CoChat
Sign Up
Cited by 15Speech Recognition and SynthesisMusic and Audio ProcessingDigital Media Forensic Detection

ABC-CapsNet: Attention based Cascaded Capsule Network for Audio Deepfake Detection

Open access
Wani, Taiba Majid, Gulzar, Reeva, Amerini, Irene·2024·2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
DOI 10.1109/cvprw63382.2024.00253
Open this paper in CoChat — free accountOpen PDF

Abstract

In response to the escalating challenge of audio deepfake detection, this study introduces ABC-CapsNet (Attention-Based Cascaded Capsule Network), a novel architecture that merges the perceptual strengths of Mel spectrograms with the robust feature extraction capabilities of VGG18, enhanced by a strategically placed attention mechanism. This architecture pioneers the use of cascaded capsule networks to delve deeper into complex audio data patterns, setting a new standard in the precision of identifying manipulated audio content. Distinctively, ABC-CapsNet not only addresses the inherent limitations found in traditional CNN models but also showcases remarkable effectiveness across diverse datasets. The proposed method achieved an equal error rate EER of 0.06% on the ASVspoof2019 dataset and an EER of 0.04% on the FoR dataset, underscoring the superior accuracy and reliability of the proposed system in combating the sophisticated threat of audio deepfakes.

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 · 10.1109/cvprw63382.2024.00253

Cites

29 of 29 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?

  • Combining state-of-the-art backend models with traditional handcrafted features can outperform systems utilising learning-based features alone.

    “Studies by Wani et al. [105] and Ulutas et al. [142] have further demonstrated that combining SOTA backend models with traditional handcrafted features can outperform systems utilising learning-based features alone, underscoring the importance of a well-chosen backend in achieving optimal performance.”

    From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [105]

    The Contributions and Benchmarking sections describe a system that combines Mel spectrograms with VGG18 and capsule networks and reports better results than the cited prior systems. They do not compare that combination with a system using learning-based features alone, so the specific outperformance claim is not demonstrated by this paper. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

  • The MelSpec+ABC-CapsNet system achieved state-of-the-art performance on the ASVspoof 2019 LA dataset without employing any data augmentation.

    “Notably, the MelSpec+ABC-CapsNet system proposed by Wani et al. [105] achieved SOTA performance on the ASVspoof 2019 LA dataset without employing any DA.”

    From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [105]

    Section 4.3 reports results on ASVspoof 2019 LA, and Table 1 compares the method’s 0.06% EER with higher EERs for the listed state-of-the-art methods. The paper describes ABC-CapsNet as using Mel spectrograms, but our copy does not state that its training used no data augmentation; its mentions of augmentation in Related Works concern other systems. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Wani et al. used an Attention-Based Cascaded Capsule Network as a backend model, achieving exceptional results that surpassed other SOTA methods.

    “Wani et al. [105] utilised an Attention-Based Cascaded Capsule Network (ABC-CapsNet) as a backend model, achieving exceptional results that surpassed other SOTA methods.”

    From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [105]

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

Research works printed but not linked: 3

  1. [16]16. Sara Sabour, Nicholas Frosst, and Geoffrey E Hinton. Dy- namic routing between capsules. Advances in neural infor- mation processing systems, 30, 2017.
  2. [21]21. Kishan Vyas, Preksha Pareek, Ruchi Jayaswal, and Shruti Patil. Analysing the landscape of deep fake detection: A survey. International Journal of Intelligent Systems and Ap- plications in Engineering, 12(11s):40-55, 2024.
  3. [22]22. Taiba Majid Wani and Irene Amerini. Deepfakes audio de- tection leveraging audio spectrogram and convolutional neu- ral networks. In International Conference on Image Analysis and Processing, pages 156-167. Springer, 2023.

Cited by

Sorted by citations

Related

Authors
Wani, T. M., Gulzar, R., & Amerini, I.
Affiliations
Sapienza University of Rome (IT)
ORCID
Venue
2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
Publisher
IEEE
Published
2024-06-17
Pages
2464-2472
DOI
10.1109/cvprw63382.2024.00253
OpenAlex
W4402916306
Cited by
14
References
24
Citations by year
2026: 4 · 2025: 7 · 2024: 3
FWCI
4.04
Type
conference paper
Language
en
Open access
Licence
Locations
  • IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) (publishedVersion)
  • IRIS Research product catalog (Sapienza University of Rome) (submittedVersion, other-oa)
Funders
  • European Commission (PE00000014)
  • Sapienza Università di Roma
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
  • Physical Sciences › Computer Science › Signal Processing › Music and Audio Processing
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
Keywords
audio deepfake detection, attention mechanism, Mel spectrogram, ASVspoof 2019

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