ToyADMOS: A Dataset of Miniature-Machine Operating Sounds for Anomalous Sound Detection

@article{Koizumi2019ToyADMOSAD,
  title={ToyADMOS: A Dataset of Miniature-Machine Operating Sounds for Anomalous Sound Detection},
  author={Yuma Koizumi and Shoichiro Saito and Hisashi Uematsu and Noboru Harada and Keisuke Imoto},
  journal={2019 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA)},
  year={2019},
  pages={313-317}
}
This paper introduces a new dataset called "ToyADMOS" designed for anomaly detection in machine operating sounds (ADMOS). To the best our knowledge, no large-scale datasets are available for ADMOS, although large-scale datasets have contributed to recent advancements in acoustic signal processing. This is because anomalous sound data are difficult to collect. To build a large-scale dataset for ADMOS, we collected anomalous operating sounds of miniature machines (toys) by deliberately damaging… 

Figures and Tables from this paper

ToyADMOS2: Another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions
TLDR
A large number of operating sounds of miniature machines (toys) under normal and anomaly conditions by deliberately damaging them but extended with providing controlled depth of damages in anomaly samples to propose a new large-scale dataset for anomaly detection in machine operating sounds (ADMOS).
ToyADMOS2 dataset: Another dataset of miniature-machine operating sounds for anomalous sound detection under domain shift conditions
TLDR
The ToyADMOS2 dataset is designed for evaluating systems under typical application scenarios of ADMOS often require robust performance under domain-shift conditions, and consists of two sub-datasets for machine-condition inspection: fault diagnosis of machines with geometrically fixed tasks and fault diagnosisof machines with moving tasks.
MIMII Dataset: Sound Dataset for Malfunctioning Industrial Machine Investigation and Inspection
TLDR
A new dataset of industrial machine sounds that is called a sound dataset for malfunctioning industrial machine investigation and inspection (MIMII dataset) is presented to assist the machine-learning and signal-processing community with their development of automated facility maintenance.
IAEO3-COMBINING OPENL3 EMBEDDINGS AND INTERPOLATION AUTOENCODER FOR ANOMALOUS SOUND DETECTION Technical Report
TLDR
This proposed method combines pre-trained OpenL3 embeddings with the reconstruction error of an interpolation autoencoder using a gaussian mixture model as the final predictor and outperformed the published baseline by 14.9% AUC and 17.2% pAUC.
DCASE CHALLENGE 2020: UNSUPERVISED ANOMALOUS SOUND DETECTION OF MACHINERY WITH DEEP AUTOENCODERS Technical Report
TLDR
This work presents an unsupervised anomalous sound detection framework trained on DCASE2020 audio dataset, which uses the state of the art anomaly detection approach, deep autoencoder architecture trained on Mel-spectrograms to learn the normal condition of the machine.
Sub-Cluster AdaCos: Learning Representations for Anomalous Sound Detection
  • Kevin Wilkinghoff
  • Computer Science
    2021 International Joint Conference on Neural Networks (IJCNN)
  • 2021
TLDR
A modified AdaCos loss called sub-cluster AdaCos specifically designed for detecting anomalous data is presented and significantly outperforms all other published systems on this dataset.
Anomalous Sound Detection with Machine Learning: A Systematic Review
TLDR
A Systematic Review about studies related to Anamolous Sound Detection using Machine Learning (ML) techniques showed that the ToyADMOS, MIMII, and Mivia datasets, the Mel-frequency cepstral coefficients (MFCC) method for extracting features, the Autoencoder (AE) and Convolutional Neural Network (CNN) models of ML, the AUC and F1-score evaluation methods were most cited.
Unsupervised Detection of Anomalous Sound for Machine Condition Monitoring using Fully Connected U-Net
TLDR
The Mixed Feature is created, which is a mixture of different sound representations, and a new deep learning method called Fully-Connected U-Net, a form of autoencoder architecture, which outperformed the baseline system in terms of the AUC and pAUC evaluation metrics.
ANOMALOUS SOUND DETECTION WITH LOOK, LISTEN, AND LEARN EMBEDDINGS Technical Report
TLDR
The experimental results show that the presented system significantly outperforms the baseline system of the challenge both in detecting outliers and in recognizing the correct machine type or exact machine id.
Anomalous Sound Detection Using Attentive Neural Processes
TLDR
This work presents an iterative approach that finds difficult-to-reconstruct spectrogram regions, and uses the reconstruction error over only those regions as the anomaly score, and outperforms baseline approaches in predicting anomalies for unseen machine instances.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 26 REFERENCES
How can we detect anomalies from subsampled audio signals?
  • Y. Kawaguchi, Takashi Endo
  • Computer Science
    2017 IEEE 27th International Workshop on Machine Learning for Signal Processing (MLSP)
  • 2017
TLDR
Experimental results indicate that the proposed method for reducing the sampling rate and applying a long short-term memory-(LSTM)-based autoencoder network for detecting anomalies is suitable for anomaly detection from the subsampled signal.
Unsupervised Detection of Anomalous Sound Based on Deep Learning and the Neyman–Pearson Lemma
TLDR
This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound (ADS) using an autoencoder (AE) and defines an objective function based on the Neyman–Pearson lemma by considering the ADS as a statistical hypothesis test.
Optimizing acoustic feature extractor for anomalous sound detection based on Neyman-Pearson lemma
TLDR
The proposed method improved the F-measure score from 0.02 to 0.06 points compared to those of conventional methods, and ASD results of a stereolithography 3D-printer in a real-environment show that the proposed method is effective in identifying anomalous sounds.
Batch Uniformization for Minimizing Maximum Anomaly Score of Dnn-Based Anomaly Detection in Sounds
TLDR
Batch uniformization is proposed, a training method for unsupervised-ADS for minimizing a weighted average of the anomaly score on each sample in a mini-batch, using the reciprocal of the probabilistic density of each sample as the weight.
Anomaly Detection Based on an Ensemble of Dereverberation and Anomalous Sound Extraction
TLDR
Experimental results indicate that the proposed method for detecting anomalous sounds based on a front-end ensemble consisting of a blind-dereverberation algorithm and multiple anomalous-sound-extraction algorithms improves detection performance significantly.
Probabilistic Novelty Detection for Acoustic Surveillance Under Real-World Conditions
TLDR
Probabilistic novelty detection can provide an accurate analysis of the audio scene to identify abnormal events and is explored as applied to acoustic surveillance of abnormal situations.
Scream and gunshot detection and localization for audio-surveillance systems
This paper describes an audio-based video surveillance system which automatically detects anomalous audio events in a public square, such as screams or gunshots, and localizes the position of the
SNIPER: Few-shot Learning for Anomaly Detection to Minimize False-negative Rate with Ensured True-positive Rate
TLDR
This work proposes a training method for a cascaded specific anomaly detector using few-shot (just 1 to 3) samples and shows that the proposed method outperformed conventional cross-entropy-based few- shot learning methods.
Complementary Set Variational Autoencoder for Supervised Anomaly Detection
TLDR
The proposed model defines the normal and anomaly distributions using the analogy between a set and the complementary set and was applied to an unsupervised variational autoencoder (VAE)-based method and turned it into a supervised VAE-based method.
A novel approach for automatic acoustic novelty detection using a denoising autoencoder with bidirectional LSTM neural networks
TLDR
This paper presents a novel unsupervised approach based on a denoising autoencoder which significantly outperforms existing methods by achieving up to 93.4% F-Measure.
...
1
2
3
...