, “Data-driven smart manufacturing: Tool wear monitoring with audio signals and machine learning,” Journal of Manufacturing Processes, vol. 48, pp. 66–76, Dec. 2019

    2019

  8. [16]

    Smart audio sensors in the Internet of things edge for anomaly detection,

    M. Antonini, M. Vecchio, F. Antonelli, P. Ducange, and C. Perera, “Smart audio sensors in the Internet of things edge for anomaly detection,” IEEE Access, vol. 6, pp. 67 594–67 610, 2018

    2018

  9. [17]

    Anomaly de- tection in milling tools using acoustic signals and generative adversarial networks,

    C. Cooper, J. Zhang, R. X. Gao, P. Wang, and I. Ragai, “Anomaly de- tection in milling tools using acoustic signals and generative adversarial networks,” Procedia Manufacturing, vol. 48, pp. 372–378, 2020

    2020

  10. [18]

    Acoustic anomaly detection of mechanical failures in noisy real-life factory environments,

    Y . Tagawa, R. Maskeli ¯unas, and R. Dama ˇseviˇcius, “Acoustic anomaly detection of mechanical failures in noisy real-life factory environments,” Electronics, vol. 10, no. 19, p. 2329, 2021

    2021

  11. [19]

    Description and discussion on DCASE 2024 challenge task 2: First-shot unsupervised anomalous sound detection for machine condition monitoring,

    T. Nishida, N. Harada, D. Niizumi, D. Albertini, R. Sannino, S. Pradolini, F. Augusti, K. Imoto, K. Dohi, H. Purohit, T. Endo, and Y . Kawaguchi, “Description and discussion on DCASE 2024 challenge task 2: First-shot unsupervised anomalous sound detection for machine condition...

    2024 arXiv

  12. [20]

    Residual error based anomaly detection using auto-encoder in SMD machine sound,

    D. Y . Oh and I. D. Yun, “Residual error based anomaly detection using auto-encoder in SMD machine sound,” Sensors, vol. 18, no. 5, p. 1308, 2018

    2018

  13. [21]

    First-shot anomaly detection for machine condition monitoring: A domain generalization baseline,

    N. Harada, D. Niizumi, D. Takeuchi, Y . Ohishi, and M. Yasuda, “First-shot anomaly detection for machine condition monitoring: A domain generalization baseline,” Proceedings of 31st European Signal Processing Conference (EUSIPCO), pp. 191–195, 2023

    2023

  14. [22]

    Attention is All you Need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is All you Need,” Advances in Neural Information Processing Systems , 2017

    2017

  15. [23]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” in International Conference on Learning Representations , 2019

    2019

  16. [24]

    SGDR: Stochastic gradient descent with warm restarts,

    ——, “SGDR: Stochastic gradient descent with warm restarts,” in International Conference on Learning Representations , 2017. [Online]. Available: https://openreview.net/forum?id=Skq89Scxx

    2017

  17. [25]

    Scikit-learn: Machine learning in Python,

    F. Pedregosa, G. Varoquaux, A. Gramfort, V . Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V . Dubourg, J. Vander- plas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duch- esnay, “Scikit-learn: Machine learning in Python,” Journal of Machine ...

    2011

Pith tools

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