Pith. sign in

REVIEW 4 major objections 5 minor 30 references

Towards Source Attribution of Singing Voice Deepfake with Multimodal Foundation Models

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Multimodal foundation models can trace singing-voice deepfakes to their source generator.

desk verdict First SVDSA benchmark with a broad FM comparison; the MMFM finding is plausible, but the evaluation protocol and small fusion gain need hardening before the headline numbers can be trusted. read the letter →

arxiv 2506.03364 v1 pith:4I6O4NPM submitted 2025-06-03 eess.AS cs.MMcs.SD

Orchid Chetia Phukan , Girish , Mohd Mujtaba Akhtar , Swarup Ranjan Behera , Priyabrata Mallick , Pailla Balakrishna Reddy , Arun Balaji Buduru , Rajesh Sharma This is my paper · ORCID
classification eess.AScs.MMcs.SD
keywords singingvoicedeepfakesourceattributionmultimodalfoundationmodelsImageBindLanguageChernoffdistancedetectionCtrSVDD
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

This paper introduces singing voice deepfake source attribution (SVDSA), the task of identifying which generative model produced a fake singing clip rather than merely flagging it as synthetic. The authors hypothesize that multimodal foundation models, pre-trained across audio, image, and text, are better suited than speech-only or music-only models for capturing the subtle signature of each synthesis source. Testing ImageBind and LanguageBind against a battery of speech and music foundation models on the CtrSVDD benchmark, they find the multimodal models perform best. Their fusion framework COFFE, which aligns two models' feature distributions with a Chernoff distance loss, pushes performance to 91.16% accuracy and 3.63% equal error rate for the eight seen generators. If these numbers hold, the approach gives audio forensics a practical way to trace the origin of synthetic singing voices.

What carries the argument

The machinery is COFFE, a fusion framework that combines frozen embeddings from two foundation models. Each embedding passes through a small 1D-convolutional block; the two feature distributions are then aligned by minimizing the Chernoff distance $L_{CD} = -\log \sum_i p_i^s q_i^{1-s}$ between them, with $s$ balancing the two distributions. After alignment the features are concatenated and fed to a fully connected classifier. The total loss is $L = L_{CE} + \lambda L_{CD}$, with $\lambda = 0.1$. The Chernoff distance acts as an alignment loss: it is small when the two models' feature distributions overlap, so minimizing it pushes the models to represent the same source classes in compatible ways before their information is merged.

What would settle it

Re-running the same system on a fresh test partition whose clips are verified to be disjoint from training, or adding the held-out generators A09-A14 as a test set, would settle whether the 91.16% accuracy reflects genuine source attribution or leakage and closed-set optimism.

Watch

Extended reading notes

Core claim

The paper's central claim is that cross-modality pretraining, rather than scale or domain-specific music training, is what makes a foundation model effective for singing voice deepfake source attribution. On the CtrSVDD dataset, the multimodal models LanguageBind (82.37% accuracy) and ImageBind (81.92%) outperform the best speech model (MMS at 80.41%) and the best music model (MERT-v1-330M at 67.96%). The gap grows when the two multimodal models are fused: COFFE with LanguageBind and ImageBind reaches 91.16% accuracy, 90.03 F1, and 3.63% EER, beating every individual model and every concatenation-based fusion baseline. The authors interpret this as evidence that multimodal alignment captures source-specific traits—timbre, pitch manipulation, synthesis artifacts—that unimodal representations miss.

Load-bearing premise

The evaluation assumes the official CtrSVDD training and development splits can be used as training and testing sets for the same eight generators without any overlap or near-duplicate clips, and that attributing only these seen generators is the right benchmark.

Editorial extensions

If this is right

  • A forensic system could attribute a fake singing clip to one of eight known generators with over 91% accuracy using frozen multimodal embeddings and a small trainable head.
  • Fusion with an explicit alignment loss such as Chernoff distance is more effective than simple concatenation, so future fusion work should treat alignment as a first-class design choice.
  • Multimodal pretraining should be preferred over speech-only or music-only pretraining for forensic tasks involving singing voice, since it captures source artifacts unimodal models miss.
  • The reported CtrSVDD numbers become a benchmark that future SVDSA systems can be compared against.

Reading between the lines

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

  • The paper evaluates only the eight seen generators (A01-A08); a natural extension is open-set attribution, where an unknown generator appears at test time, but whether Chernoff alignment keeps unseen sources distinguishable is undemonstrated.
  • The two winning models anchor to different modalities (ImageBind to images, LanguageBind to language), so fusing models with different anchors may be the reason for the gain; other cross-modal pairs could show similar improvements.
  • Because the system uses frozen embeddings, it could be combined with existing singing voice deepfake detection models as a post-hoc attribution stage without retraining the detector, though the paper does not explore this.
  • The paper does not report per-source accuracies for the eight generators, so it is unknown whether attribution errors concentrate on acoustically similar synthesis methods; a per-source breakdown would clarify the failure modes.
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 / 5 minor

Summary. The paper introduces singing voice deepfake source attribution (SVDSA) as a task, evaluates 14 speech, music, and multimodal foundation models as frozen feature extractors with FCN and CNN downstream heads, and proposes COFFE, a fusion framework that uses a Chernoff-distance loss to align and concatenate FM embeddings. On the CtrSVDD benchmark, it reports that multimodal foundation models (LanguageBind, ImageBind) outperform unimodal models, and that COFFE with LB+IB attains the best numbers: 91.16% accuracy, 90.03 F1, and 3.63 EER. The central claims are that MMFMs are the most effective feature extractors for SVDSA and that the proposed fusion loss improves over simple concatenation.

Significance. If the results hold, the paper provides the first benchmark for singing voice deepfake source attribution and a useful empirical comparison across 14 foundation models. The released code is a concrete reproducibility asset, and the finding that multimodal models outperform unimodal speech/music models is plausible and of interest to the audio forensics community. The methodological novelty of COFFE is more modest, and the reported gain over concatenation fusion is small, so the strength of the contribution depends on the evaluation protocol being clean and the Chernoff-loss implementation being well defined.

major comments (4)
  1. [Section 4.1, Table 2] The evaluation protocol is not defined clearly enough to support the headline claim. The text says the official train and dev partitions (sources A01-A08) were 'used as training and testing', but it does not state how the two partitions were combined, whether utterances or recordings are disjoint across the train/test boundary, or whether the dev partition was used for early stopping or for selecting s, lambda, and the downstream architecture. If any of those decisions used the test portion, the 91.16% accuracy and the COFFE-vs-concatenation gap (91.16 vs 89.62 in Table 2) are optimistically biased. Please state the exact split, verify that there is no content overlap between training and testing clips, and either report results on the held-out eval generators A09-A14 or explicitly qualify the claims as closed-set seen-source attribution.
  2. [Section 3.1, Eq. (1)] Equation (1) defines the Chernoff loss as a sum over a shared index i of p_i^s * q_i^{1-s}, but the two FMs in the headline COFFE run (LB and IB) produce 768- and 1024-dimensional vectors, so the sum over a common i is not defined as written. The paper must specify the projection, pooling, or normalization used to obtain common-dimension probability distributions from the two feature vectors, and explain how s is applied in that construction. Without this, COFFE cannot be reproduced from the text, and the claimed advantage over concatenation cannot be attributed to the Chernoff loss.
  3. [Section 4.1] The hyperparameters s=0.3 and lambda=0.1 are reported as chosen 'after some initial experimentation', but no held-out validation set is described. Since the same data appear to be used for model selection and for final testing, the comparison against concatenation baselines (which share the same training details but not the same loss) is not a fully controlled comparison. Please report sensitivity curves over s and lambda, or select these values on a separate validation split before computing the final test numbers.
  4. [Section 4.2, Table 2] All reported numbers are point estimates with no error bars or significance tests. The difference between COFFE and concatenation for LB+IB is 1.54 accuracy points, so without multiple seeds or a statistical test, the claim that COFFE 'outperforms' baseline fusion is not established beyond run-to-run variation. Please report mean and standard deviation over at least a few independent training runs, or otherwise provide evidence that the gap is not noise.
minor comments (5)
  1. [Figure 1 caption] The caption spells the proposed framework as 'COFEE' instead of 'COFFE'.
  2. [Table 1 caption] The abbreviation list includes 'HUB: HuBERT', but no HuBERT row appears in Table 1; remove the entry or add the corresponding results.
  3. [Table 2 heading] The word 'Evalution' should be 'Evaluation'.
  4. [Figure 3] The confusion matrices compare COFFE(LB+IB) with CNN(LB), but the more informative baseline for the fusion claim is the concatenation-based LB+IB fusion; please include that comparison.
  5. [Abstract and Section 1] The phrase 'symphony of MMFMs' is vague; the paper should state explicitly that the best result is obtained by fusing LanguageBind and ImageBind.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's claims are empirical benchmark results, not derivations that reduce to their inputs.

full rationale

The central claims—that MMFMs outperform SFMs/MFMs for singing-voice deepfake source attribution, and that COFFE with LB+IB fusion reaches 91.16% accuracy/90.03 F1/3.63 EER—are established by training classifiers and reporting test-set numbers on CtrSVDD. No equation or fitted parameter is renamed as a prediction: Eq. (1) defines a Chernoff-distance alignment loss used during training, and the reported accuracy is the downstream classifier's measured performance, not a quantity implied by the loss definition. The hyperparameters s=0.3 and λ=0.1 were chosen 'after some initial experimentation' (Section 4.1), which raises a legitimate validation-protocol concern if the dev split was used for model selection, but this is a tuning/reproducibility issue rather than a definitional circularity. The only overlapping-author citation ([10], used to motivate fusion of foundation models) is accompanied by additional independent citations and is not load-bearing: the paper does not invoke a uniqueness theorem or an unverified result from the authors' prior work to force its choice of COFFE. The derivation chain is therefore self-contained with respect to circularity; the ambiguous train/dev usage belongs under correctness risk, not under circularity.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new particles or forces; it introduces a task, a benchmark protocol, and a loss-based fusion method. The main assumptions are about dataset partitioning, feature pooling, and the validity of closed-set evaluation.

free parameters (5)
  • s (Chernoff exponent) = 0.3
    Set after 'some initial experimentation' (Section 4.1); no sensitivity analysis or held-out tuning reported.
  • lambda (CD loss weight) = 0.1
    Set after initial experimentation (Section 4.1); no sensitivity analysis.
  • learning rate = 1e-3
    Standard Adam setting, not problem-specific.
  • dropout and early stopping = not specified
    Mentioned but no quantitative stopping criterion, which affects reproducibility.
  • downstream head architecture = CNN: 64-128 filters, FCN: 128 dense
    Chosen by the authors without justification or comparison with other head designs.
assumptions (4)
  • domain assumption The CtrSVDD official train and dev splits can be used as training and test partitions without data leakage.
    Section 4.1 says 'we used them as training and testing' but does not state how the split was performed; if this premise fails, reported accuracies are inflated.
  • ad hoc to paper Average pooling of the final hidden layer from frozen FMs preserves source-specific discriminative information.
    Feature aggregation is a design choice not validated against alternatives like attention pooling (Section 2).
  • domain assumption Closed-set accuracy on sources A01-A08 is a valid measure of SVDSA capability.
    The eval-set sources A09-A14 are excluded, so the benchmark does not test attribution of unseen generators (Section 4.1).
  • ad hoc to paper The Chernoff coefficient with s=0.3 is a suitable alignment objective for fusing FM embeddings.
    No comparison with other divergence losses and no justification for why s=0.3 is optimal (Section 3.1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Source Attribution of Singing Voice Deepfake with Multimodal Foundation Models." pith.science (2026). https://pith.science/paper/4I6O4NPM

@misc{pith2026250603364,
  author       = {Pith},
  title        = {Pith review of: Towards Source Attribution of Singing Voice Deepfake with Multimodal Foundation Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4I6O4NPM}},
  note         = {Machine review of arXiv:2506.03364}
}
read the original abstract

In this work, we introduce the task of singing voice deepfake source attribution (SVDSA). We hypothesize that multimodal foundation models (MMFMs) such as ImageBind, LanguageBind will be most effective for SVDSA as they are better equipped for capturing subtle source-specific characteristics-such as unique timbre, pitch manipulation, or synthesis artifacts of each singing voice deepfake source due to their cross-modality pre-training. Our experiments with MMFMs, speech foundation models and music foundation models verify the hypothesis that MMFMs are the most effective for SVDSA. Furthermore, inspired from related research, we also explore fusion of foundation models (FMs) for improved SVDSA. To this end, we propose a novel framework, COFFE which employs Chernoff Distance as novel loss function for effective fusion of FMs. Through COFFE with the symphony of MMFMs, we attain the topmost performance in comparison to all the individual FMs and baseline fusion methods.

Figures

Figures reproduced from arXiv: 2506.03364 by the authors.

Figure 1
Figure 1. Proposed Modeling Framework for fusion of FMs: COFEE; CD represents the Chernoff Distance; LCD, LCE, L stands for Chernoff Distance Loss, Cross-entropy loss, Total loss respectively https://github.com/Helix-IIIT-Delhi/COFFE-Singing_ Voice_Deepfake 2. Foundation Models In this section, we first discuss the SFMs followed by MFMs and MMFMs considered in our study. Speech Foundation Models: We consider WavLM1 [14] and U… view at source ↗
Figure 2
Figure 2. t-SNE Plots- (a) XLS-R (b) MERT-v1-330M (c) Lan￾guageBind (d) ImageBind (a) (b) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Confusion matrices: (a) COFEE(LB + IB) (b) CNN(LB); The x-axis represents the predicted values, while the y-axis denotes the true values all the combinations of FMs as well as individual FMs. This shows that combination of MMFMs further brings out the com￾plementary behavior among them. These results show the effi￾cacy of COFFE for effective fusion of FMs for improved SVDSA. Additionally, the confusion matrices in … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 17 canonical work pages

  1. [1]

    Towards Source Attribution of Singing Voice Deepfake with Multimodal Foundation Models

    Introduction "Imagine discovering a new song by your favorite artist, only to learn they never recorded it." With generative technologies advancing at an unprecedented pace, this scenario is no longer hypothetical. Singing voice deepfakes (SVDs) have evolved to a level where they can convincingly mimic an artist’s vocal timbre, seamlessly intertwining spe...

    work page Pith review arXiv 2025

  2. [2]

    Speech Foundation Models : We consider WavLM1 [14] and Unispeech-SAT2 [15] which are SOTA SFMs in SUPERB

    Foundation Models In this section, we first discuss the SFMs followed by MFMs and MMFMs considered in our study. Speech Foundation Models : We consider WavLM1 [14] and Unispeech-SAT2 [15] which are SOTA SFMs in SUPERB. Unispeech-SAT incorporates contrastive utterance-wise loss and speaker-aware training while WavLM does its pre-training through masked spe...

    work page

  3. [3]

    We implemented two distinct downstream for individual FMs—Fully Connected Network (FCN) and CNN

    Modeling In this section, we first discuss the downstream network with individual FMs followed by our proposed framework for fusion of FMs, COFFE. We implemented two distinct downstream for individual FMs—Fully Connected Network (FCN) and CNN. The CNN consists of two 1D convolutional layers with 64 and 128 filters (kernel size = 3), each followed by max p...

    work page

  4. [4]

    Dataset We utilized the CtrSVDD [24], a benchmark dataset specifically designed for SVDD and the audio samples are in Chinese and Japanese

    Experiment 4.1. Dataset We utilized the CtrSVDD [24], a benchmark dataset specifically designed for SVDD and the audio samples are in Chinese and Japanese. We only consider the synthetic samples for our experi- ments and it includes 188,486 clips, totaling 260.34 hours. The synthetic samples are generated through 14 distinct synthesis methods (A01–A14). T...

    work page

  5. [5]

    Table 2 presents the evaluation scores for modeling with various combinations of SFMs

    Through these plots, we observe better clustering across the source classes in MMFMs and providing support to our hypothe- sis and obtained experimental results. Table 2 presents the evaluation scores for modeling with various combinations of SFMs. We use concatenation-based fusion as baseline technique. We keep the modeling the same as COFFE except the C...

    work page

  6. [6]

    MMFMs such as IB and LB, excel in capturing source-specific traits like timbre, pitch manipulation, and synthesis artifacts due to their cross- modal pretraining

    Conclusion In this study, we introduce the task of SVDSA and demonstrate that MMFMs are the most effective for SVDSA. MMFMs such as IB and LB, excel in capturing source-specific traits like timbre, pitch manipulation, and synthesis artifacts due to their cross- modal pretraining. Building on these insights, we proposeCOFFE, a novel fusion framework that l...

    work page

  7. [7]

    Singfake: Singing voice deepfake detection,

    Y . Zang, Y . Zhang, M. Heydari, and Z. Duan, “Singfake: Singing voice deepfake detection,” in ICASSP 2024-2024 IEEE Interna- tional Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2024, pp. 12 156–12 160

    work page 2024

  8. [8]

    Ctrsvdd: A benchmark dataset and baseline analysis for controlled singing voice deepfake detection,

    Y . Zang, J. Shi, Y . Zhang, R. Yamamoto, J. Han, Y . Tang, S. Xu, W. Zhao, J. Guo, T. Toda, and Z. Duan, “Ctrsvdd: A benchmark dataset and baseline analysis for controlled singing voice deepfake detection,” in Interspeech 2024, 2024, pp. 4783–4787

    work page 2024

Show all 30 references
  1. [9]

    Svdd 2024: The inaugural singing voice deepfake detection chal- lenge,

    Y . Zhang, Y . Zang, J. Shi, R. Yamamoto, T. Toda, and Z. Duan, “Svdd 2024: The inaugural singing voice deepfake detection chal- lenge,” in 2024 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2024, pp. 782–787

    2024

  2. [10]

    Source tracing: Detect- ing voice spoofing,

    T. Zhu, X. Wang, X. Qin, and M. Li, “Source tracing: Detect- ing voice spoofing,” in Proc. Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), 2022

    2022

  3. [11]

    An initial investigation for detecting vocoder fingerprints of fake audio,

    X. Yan, J. Yi, J. Tao, C. Wang, H. Ma, T. Wang, S. Wang, and R. Fu, “An initial investigation for detecting vocoder fingerprints of fake audio,” in Proc. of the 1st International Workshop on Deepfake Detection for Audio Multimedia, 2022

    2022

  4. [12]

    Distinguish- ing neural speech synthesis models through fingerprints in speech waveforms,

    C. Y . Zhang, J. Yi, J. Tao, C. Wang, and X. Yan, “Distinguish- ing neural speech synthesis models through fingerprints in speech waveforms,” in China National Conference on Chinese Computa- tional Linguistics. Springer, 2024, pp. 259–273

    2024

  5. [13]

    Attacker attribution of audio deepfakes,

    N. Müller, F. Diekmann, and J. Williams, “Attacker attribution of audio deepfakes,” in Interspeech 2022, 2022, pp. 2788–2792

    2022

  6. [14]

    Source tracing of audio deepfake systems,

    N. Klein, T. Chen, H. Tak, R. Casal, and E. Khoury, “Source tracing of audio deepfake systems,” in Interspeech 2024, 2024, pp. 1100–1104

    2024

  7. [15]

    Attribu- tion of diffusion based deepfake speech generators,

    K. Bhagtani, A. K. S. Yadav, P. Bestagini, and E. J. Delp, “Attribu- tion of diffusion based deepfake speech generators,” in 2024 IEEE International Workshop on Information Forensics and Security (WIFS). IEEE, 2024, pp. 1–6

    2024

  8. [16]

    Heterogeneity over homogeneity: Investigating multilingual speech pre-trained models for detecting audio deepfake,

    O. Chetia Phukan, G. Kashyap, A. B. Buduru, and R. Sharma, “Heterogeneity over homogeneity: Investigating multilingual speech pre-trained models for detecting audio deepfake,” in Findings of the Association for Computational Linguistics: NAACL 2024, K. Duh, H. Gomez, and S. Be...

    2024

  9. [17]

    Singing voice graph modeling for singfake detection,

    X.-B. Chen, H. Wu, J.-S. R. Jang, and H. yi Lee, “Singing voice graph modeling for singfake detection,” Interspeech 2024, 2024

    2024

  10. [18]

    Investi- gation of ensemble features of self-supervised pretrained models for automatic speech recognition,

    A. Arunkumar, V . Nileshkumar Sukhadia, and S. Umesh, “Investi- gation of ensemble features of self-supervised pretrained models for automatic speech recognition,” in Interspeech 2022, 2022, pp. 5145–5149

    2022

  11. [19]

    Speech foundation model ensembles for the controlled singing voice deep- fake detection (ctrsvdd) challenge 2024,

    A. Guragain, T. Liu, Z. Pan, H. B. Sailor, and Q. Wang, “Speech foundation model ensembles for the controlled singing voice deep- fake detection (ctrsvdd) challenge 2024,” in 2024 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2024, pp. 774–781

    2024

  12. [20]

    wav2vec 2.0: A framework for self-supervised learning of speech representations,

    A. Baevski, Y . Zhou, A. Mohamed, and M. Auli, “wav2vec 2.0: A framework for self-supervised learning of speech representations,” Advances in neural information processing systems , vol. 33, pp. 12 449–12 460, 2020

    2020

  13. [21]

    Unispeech-sat: Universal speech repre- sentation learning with speaker aware pre-training,

    S. Chen, Y . Wu, C. Wang, Z. Chen, Z. Chen, S. Liu, J. Wu, Y . Qian, F. Wei, J. Li, and X. Yu, “Unispeech-sat: Universal speech repre- sentation learning with speaker aware pre-training,” ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processi...

    2022

  14. [22]

    Xls-r: Self-supervised cross-lingual speech representation learning at scale,

    A. Babu, C. Wang, A. Tjandra, K. Lakhotia, Q. Xu, N. Goyal, K. Singh, P. von Platen, Y . Saraf, J. Pino, A. Baevski, A. Con- neau, and M. Auli, “Xls-r: Self-supervised cross-lingual speech representation learning at scale,” in Interspeech 2022, 2022, pp. 2278–2282

    2022

  15. [23]

    Robust speech recognition via large-scale weak supervision,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak supervision,” in International conference on machine learning . PMLR, 2023, pp. 28 492–28 518

    2023

  16. [24]

    Scaling speech technology to 1,000+ languages,

    V . Pratap, A. Tjandra, B. Shi, P. Tomasello, A. Babu, S. Kundu, A. Elkahky, Z. Ni, A. Vyas, M. Fazel-Zarandi et al., “Scaling speech technology to 1,000+ languages,” Journal of Machine Learning Research, vol. 25, no. 97, pp. 1–52, 2024

    2024

  17. [25]

    X-vectors: Robust dnn embeddings for speaker recognition,

    D. Snyder, D. Garcia-Romero, G. Sell, D. Povey, and S. Khudanpur, “X-vectors: Robust dnn embeddings for speaker recognition,”2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 5329–5333, 2018

    2018

  18. [26]

    Mert: Acoustic music understanding model with large-scale self-supervised training,

    Y . Li, R. Yuan, G. Zhang, Y . Ma, X. Chen, H. Yin, C. Xiao, C. Lin, A. Ragni, E. Benetos et al., “Mert: Acoustic music understanding model with large-scale self-supervised training,” arXiv preprint arXiv:2306.00107, 2023

    2023 arXiv

  19. [27]

    Map- music2vec: A simple and effective baseline for self-supervised music audio representation learning,

    Y . Li, R. Yuan, G. Zhang, Y . Ma, C. Lin, X. Chen, A. Ragni, H. Yin, Z. Hu, H. He, E. Benetos, N. Gyenge, R. Liu, and J. Fu, “Map- music2vec: A simple and effective baseline for self-supervised music audio representation learning,” ArXiv, vol. abs/2212.02508, 2022

    2022 arXiv

  20. [28]

    Imagebind: One embedding space to bind them all,

    R. Girdhar, A. El-Nouby, Z. Liu, M. Singh, K. V . Alwala, A. Joulin, and I. Misra, “Imagebind: One embedding space to bind them all,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 15 180–15 190

    2023

  21. [29]

    Languagebind: Extending video-language pretraining to n-modality by language-based semantic alignment,

    B. Zhu, B. Lin, M. Ning, Y . Yan, J. Cui, H. Wang, Y . Pang, W. Jiang, J. Zhang, Z. Li et al., “Languagebind: Extending video-language pretraining to n-modality by language-based semantic alignment,” arXiv preprint arXiv:2310.01852, 2023

    arXiv

  22. [30]

    Svdd challenge 2024: A singing voice deepfake detection challenge (ctrsvdd track, training/development set),

    Y . Zhang, Y . Zang, J. Shi, R. Yamamoto, J. Han, Y . Tang, T. Toda, and Z. Duan, “Svdd challenge 2024: A singing voice deepfake detection challenge (ctrsvdd track, training/development set),” 2024. [Online]. Available: https://doi.org/10.5281/zenodo. 10467648

    2024 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.