Leveraging Mixture of Experts for Improved Speech Deepfake Detection
Open accessThe DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index.
Abstract
Speech deepfakes pose a significant threat to personal security and content authenticity. Several detectors have been proposed in the literature, and one of the primary challenges these systems have to face is the generalization over unseen data to identify fake signals across a wide range of datasets. In this paper, we introduce a novel approach for enhancing speech deepfake detection performance using a Mixture of Experts architecture. The Mixture of Experts framework is well-suited for the speech deepfake detection task due to its ability to specialize in different input types and handle data variability efficiently. This approach offers superior generalization and adaptability to unseen data compared to traditional single models or ensemble methods. Additionally, its modular structure supports scalable updates, making it more flexible in managing the evolving complexity of deepfake techniques while maintaining high detection accuracy. We propose an efficient, lightweight gating mechanism to dynamically assign expert weights for each input, optimizing detection performance. Experimental results across multiple datasets demonstrate the effectiveness and potential of our proposed approach.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
“Librispeech: an ASR corpus based on public domain audio books,”
Panayotov, V., Chen, G., Povey, D., & Khudanpur, S. (2015) · 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2015.7178964 · Cited by 6,072 · cited 1×
“The real 1 Code will be released upon acceptance. speech data were sourced from LJSpeech [26] and Lib-riSpeech [27].”
MethodAdaptive Mixtures of Local Experts
Jacobs, R. A., Jordan, M. I., Nowlan, S. J., & Hinton, G. E. (1991) · Neural Computation · doi:10.1162/neco.1991.3.1.79 · Cited by 5,384 · cited 3×
“MoEs are hierarchical models comprising an ensemble of N experts governed by a probabilistic gating function [12]–[15].”
BackgroundMethodA Light CNN for Deep Face Representation With Noisy Labels
Wu, X., He, R., Sun, Z., & Tan, T. (2018) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2018.2833032 · Cited by 1,186 · cited 1×
“To ensure consistency across all our experiments, we used a single speech deepfake detection model, a LCNN model [18], as both a baseline and expert within the MoE framework.”
MethodASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detection
Todisco, M., Wang, X., Vestman, V., Sahidullah, M., Delgado, H., Nautsch, A., Yamagishi, J., Evans, N., Kinnunen, T., & Lee, K. A. (2019) · arXiv · doi:10.48550/arxiv.1904.05441 · Cited by 688 · cited 1×
“ASVspoof 2019 [21] ( D ASV ).”
ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detection
Todisco, M., Wang, X., Vestman, V., Sahidullah, M., Delgado, H., Nautsch, A., Yamagishi, J., Evans, N., Kinnunen, T. H., & Lee, K. A. (2019) · Interspeech 2019 · doi:10.21437/interspeech.2019-2249 · Cited by 687
Twenty Years of Mixture of Experts
Yuksel, S. E., Wilson, J. N., & Gader, P. D. (2012) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2012.2200299 · Cited by 606
Mixture of experts: a literature survey
Masoudnia, S., & Ebrahimpour, R. (2014) · Artificial Intelligence Review · doi:10.1007/s10462-012-9338-y · Cited by 561 · cited 1×
“Indeed, this family of MoEs, also known as mixtures of implicitly localized experts [14], are rooted in the idea of partitioning a complex task into sub-problems easier for each expert to learn and solve, fundamentally amounting to a divide and conquer strategy.”
BackgroundASVspoof 2021: accelerating progress in spoofed and deepfake speech detection
Yamagishi, J., Wang, X., Todisco, M., Sahidullah, M., Patino, J., Nautsch, A., Liu, X., Lee, K. A., Kinnunen, T., Evans, N., et al. (2021) · 2021 Edition of the Automatic Speaker Verification and Spoofing Countermeasures Challenge · doi:10.21437/asvspoof.2021-8 · Cited by 380
Ito, K. A. (2024) · TIB Data Manager · doi:10.57702/yamk7ojd · Cited by 259
Does Audio Deepfake Detection Generalize?
Müller, N., Czempin, P., Diekmann, F., Froghyar, A., & Böttinger, K. (2022) · Interspeech 2022 · doi:10.21437/interspeech.2022-108 · Cited by 258
Research works printed but not linked: 1
- Veaux, 2016, Superseded-CSTR VCTK Corpus: English Multi-Speaker Corpus for CSTR Voice Cloning Toolkit
Cited by
Deepfake Media Forensics: Status and Future Challenges
Amerini, I., Barni, M., Battiato, S., Bestagini, P., Boato, G., Bruni, V., Caldelli, R., De Natale, F., De Nicola, R., Guarnera, L., et al. (2025) · Journal of Imaging · doi:10.3390/jimaging11030073 · Cited by 113
“To enhance the practicality of existing detection methods in real-world scenarios, new strategies have been proposed in [118–120].”
Self-citationPierno, A. D., Guarnera, L., Allegra, D., & Battiato, S. (2025) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn64981.2025.11229087 · Cited by 24
“In this work, we proposed RawNetLite , a lightweight end-to-end model for audio deepfake detection operating directly Authorized licensed use limited to the terms of the applicable license agreement with IEEE.”
Phoneme-Level Analysis for Person-of-Interest Speech Deepfake Detection
Salvi, D., Negroni, V., Mandelli, S., Bestagini, P., & Tubaro, S. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00169 · Cited by 15
“To this end, a variety of approaches have been explored, including transformer-based architectures [6, 14, 45], Mixture of Experts (MoE) models [26, 42], and others DL-based techniques [4, 19, 37, 44, 47].”
Self-citationBIM-Based Adversarial Attacks Against Speech Deepfake Detectors
Wang, W. E., Salvi, D., Negroni, V., Leonzio, D. U., Bestagini, P., & Tubaro, S. (2025) · Electronics · doi:10.3390/electronics14152967 · Cited by 9
Self-citationAnomaly Detection and Localization for Speech Deepfakes via Feature Pyramid Matching
Coletta, E., Salvi, D., Negroni, V., Leonzio, D. U., & Bestagini, P. (2025) · European Signal Processing Conference (EUSIPCO) · doi:10.23919/eusipco63237.2025.11226156 · Cited by 7
Self-citation
Metadata
- Authors
- Negroni, V., Salvi, D., Mezza, A. I., Bestagini, P., & Tubaro, S.
- Affiliations
- Politecnico di Milano (IT)
- Venue
- ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
- Publisher
- IEEE
- Published
- 2025-04-06
- Pages
- 1-5
- ISSN
- 1520-6149 (ISSN-L)
- OpenAlex
- W4408353669
- Cited by
- 14
- References
- 28
- Citations by year
- 2026: 4 · 2025: 10
- FWCI
- 13.28
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE International Conference on Acoustics Speech and Signal Processing (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- Virtual Community of Pathological Anatomy (University of Castilla La Mancha) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- Air Force Research Laboratory
- Ministry of Education
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Physical Sciences › Computer Science › Signal Processing › Speech and Audio Processing
- Keywords
- speech deepfake detection, mixture of experts, gating mechanism, generalization to unseen data, content authenticity
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Leveraging Mixture of Experts for Improved Speech Deepfake Detection. https://app.cochat.ai/papers/10.1109/icassp49660.2025.10890398
- The paper
- Negroni, V., Salvi, D., Mezza, A. I., Bestagini, P., & Tubaro, S. (2025). Leveraging Mixture of Experts for Improved Speech Deepfake Detection. ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). https://doi.org/10.1109/icassp49660.2025.10890398
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.