Pith. sign in

REVIEW 2 major objections 3 minor 43 references

Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge

T0 review · 2 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read MC-S2SND, a two-pass multi-channel extension of the sequence-to-sequence diarization framework, posts an 8.09% diarization error rate on the MISP 2025 evaluation set — the best score in the challenge's speaker-diarization task.

desk verdict A clean, honest competition system paper whose central comparison is confounded: MC-S2SND's gain over S2SND includes extra real-data fine-tuning, so the channel-attention contribution itself is not isolated. read the letter →

arxiv 2505.16387 v1 pith:GPMEZITV submitted 2025-05-22 eess.AS

Ming Cheng , Fei Su , Cancan Li , Juan Liu , Ming Li This is my paper · ORCID
classification eess.AS
keywords speakerdiarizationmulti-channelaudiosequence-to-sequenceneuralchannelattentiontarget-speakervoiceactivitydetectionmicrophonearrayfusionerrorratetwo-passinference
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

The paper claims that speaker diarization — deciding who spoke when in a meeting — improves when a single model does the job in two passes over a microphone array. The first pass uses one channel to produce a rough "who spoke when" draft; the second pass feeds that draft back into a paired decoder that extracts each listed speaker's voice profile from all eight channels, then re-answers the question using that profile as enrollment. On the MISP 2025 Challenge evaluation set this two-pass design reaches an 8.09% diarization error rate, down from 11.48% for the single-channel model and good for first place on the task's leaderboard. A careful reader would care because the design shows that the sequence-to-sequence way of doing diarization can be extended from single-channel online use to offline multi-channel use without giving up its unified, clustering-free architecture.

What carries the argument

The load-bearing object is the paired decoder design inherited from S2SND and turned into a two-pass loop. The representation decoder consumes the initial diarization labels as auxiliary queries and outputs per-speaker voice embeddings; the detection decoder consumes those embeddings as auxiliary queries and outputs refined voice activities, so extraction and detection are inverse tasks trained inside one model with a binary cross-entropy loss and an ArcFace loss. The single architectural addition is the channel-attention block, a two-layer Transformer that runs self-attention on the channel axis of the multi-channel features and averages across channels, turning the eight-microphone array into one fused feature sequence. Two inference-time mechanisms carry much of the measured gain: K-Means clustering of block-wise speaker embeddings seeded by each speaker's average embedding, and score-level fusion that averages the overlapping predictions of neighbouring audio blocks.

What would settle it

Run the two-pass pipeline on the MISP-Meeting evaluation audio twice: once with the first-pass draft as the enrollment source and once with the human-annotated speaker labels in its place. If the refined error rate does not improve (or worsens) with the cleaner labels, the claim that the representation decoder extracts usable voice profiles from the initial draft is not supported; and if deliberately deleting one speaker from the draft leaves the final output effectively unchanged, the refinement stage is not actually carrying the result.

Watch

Extended reading notes

Core claim

The paper's central claim is that its Multi-Channel Sequence-to-Sequence Neural Diarization (MC-S2SND) model significantly outperforms the original single-channel S2SND model in offline multi-channel inference. The system works in two passes: the original S2SND reads the first channel and produces initial voice-activity labels; MC-S2SND then takes multi-channel features plus those labels, uses its representation decoder to extract target-voice speaker embeddings, and uses its detection decoder to predict a refined diarization result. The only architectural difference from the parent model is a channel-attention block that applies Transformer self-attention across the microphone axis before fusing the channels. Trained jointly with binary cross-entropy on voice activities and ArcFace on the extracted embeddings, and boosted by embedding clustering and score-level fusion over block-wise predictions, the system reports an 8.09% diarization error rate on the MISP-Meeting evaluation set after adaptation, the best result in the MISP 2025 speaker-diarization task.

Load-bearing premise

The whole refinement loop depends on the first pass being roughly right, because the second pass extracts each speaker's voice profile from that initial draft, so any speaker the draft misses or mislabels in overlapped speech is likely to stay wrong in the final output.

Editorial extensions

If this is right

  • If the two-pass loop holds up beyond this challenge set, multi-channel diarization no longer needs a separate clustering or speaker-counting module: the same network both enrolls speakers and detects their activity.
  • The reported ablations tie the gain to the channel-attention front end and to the two block-wise inference tricks, so the same recipe (attention-based channel fusion plus embedding clustering and score fusion) should transfer to other array-equipped diarization systems.
  • The 8.62% to 8.09% drop after adapting on real meeting data indicates that a model trained partly on simulated single-channel data can be cheaply re-tuned to a real multi-channel corpus once the channel-attention path exists.
  • Because MC-S2SND is initialized from the S2SND weights, upgrading an existing single-channel system to multi-channel is a fine-tuning step rather than a full re-training from scratch.

Reading between the lines

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

  • A direct consequence the authors do not draw is that the same two-pass enrollment loop could feed a multi-talker speech recognizer, since the refined output carries per-speaker identities and time boundaries in one artifact, exactly what meeting transcription systems consume.
  • The load-bearing dependence on the initial draft suggests a testable extension: corrupt the initial labels during training (drop whole speakers, scramble speaker order) to make the representation decoder robust to first-pass errors, which would remove the main failure mode of the pipeline.
  • Because the channel-attention module is small relative to the rest of the network, the approach likely transfers to other array geometries (4, 16, or binaural channels) by fine-tuning only the new block; that is an experiment the paper does not report.
  • Since the MISP 2025 data also includes a panoramic camera track, an untested variant would be to treat the video stream as extra channels in the same attention-fusion design, extending audio-only fusion to audio-visual fusion.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 3 minor

Summary. The paper describes a speaker diarization system for the MISP 2025 Challenge. It first applies the authors' prior Sequence-to-Sequence Neural Diarization (S2SND) system to single-channel audio to obtain an initial diarization result, then extends S2SND with a channel-attention module and multi-channel input to create MC-S2SND, which refines the initial predictions. The paper reports a final DER of 8.09% on the challenge evaluation set, ranking first in the diarization task. The main experimental evidence is Table 1, which compares S2SND and MC-S2SND under different inference settings (embedding clustering and block-shift fusion).

Significance. If the reported gains were attributable to the proposed architecture, the paper would be a useful contribution to multi-channel diarization, and the two-pass design of using an initial single-channel diarization result to drive embedding extraction for a second-stage refinement is an interesting and practical idea. The paper is also honest in reporting that the final system is obtained by adapting on real training plus development data. The ablation of inference-stage techniques (embedding clustering and score-level fusion) is internally consistent and shows clear improvements for both S2SND and MC-S2SND. However, the central comparative claim is weakened by a training-protocol confound, and the evaluation is based on a single 3-hour evaluation set with no statistical uncertainty quantification. The contribution is therefore best viewed as a challenge system description rather than an established architectural claim until the confound is addressed.

major comments (2)
  1. [Table 1; Sections 3.3.2 and 3.3.3] The comparison of S2SND System #4 (11.48% DER) with MC-S2SND System #8 (8.62% DER) is confounded by differences in training schedule. S2SND is trained in three stages with 50% simulated and 50% real data, while MC-S2SND is initialized from the trained S2SND checkpoint and then undergoes two additional stages on real data only, including full unfreezing of all weights at learning rate 1e-5 (Section 3.3.3). System #9 shows that an additional adaptation pass on training and development data alone reduces DER by 0.53% absolute, demonstrating that the extra real-data fine-tuning can produce nontrivial gains. Without a control that isolates the channel-attention module (e.g., an S2SND model fine-tuned on the same real-data schedule without CH-Attention, or an MC-S2SND model fed only a single channel), the paper's claim that MC-S2SND 'significantly outperforms' the original S2SND is not established; the observed gain may be due in whole or in part to the additional training on real data.
  2. [Section 4; Table 1] The evaluation set contains only 3 hours of audio (Section 3.1), and Table 1 reports single DER point estimates with no confidence intervals, bootstrap resampling, or significance tests. The word 'significantly' in Section 4 ('our newly proposed MC-S2SND model significantly outperforms the original S2SND model') is therefore not supported statistically. Given the small evaluation set, even the 0.53% gap between System #8 and System #9 could be within sampling variability, and the 2.86% gap between System #4 and System #8 should be accompanied by an uncertainty estimate before being described as significant. I request either confidence intervals (e.g., per-recording bootstrap) or at least a per-recording breakdown of DER, and a softening of 'significantly' if no statistical test is provided.
minor comments (3)
  1. [Figure 2; Section 2.2.1] The headings 'T arget-Speaker Voice Activities' and 'V oice Activities' appear with stray spaces in the rendered figure and text; please fix the typography.
  2. [Section 3.4.1] The K-means step is described as using the average embedding of each speaker as initial centroids, but it is not stated how the number of speakers N is determined during inference when the initial diarization result may contain a variable number of speakers. Please clarify whether N is taken from the initial S2SND output or from a separate speaker-counting step.
  3. [Section 1] The introduction states that S2SND models 'significantly outperform' previous state-of-the-art methods on DIHARD-II and DIHARD-III, citing a preprint. Please provide the exact DER numbers or a pointer to the relevant table in that work, since 'significantly' here is a quantitative claim that should be verifiable without accessing an unpublished companion.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported DER improvements are measured against a held-out evaluation benchmark and do not reduce to fitted parameters or self-citation chains.

full rationale

The paper's central claim is an empirical system description for the MISP 2025 Challenge: the MC-S2SND model is trained to predict speaker voice activities from annotations and then evaluated on a held-out evaluation set with an external DER metric. No equation in the paper defines the predicted output in terms of the evaluation target, and no fitted parameter is renamed as a prediction. The system does rely on the authors' prior S2SND work [17] for the base architecture and training procedure, and the MC-S2SND weights are initialized from the authors' own S2SND checkpoint; however, this self-citation and initialization are not used to establish the reported 8.09% DER, which comes from the challenge leaderboard ground truth. The two-pass inference uses an initial diarization result Y as an input to the representation decoder; this is a stated conditional design, not a circular derivation, because the final output is still compared against external annotations and the initial Y is not a fitted parameter of the evaluation target. The main experimental weakness is a potential confound: MC-S2SND receives additional real-data fine-tuning relative to the S2SND baseline (Sec. 3.3.3), so the Table 1 gain may not be attributable solely to the channel-attention module. That is a correctness or ablation concern, not circularity. Therefore the derivation chain is self-contained with respect to circularity.

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

The paper is an empirical systems paper; no free parameters are fitted to the evaluation set, and hyperparameters are inherited from prior work. The main axioms are standard domain assumptions about the effectiveness of pretrained speaker features, the validity of simulated training data, the correctness of challenge annotations, and the ability of channel attention to fuse microphone channels.

assumptions (4)
  • domain assumption The MISP-Meeting evaluation set annotations are accurate and the reported DER is computed correctly.
    The paper does not provide error analysis or independent verification; the leaderboard is self-reported.
  • domain assumption The pretrained speaker verification extractor (trained on VoxBlink2 and VoxCeleb2) provides a generalizable embedding space for the challenge data.
    The extractor is pretrained on different corpora and is used as the front-end for both S2SND and MC-S2SND.
  • domain assumption The on-the-fly simulated data generation from VoxCeleb2, VoxBlink2, KeSpeech, and 3D-Speaker is effective for training S2SND.
    The simulation method is not described in this paper and is inherited from prior works.
  • domain assumption Channel attention along the channel axis can effectively combine multi-channel information.
    No analysis of channel attention behavior is provided; the improvement is inferred from the ablation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge." pith.science (2026). https://pith.science/paper/GPMEZITV

@misc{pith2026250516387,
  author       = {Pith},
  title        = {Pith review of: Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GPMEZITV}},
  note         = {Machine review of arXiv:2505.16387}
}
read the original abstract

This paper describes the speaker diarization system developed for the Multimodal Information-Based Speech Processing (MISP) 2025 Challenge. First, we utilize the Sequence-to-Sequence Neural Diarization (S2SND) framework to generate initial predictions using single-channel audio. Then, we extend the original S2SND framework to create a new version, Multi-Channel Sequence-to-Sequence Neural Diarization (MC-S2SND), which refines the initial results using multi-channel audio. The final system achieves a diarization error rate (DER) of 8.09% on the evaluation set of the competition database, ranking first place in the speaker diarization task of the MISP 2025 Challenge.

Figures

Figures reproduced from arXiv: 2505.16387 by the authors.

Figure 1
Figure 1. Overview of our developed system. The dashed parts represent the prior single-channel diarization method [17]. Recently, a novel Sequence-to-Sequence Neural Diariza￾tion (S2SND) framework [17] has been proposed for speaker di￾arization. By utilizing an automatic speaker detection and rep￾resentation technique, it eliminates the need for unsupervised clustering methods (e.g., K-Means [3], SC [4], AHC [18]) and the PI… view at source ↗
Figure 2
Figure 2. Multi-Channel Sequence-to-Sequence Neural Diarization (MC-S2SND) framework. Det. and Rep. denote the abbreviations of detection and representation, respectively. 2.1. Architecture 2.1.1. Front-End The front-end module comprises the extractor and channel at￾tention (CH-Attention) block. The extractor is based on a ResNet model [22] integrated with segmental statistical pooling (SSP) [11]. For the input audio with C c… view at source ↗
Figure 3
Figure 3. Leaderboard of the MISP 2025 Challenge (Task 1). The official baseline and top-ranked systems are plotted, where team names have been anonymized. Systems #1-4 represent the S2SND model with different in￾ference settings. The ablation results show that the embedding clustering and dense score-level fusion strategies contribute to improved diarization performance. System #4, which combines both strategies, achieves th… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 30 canonical work pages

  1. [1]

    Who-Spoke-When

    Introduction Speaker diarization refers to the process of identifying each speaker’s utterance boundaries in conversational data, address- ing the “Who-Spoke-When” problem [1]. It is essential for various downstream speech-related tasks, such as multi-talker speech recognition [2]. Many classical studies have been proposed for speaker di- arization. Conve...

    work page 2025

  2. [2]

    Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge

    Methodology Fig. 2 demonstrates the proposed Multi-Channel Sequence-to- Sequence Neural Diarization (MC-S2SND) framework, which is partially modified from the original S2SND [17] model. The details are described as follows. arXiv:2505.16387v1 [eess.AS] 22 May 2025 Front-End Det. DecoderEncoder Rep. Decoder Speaker Embeddings as Auxiliary Queries V oice Ac...

    work page Pith review arXiv 2025

  3. [3]

    Datasets For the simulated data, we combine the V oxCeleb2 [26], V oxBlink2 [27], KeSpeech [28], and 3D-Speaker [29] datasets to create a large-scale corpus with 153,738 identities

    Experimental Setup 3.1. Datasets For the simulated data, we combine the V oxCeleb2 [26], V oxBlink2 [27], KeSpeech [28], and 3D-Speaker [29] datasets to create a large-scale corpus with 153,738 identities. Then, an on-the-fly simulation method can generate new training data, which is used in our previous works [12, 17, 30]. For the real data, the MISP-Mee...

    work page 2025

  4. [4]

    Table 1 illustrates the performance of the systems we developed on the MISP-Meeting evaluation set

    Results The evaluation metric is the diarization error rate (DER) with- out collar tolerance. Table 1 illustrates the performance of the systems we developed on the MISP-Meeting evaluation set. 2https://scikit-learn.org/stable/ Ours Team A Team B Team C Baseline 0 2 4 6 8 10 12 14 16DER (%) 8.09 8.18 8.33 8.88 15.52 Figure 3:Leaderboard of the MISP 2025 C...

    work page 2025

  5. [5]

    Com- pared to the original S2SND method, the modified MC-S2SND model effectively processes multi-channel audio to enhance di- arization performance in offline scenarios

    Conclusions This paper describes our proposed Multi-Channel Sequence-to- Sequence Neural Diarization (MC-S2SND) framework for the speaker diarization task in the MISP 2025 Challenge. Com- pared to the original S2SND method, the modified MC-S2SND model effectively processes multi-channel audio to enhance di- arization performance in offline scenarios. As s...

    work page 2025

  6. [6]

    Many thanks for the computational resource provided by the Advanced Computing East China Sub-Center

    Acknowledgements This research is funded in part by the National Natural Sci- ence Foundation of China (62171207), Yangtze River Delta Science and Technology Innovation Community Joint Research Project (2024CSJGG01100), Science and Technology Program of Suzhou City(SYC2022051) and Guangdong Science and Technology Plan (2023A1111120012). Many thanks for th...

    work page

  7. [7]

    A review of speaker diarization: Recent advances with deep learning,

    T. J. Park, N. Kanda, D. Dimitriadis, K. J. Han, S. Watanabe, and S. Narayanan, “A review of speaker diarization: Recent advances with deep learning,”Computer Speech & Language, vol. 72, p. 101317, 2022

    2022

  8. [8]

    Joint speaker counting, speech recognition, and speaker identification for overlapped speech of any number of speakers,

    N. Kanda, Y . Gaur, X. Wang, Z. Meng, Z. Chen, T. Zhou, and T. Yoshioka, “Joint speaker counting, speech recognition, and speaker identification for overlapped speech of any number of speakers,” inProc. INTERSPEECH, 2020, pp. 36–40

    work page 2020

Show all 43 references
  1. [9]

    Speaker diarization with lstm,

    Q. Wang, C. Downey, L. Wan, P. A. Mansfield, and I. L. Moreno, “Speaker diarization with lstm,” inProc. ICASSP, 2018, pp. 5239–5243

    2018

  2. [10]

    Lstm based sim- ilarity measurement with spectral clustering for speaker diariza- tion,

    Q. Lin, R. Yin, M. Li, H. Bredin, and C. Barras, “Lstm based sim- ilarity measurement with spectral clustering for speaker diariza- tion,” inProc. INTERSPEECH, 2019, pp. 366–370

    2019

  3. [11]

    Bayesian hmm clustering of x-vector sequences (vbx) in speaker diarization: Theory, implementation and analysis on standard tasks,

    F. Landini, J. Profant, M. Diez, and L. Burget, “Bayesian hmm clustering of x-vector sequences (vbx) in speaker diarization: Theory, implementation and analysis on standard tasks,”Com- puter Speech & Language, vol. 71, p. 101254, 2022

    2022

  4. [12]

    End-to-end neural speaker diarization with permutation-free objectives,

    Y . Fujita, N. Kanda, S. Horiguchi, K. Nagamatsu, and S. Watan- abe, “End-to-end neural speaker diarization with permutation-free objectives,” inProc. INTERSPEECH, 2019, pp. 4300–4304

    2019

  5. [13]

    End-to-end speaker diarization for an unknown number of speakers with encoder-decoder based attractors,

    S. Horiguchi, Y . Fujita, S. Watanabe, Y . Xue, and K. Naga- matsu, “End-to-end speaker diarization for an unknown number of speakers with encoder-decoder based attractors,” inProc. IN- TERSPEECH, 2020, pp. 269–273

    2020

  6. [14]

    Encoder-decoder based attractors for end-to-end neural diariza- tion,

    S. Horiguchi, Y . Fujita, S. Watanabe, Y . Xue, and P. Garc ´ıa, “Encoder-decoder based attractors for end-to-end neural diariza- tion,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 1493–1507, 2022

    2022

  7. [15]

    Deep clus- tering: Discriminative embeddings for segmentation and separa- tion,

    J. R. Hershey, Z. Chen, J. Le Roux, and S. Watanabe, “Deep clus- tering: Discriminative embeddings for segmentation and separa- tion,” inProc. ICASSP, 2016, pp. 31–35

    2016

  8. [16]

    Target- speaker voice activity detection: A novel approach for multi- speaker diarization in a dinner party scenario,

    I. Medennikov, M. Korenevsky, T. Prisyach, Y . Khokhlov, M. Ko- renevskaya, I. Sorokin, T. Timofeeva, A. Mitrofanov, A. An- drusenko, I. Podluzhny, A. Laptev, and A. Romanenko, “Target- speaker voice activity detection: A novel approach for multi- speaker diarization in a dinn...

    2020

  9. [17]

    Similarity measure- ment of segment-level speaker embeddings in speaker diariza- tion,

    W. Wang, Q. Lin, D. Cai, and M. Li, “Similarity measure- ment of segment-level speaker embeddings in speaker diariza- tion,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 2645–2658, 2022

    2022

  10. [18]

    Target- speaker voice activity detection via sequence-to-sequence predic- tion,

    M. Cheng, W. Wang, Y . Zhang, X. Qin, and M. Li, “Target- speaker voice activity detection via sequence-to-sequence predic- tion,” inProc. ICASSP, 2023, pp. 1–5

    2023

  11. [19]

    Ustc-nelslip system description for dihard-iii chal- lenge,

    Y . Wang, M. He, S. Niu, L. Sun, T. Gao, X. Fang, J. Pan, J. Du, and C.-H. Lee, “Ustc-nelslip system description for dihard-iii chal- lenge,”arXiv preprint arXiv:2103.10661, 2021

    2021 arXiv

  12. [20]

    The dku-dukeece-lenovo system for the diarization task of the 2021 voxceleb speaker recognition challenge,

    W. Wang, D. Cai, Q. Lin, L. Yang, J. Wang, J. Wang, and M. Li, “The dku-dukeece-lenovo system for the diarization task of the 2021 voxceleb speaker recognition challenge,”arXiv preprint arXiv:2109.02002, 2021

    2021 arXiv

  13. [21]

    The dku-dukeece diarization system for the voxceleb speaker recogni- tion challenge 2022,

    W. Wang, X. Qin, M. Cheng, Y . Zhang, K. Wang, and M. Li, “The dku-dukeece diarization system for the voxceleb speaker recogni- tion challenge 2022,”arXiv preprint arXiv:2210.01677, 2022

    2022 arXiv

  14. [22]

    The dku-msxf diarization system for the voxceleb speaker recog- nition challenge 2023,

    M. Cheng, W. Wang, X. Qin, Y . Lin, N. Jiang, G. Zhao, and M. Li, “The dku-msxf diarization system for the voxceleb speaker recog- nition challenge 2023,” inProc. NCMMSC, 2024, pp. 330–337

    2023

  15. [23]

    Sequence-to-sequence neural di- arization with automatic speaker detection and representation,

    M. Cheng, Y . Lin, and M. Li, “Sequence-to-sequence neural di- arization with automatic speaker detection and representation,” arXiv preprint arXiv:2411.13849, 2024

    2024 arXiv

  16. [24]

    Diarization is hard: Some experiences and lessons learned for the jhu team in the inaugural dihard challenge,

    G. Sell, D. Snyder, A. McCree, D. Garcia-Romero, J. Villalba, M. Maciejewski, V . Manohar, N. Dehak, D. Povey, S. Watanabe, and S. Khudanpur, “Diarization is hard: Some experiences and lessons learned for the jhu team in the inaugural dihard challenge,” inProc. INTERSPEECH, 20...

    2018

  17. [25]

    The second dihard diarization challenge: Dataset, task, and baselines,

    N. Ryant, K. Church, C. Cieri, A. Cristia, J. Du, S. Ganapathy, and M. Liberman, “The second dihard diarization challenge: Dataset, task, and baselines,” inProc. INTERSPEECH, 2019, pp. 978–982

    2019

  18. [26]

    The third di- hard diarization challenge,

    N. Ryant, P. Singh, V . Krishnamohan, R. Varma, K. Church, C. Cieri, J. Du, S. Ganapathy, and M. Liberman, “The third di- hard diarization challenge,” inProc. INTERSPEECH, 2021, pp. 3570–3574

    2021

  19. [27]

    The multimodal information based speech processing (misp) 2025 challenge: Audio-visual di- arization and recognition,

    M. Gao, S. Wu, H. Chen, J. Du, C.-H. Lee, S. Watanabe, J. Chen, S. S. Marco, and O. Scharenborg, “The multimodal information based speech processing (misp) 2025 challenge: Audio-visual di- arization and recognition,” 2025

    2025

  20. [28]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProc. CVPR, 2016

    2016

  21. [29]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. u. Kaiser, and I. Polosukhin, “Attention is all you need,” inProc. NeurIPS, vol. 30, 2017

    2017

  22. [30]

    Conformer: Convolution-augmented transformer for speech recognition,

    A. Gulati, J. Qin, C.-C. Chiu, N. Parmar, Y . Zhang, J. Yu, W. Han, S. Wang, Z. Zhang, Y . Wu, and R. Pang, “Conformer: Convolution-augmented transformer for speech recognition,” in Proc. INTERSPEECH, 2020, pp. 5036–5040

    2020

  23. [31]

    Arcface: Additive angular margin loss for deep face recognition,

    J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “Arcface: Additive angular margin loss for deep face recognition,” inProc. CVPR, 2019

    2019

  24. [32]

    V oxceleb2: Deep speaker recognition,

    J. S. Chung, A. Nagrani, and A. Zisserman, “V oxceleb2: Deep speaker recognition,” inProc. INTERSPEECH, 2018, pp. 1086– 1090

    2018

  25. [33]

    V oxblink2: A 100k+ speaker recognition corpus and the open- set speaker-identification benchmark,

    Y . Lin, M. Cheng, F. Zhang, Y . Gao, S. Zhang, and M. Li, “V oxblink2: A 100k+ speaker recognition corpus and the open- set speaker-identification benchmark,” inProc. INTERSPEECH, 2024, pp. 4263–4267

    2024

  26. [34]

    Kespeech: An open source speech dataset of mandarin and its eight subdialects,

    Z. Tang, D. Wang, Y . Xu, J. Sun, X. Lei, S. Zhao, C. Wen, X. Tan, C. Xie, S. Zhouet al., “Kespeech: An open source speech dataset of mandarin and its eight subdialects,” inThirty-fifth Conference on Neural Information Processing Systems Datasets and Bench- marks Track (Round 2), 2021

    2021

  27. [35]

    3d-speaker: A large-scale multi-device, multi-distance, and multi-dialect cor- pus for speech representation disentanglement,

    S. Zheng, L. Cheng, Y . Chen, H. Wang, and Q. Chen, “3d-speaker: A large-scale multi-device, multi-distance, and multi-dialect cor- pus for speech representation disentanglement,”arXiv preprint arXiv:2306.15354, 2023

    2023 arXiv

  28. [36]

    Multi-input multi-output target-speaker voice activity detection for unified, flexible, and robust audio- visual speaker diarization,

    M. Cheng and M. Li, “Multi-input multi-output target-speaker voice activity detection for unified, flexible, and robust audio- visual speaker diarization,”arXiv preprint arXiv:2401.08052, 2024

    2024 arXiv

  29. [37]

    MISP-Meeting: A real-world dataset with multimodal cues for long-form meeting transcription and summarization,

    H. Chen, C.-H. H. Yang, J.-C. Gu, S. M. Siniscalchi, and J. Du, “MISP-Meeting: A real-world dataset with multimodal cues for long-form meeting transcription and summarization,” inProc. ACL, 2025, pp. 1–14

    2025

  30. [38]

    Musan: A music, speech, and noise corpus,

    D. Snyder, G. Chen, and D. Povey, “Musan: A music, speech, and noise corpus,”arXiv preprint arXiv:1510.08484, 2015

    2015 arXiv

  31. [39]

    A study on data augmentation of reverberant speech for robust speech recognition,

    T. Ko, V . Peddinti, D. Povey, M. L. Seltzer, and S. Khudanpur, “A study on data augmentation of reverberant speech for robust speech recognition,” inProc. ICASSP, 2017, pp. 5220–5224

    2017

  32. [40]

    Decoupled weight decay regularization,

    I. Loshchilov, “Decoupled weight decay regularization,”arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  33. [41]

    X-vectors: Robust dnn embeddings for speaker recognition,

    D. Snyder, D. Garcia-Romero, G. Sell, D. Povey, and S. Khudan- pur, “X-vectors: Robust dnn embeddings for speaker recognition,” inProc. ICASSP, 2018, pp. 5329–5333

    2018

  34. [42]

    The idlab voxsrc-20 submission: Large margin fine-tuning and quality- aware score calibration in dnn based speaker verification,

    J. Thienpondt, B. Desplanques, and K. Demuynck, “The idlab voxsrc-20 submission: Large margin fine-tuning and quality- aware score calibration in dnn based speaker verification,” inProc. ICASSP, 2021, pp. 5814–5818

    2021

  35. [43]

    V oxceleb: A large- scale speaker identification dataset,

    A. Nagrani, J. S. Chung, and A. Zisserman, “V oxceleb: A large- scale speaker identification dataset,” inProc. INTERSPEECH, 2017, pp. 2616–2620

    2017

Pith tools

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