CoChat
Sign Up
Cited by 132Music and Audio ProcessingSpeech Recognition and SynthesisAdvanced Neural Network Applications

Comparison of Pre-Trained CNNs for Audio Classification Using Transfer Learning

Open access
Tsalera, Eleni, Papadakis, Andreas, Samarakou, Maria·2021·Journal of Sensor and Actuator Networks
DOI 10.3390/jsan10040072
Open this paper in CoChat — free accountOpen PDF

Abstract

The paper investigates retraining options and the performance of pre-trained Convolutional Neural Networks (CNNs) for sound classification. CNNs were initially designed for image classification and recognition, and, at a second phase, they extended towards sound classification. Transfer learning is a promising paradigm, retraining already trained networks upon different datasets. We selected three ‘Image’- and two ‘Sound’-trained CNNs, namely, GoogLeNet, SqueezeNet, ShuffleNet, VGGish, and YAMNet, and applied transfer learning. We explored the influence of key retraining parameters, including the optimizer, the mini-batch size, the learning rate, and the number of epochs, on the classification accuracy and the processing time needed in terms of sound preprocessing for the preparation of the scalograms and spectrograms as well as CNN training. The UrbanSound8K, ESC-10, and Air Compressor open sound datasets were employed. Using a two-fold criterion based on classification accuracy and time needed, we selected the ‘champion’ transfer-learning parameter combinations, discussed the consistency of the classification results, and explored possible benefits from fusing the classification estimations. The Sound CNNs achieved better classification accuracy, reaching an average of 96.4% for UrbanSound8K, 91.25% for ESC-10, and 100% for the Air Compressor dataset.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3390/jsan10040072

Cites

40 of 40 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • When trained on large general audio datasets, these models often exceed 90% accuracy and transfer remarkably well to wildlife monitoring.

    “When trained on large general audio datasets, these models often exceed 90% accuracy and transfer remarkably well to wildlife monitoring”

    From Listening forward: emerging roles of bioacoustics in ecology, evolution, and conservation (2026) · cites this paper as Tsalera et al. 2021

  • Pre-trained models for audio recognition (YAMNet, VGGish) are trained on sounds of limited time duration and resolution (rather than on images of sounds), which can impact their applicability to noisy and complex signals such as found in biodiverse soundscapes.

    “While pre-trained models for audio recognition have recently been proposed (YAMNet, VGGish62), they are trained on sounds of limited time duration and resolution (rather than on images of sounds), which can impact their applicability to noisy and complex signals such as found in biodiverse soundscapes.”

    From Soundscapes and deep learning enable tracking biodiversity recovery in tropical forests (2023) · cites this paper as [62]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

1 of the 41 references couldn’t be matched to a known paper.

Research works printed but not linked: 10

  1. Abu-El-Haija, S.; Kothari, N.; Lee, J.; Natsev, P.; Toderici, G.; Varadarajan, B.; Vijayanarasimhan, S. Youtube-8m: A large-scale video classification benchmark. arXiv 2016, arXiv:1609.08675. · arXiv
  2. Simonyan, K.; Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv 2014, arXiv:1409.1556. · arXiv
  3. Iandola, F.N.; Han, S.; Moskewicz, M.W.; Ashraf, K.; Dally, W.J.; Keutzer, K. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5 MB model size. arXiv 2016, arXiv:1602.07360. · arXiv
  4. Howard, A.G.; Zhu, M.; Chen, B.; Kalenichenko, D.; Wang, W.; Weyand, T.; Adam, H. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv 2017, arXiv:1704.04861. · arXiv
  5. Howard, A.; Sandler, M.; Chu, G.; Chen, L.C.; Chen, B.; Tan, M.; Adam, H. Searching for mobilenetv3. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Korea, 27 October-2 November 2019; pp. 1314-1324.
  6. Ramezani-Kebrya, A.; Khisti, A.; Liang, B. On the Generalization of Stochastic Gradient Descent with Momentum. arXiv 2021, arXiv:2102.13653. · arXiv
  7. Keskar, N.S.; Socher, R. Improving generalization performance by switching from adam to sgd. arXiv 2017, arXiv:1712.07628. · arXiv
  8. Zhou, P.; Feng, J.; Ma, C.; Xiong, C.; Hoi, S. Towards theoretically understanding why sgd generalizes better than adam in deep learning. arXiv 2020, arXiv:2010.05627. · arXiv
  9. Smith, S.L.; Kindermans, P.J.; Ying, C.; Le, Q.V. Don't decay the learning rate, increase the batch size. arXiv 2017, arXiv:1711.00489. · arXiv
  10. Hoffer, E.; Hubara, I.; Soudry, D. Train longer, generalize better: Closing the generalization gap in large batch training of neural networks. arXiv 2017, arXiv:1705.08741. · arXiv

Cited by

Sorted by citations

Related

Authors
Tsalera, E., Papadakis, A., & Samarakou, M.
Affiliations
  • University of West Attica (GR)
  • School of Pedagogical and Technological Education (GR)
  • Department of Informatics and Computer Engineering, School of Engineering, University of West Attica, 11521 Athens, Greece
Corresponding author
Papadakis, Andreas
ORCID
Venue
Journal of Sensor and Actuator Networks
Publisher
MDPI AG
Published
2021-12-10
Volume
10
Issue
4
Pages
72
ISSN
2224-2708 (ISSN-L)
DOI
10.3390/jsan10040072
OpenAlex
W4200473284
Cited by
127
References
41
Citations by year
2026: 16 · 2025: 34 · 2024: 42 · 2023: 26 · 2022: 9
FWCI
8.96
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Sensor and Actuator Networks (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Signal Processing › Music and Audio Processing
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
Keywords
audio classification, transfer learning, pre-trained convolutional neural networks, GoogLeNet, SqueezeNet, ShuffleNet, VGGish, YAMNet, spectrogram, scalogram, UrbanSound8K, ESC-10, classifier combination, training duration

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review