Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Overlap-Adaptive Hybrid Speaker Diarization and ASR-Aware Observation Addition for MISP 2025 Challenge

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 ↗

Pith's one-line read Tuning the mix of noisy, separated, and GSS audio with a module supervised by character error rate beats the baselines, and pairing it with overlap-adaptive diarization won two MISP 2025 tracks (9.48% CER, 11.56% cpCER).

desk verdict A credible MISP 2025 system report with two modest new ideas; the diarization hybrid is well supported, but the ASR-aware OA gain is underdetermined by the ablations, so treat the headline numbers as system-level. read the letter →

arxiv 2505.22013 v1 pith:AMBIQ7ZI submitted 2025-05-28 cs.SD eess.AS

Shangkun Huang , Yuxuan Du , Jingwen Yang , Dejun Zhang , Xupeng Jia , Jing Deng , Jintao Kang , Rong Zheng This is my paper
classification cs.SDeess.AS
keywords speakerdiarizationautomaticspeechrecognitionobservationadditionguidedsourceseparationhybridoverlappingMISP2025challengeMossformer2
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 is a challenge-report paper describing a meeting transcription system that won first place in two tracks of the MISP 2025 Challenge. The authors' central claim is that far-field meeting speech is recognized best when the front end is not a single enhancement method but a weighted mix of noisy audio, Mossformer2-separated audio, and guided-source-separated audio, with the mixture weights predicted per utterance by a neural module trained against the recognizer's own character error rate (CER). For diarization, they claim that no single pipeline wins across all meetings: a WavLM-based end-to-end segmentation model handles dense overlapping speech, a traditional VBx clustering pipeline handles sparse-overlap conditions, and a measured overlap ratio decides per meeting which to run. On the evaluation set the combined system reports 8.88% diarization error rate, 9.48% CER, and 11.56% concatenated cpCER. A sympathetic reader would care because the central mechanism, supervising front-end fusion with the final recognition metric, is simple to describe and portable to other enhancement and recognition stacks.

What carries the argument

Two mechanisms carry the argument. The first is overlap-adaptive hybrid diarization: the WavLM-Large end-to-end segmentation model supplies speaker-change detection, VAD, and an estimated overlap fraction per meeting, and that fraction is the switch that sends meetings with under 1% overlapping speech to the traditional pipeline (dereverberation with scaled mixing, SimAM-ResNet100 embeddings, VBx clustering) and the rest to the end-to-end path, the chosen output being fused across eight channels with Dover-lap. The second is the ASR-aware observation addition bridging module: a Conformer encoder with convolutional and attentive statistical pooling compresses the three input signals into sentence-level features, and a dense layer maps those features to logits normalized by sigmoid into the coefficients $\omega_1, \omega_2, \omega_3$. Its training loss is $L_{\mathrm{OA}} = -\log \sigma\big(\mathrm{cs}(\mathrm{logits}, \sigma(\mathrm{cers}))/\tau\big)$, where cers is the vector of CER values precomputed for roughly 220 coefficient pairs per training sample (values of $\omega_1$ stepped by 0.05, times 10 values of $\omega_2$) by running the full SS/GSS/ASR chain, so the module learns to predict the weights that actually minimize recognition error rather than weights that reconstruct clean speech.

What would settle it

Run the two diarization paths separately and the hybrid switch together on the MISP 2025 evaluation set, and count how often the 1% rule picks the subsystem with the higher DER; if it frequently selects the worse method, the adaptive selection is not carrying the reported gain. Separately, freeze the recognizer and replace the trained bridging module with fixed equal weights $\omega_1=\omega_2=\omega_3=1/3$: if evaluation CER stays near 10.09% instead of rising toward the GSS-only 11.24%, the gain comes from mixing three signals, not from ASR supervision of the coefficients.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the bottleneck in far-field meeting ASR is the mismatch between what an enhancement front end optimizes and what the recognizer needs, and that this mismatch can be closed by making the front-end fusion itself ASR-aware. The ASR-aware observation addition (OA) framework mixes multichannel noisy speech $X_{\text{sum}}$, Mossformer2-separated speech $\hat{y}$, and GSS-separated speech $X_{\text{gss}}$ as $\omega_1 X_{\text{sum}} + \omega_2 \hat{y} + \omega_3 X_{\text{gss}}$ with $\omega_1+\omega_2+\omega_3=1$, where the coefficients come from a Conformer-based bridging module that takes fbank features of the three inputs. That module is trained so that its predicted weight distribution matches, by cosine similarity, the distribution of precomputed CER values obtained by grid-searching coefficient pairs through the full separation-and-recognition pipeline, making CER itself the supervisory signal. The paper reports that this single system reaches 10.09% CER on the evaluation set, versus 11.24% for GSS-only and 10.63% for its MIM-DA front end with the same recognizer, and that the SNR-based OA baseline it reproduced scored worse on the development set (7.11% versus 5.91%). Combined with the overlap-adaptive hybrid diarization, the cascade achieves 11.56% cpCER and first place in the diarization-plus-recognition track.

Load-bearing premise

The load-bearing premise is that the 1% overlap threshold, calibrated on the development set, keeps its predictive meaning on unseen meetings; if the relationship between measured overlap and which diarization method performs better does not transfer, the hybrid system's advantage could vanish.

Editorial extensions

If this is right

  • With the same Paraformer backend, the ASR-aware OA single system cuts evaluation CER from 11.24% (GSS alone) to 10.09%, and ROVER fusion of all system variants reaches 9.48%.
  • Choosing the diarization method by measured overlap proportion lowers evaluation DER from 10.62% (end-to-end alone) and 13.25% (VBx alone) to 9.09%, and to 8.88% after eight-channel Dover-lap fusion.
  • Feeding the hybrid diarization output into the GSS front end and the ASR-aware OA recognizer yields 11.56% cpCER on the combined track, an 86.25% relative reduction from the 84.05% baseline.
  • The hybrid design works because the VBx pipeline handles sparse-overlap meetings while the end-to-end model resolves dense overlap, so per-meeting selection exploits the strength of each subsystem.

Reading between the lines

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

  • The same recipe could generalize beyond three inputs: the bridging loss does not care what the signal paths are, so a recognizer-driven coefficient learner could tune any collection of enhancement streams (beamformers, target-speaker extraction, dereverberation variants) against any downstream metric, not just CER.
  • The hard 1% overlap switch is a candidate for replacement by a learned router that regresses each subsystem's expected error from overlap and acoustic features, removing the development-set calibration and smoothing the boundary between the two diarization paths.
  • Because the authors report that visual features hurt rather than helped, a testable next step is to use the bridging module's predicted coefficients as a per-utterance confidence signal for choosing between audio-only recognition and a vision-conditional pass.
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

4 major / 5 minor

Summary. The paper reports the authors' systems for the MISP 2025 Challenge across three tracks. For speaker diarization, they propose a hybrid system that combines a WavLM-based end-to-end segmentation approach with a traditional VBx multi-module pipeline, selecting between the two per meeting based on a 1% overlapping-speech threshold tuned on the development set. For speech recognition, they propose an 'ASR-aware observation addition' method that fuses the noisy multichannel sum, Mossformer2-separated speech, and GSS-separated speech with coefficients predicted by a Conformer-based bridging module supervised with a cosine-similarity loss over precomputed CER values. They report a DER of 8.88% on Track 1, a CER of 9.48% after ROVER fusion on Track 2, and a cpCER of 11.56% on Track 3, claiming first place in Tracks 2 and 3. The paper is written as a challenge system description with brief experimental details.

Significance. If the reported results are taken at face value, the paper demonstrates a top-ranked system for a competitive multi-modal meeting transcription challenge, which is a meaningful practical achievement. The proposed hybrid diarization selection and the idea of training an observation-addition coefficient predictor directly under ASR (CER) supervision are interesting and potentially useful for downstream meeting transcription systems. The strengths are the held-out evaluation results, the use of publicly available toolkits, and the concrete integration of multiple front-end signals. However, the causal claims about the proposed components are not fully supported by the reported ablations, as detailed in the major comments. The paper is more of a systems description than a rigorous scientific study, and the empirical evidence for the specific contribution of the ASR-aware bridging module is currently underdetermined.

major comments (4)
  1. [Section 4.2.2, Table 2] The claimed benefit of the ASR-aware OA method rests on the comparison between A6 (ASR-Aware OA, Paraformer) at 10.09% eval CER and A4 (MIM-DA, Paraformer) at 10.63%, but this comparison is confounded. 'MIM-DA' is defined in Section 3.2 as a data augmentation recipe, not a front-end, so it is unclear what input signal A4 feeds to the ASR. Without a matched baseline that uses the same three input signals (Xsum, hat-y, Xgss) with fixed or SNR-based coefficients, or at least an eval-set result for A5 (SNR OA), the improvement cannot be attributed to the CER-supervised bridging module. Please add an eval-set result for A5, an oracle or fixed-weight OA baseline, and clarify the input signal used by A2 and A4.
  2. [Section 4.2.3, Table 3] The cpCER reduction from M1 (S1+A3) to M2 (S6+A6) changes both the diarization system and the ASR front-end simultaneously. No cross-ablation (e.g., S1+A6 or S6+A3) is reported, so the individual contribution of the ASR-aware OA method to the combined Track 3 result is not identified. Please report at least one cross-ablation, or explicitly state that the combined gain is not attributed to either component individually.
  3. [Section 2.1, Table 1] The hybrid diarization system selects between the end-to-end segmentation method and the traditional VBx method based on a 1% overlapping-speech threshold chosen from development-set results. The paper does not report how many evaluation meetings were assigned to each branch, nor any sensitivity analysis around the threshold. Since the adaptive selection is one of the two core contributions, the robustness of this threshold to the evaluation set should be demonstrated, or at least a caveat should be added that the threshold is an empirical choice that may not generalize.
  4. [Section 2.2, Eq. (3)] The training objective for the bridging module is not clearly justified. The loss L_OA = -log sigma(cs(logits, sigma(cers))/tau) computes cosine similarity between sigmoid(logits) and sigmoid(cers), where cers is a vector of CER values. Since lower CER is better, the sigmoid of the raw CER values compresses differences and does not directly encode 'closeness to the optimal coefficient'; the text refers to an 'optimal CER distribution' without defining it. Please specify how cers is normalized or transformed before the loss, and provide a derivation or intuition for why this objective pushes the predicted coefficients toward the low-CER region. If the objective is a soft label matching, a more standard formulation (e.g., softmax over negative CER values) would be easier to interpret.
minor comments (5)
  1. [Table 1 and Table 2 captions] The captions say 'presented the DER (%)' and 'presented the CER (%)'; the verb should be 'present' to agree with the plural subject, and the captions would read more naturally as 'DER (%) results' and 'CER (%) results'.
  2. [Section 2.2] The description of the coefficient grid is inconsistent: for omega1, the text gives '1/k + 2' fixed coefficients with step size k in [0, 0.05], while Section 4.1 states the step size was set to 0.05. With k=0.05 and range 0 to 1, there are 21 values, not 22. For omega2, '10 fixed discrete values' with step 0.1 from 0 to 1 is 11 values if the endpoints are included. Please clarify the exact grid and whether omega1+omega2 > 1 pairs are discarded.
  3. [Section 2.2, paragraph after Eq. (3)] The text says 'These WER values will guide the subsequent training of the bridging module', but the paper consistently uses CER (character error rate) elsewhere; this should say 'CER values'.
  4. [Section 4.1] No details are given about the amount of training data used for the bridging module, the ASR model used to precompute the CER vectors, or the computational cost of the grid search. A sentence on these practical aspects would help reproducibility.
  5. [Section 3.2 and Table 2] The term 'front-end' in Table 2 is misleading for the rows labeled 'MIM-DA', since MIM-DA is described as a data augmentation and training-data integration scheme rather than a signal-processing front-end. It would be clearer to label the rows by the actual input signal (e.g., Xsum with MIM-DA-trained ASR, or GSS with MIM-DA-trained ASR) and to specify the input signal in the text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline CER is measured on held-out evaluation data, and the ASR-aware OA coefficient predictor is trained with a standard CER-supervised objective rather than being defined by the reported result.

full rationale

The paper's central claimed results, a CER of 9.48% on Track 2 and a cpCER of 11.56% on Track 3, are taken from held-out challenge evaluation sets; there is no indication that evaluation-set labels were used to train the ASR or the bridging module. The OA coefficient predictor is trained on the MISP2025 training set using Eq. 3, which minimizes a cosine-similarity loss against precomputed CER vectors obtained by grid search over candidate coefficients. This is a conventional supervised objective: the module learns to select coefficients that worked well on training utterances, and its benefit is then assessed on separate dev/eval data. It is not a case of fitting a parameter to a subset of data and then relabeling that same subset as a prediction. Equation 1 defines a weighted fusion, Eq. 2 defines the coefficient predictor, and Eq. 3 defines its training loss; none of these equations reduces to another by construction. The diarization 'overlap decision' threshold of 1% is chosen from development-set results and applied to evaluation meetings; whether this threshold generalizes is an empirical robustness concern, not a circularity. The cited works by the authors (references [10], [20], [21], [26]) are background or challenge-system descriptions and are not load-bearing in the derivation of the final numbers; no same-author uniqueness theorem or ansatz is invoked to force the chosen architecture. The conclusion's admission that video features did not help is a reported experimental limitation, not a circular step. Missing ablations such as a fixed-weight OA baseline or an eval-set result for SNR-based OA could make the causal attribution of the improvement less certain, but an incomplete experimental comparison is a correctness/completeness issue, and the rules explicitly require not treating the absence of consensus or controls as circularity. Accordingly, no specific circular step can be quoted, and the appropriate score is 0.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The system is an engineering combination of existing tools; the only new free parameters are those tuned on the development set or chosen by hand.

free parameters (6)
  • Overlap decision threshold = 1% (0.01)
    Set based on development set results; scenarios with less than 1% overlap are assigned to the traditional VBx method (Section 2.1).
  • Dereverberation mixing weight = 0.7 for enhanced audio, 0.3 for original
    Manually chosen weight for mixing enhanced and original speech in diarization (Section 4.1).
  • VBx clustering parameters = F_a=0.15, F_b=5.5, loopP=0.99
    Default or dev-set tuned parameters for Bayesian HMM clustering (Section 4.1).
  • E2E clustering threshold = 0.65
    Threshold for hierarchical agglomerative clustering in end-to-end diarization (Section 4.1).
  • OA grid step sizes = k=0.05 for omega1, step 0.1 for omega2
    Discretization for precomputing CER vectors in training the bridging module (Section 2.2).
  • Temperature tau = not reported
    Temperature in cosine-similarity loss (Eq. 3); value not given in the paper.
assumptions (4)
  • domain assumption WavLM self-supervised representations are useful for speaker diarization segmentation.
    Relies on [11] and the WavLM model for feature extraction; not re-derived in this paper.
  • domain assumption VBx clustering with x-vectors and PLDA works for meeting diarization.
    Uses pretrained embeddings and VBx from prior work [15,16]; assumes their validity on MISP data.
  • ad hoc to paper The hybrid selection threshold generalizes from development to evaluation set.
    The 1% overlap threshold is chosen based on dev set and assumed to hold for eval; this is a key empirical assumption.
  • ad hoc to paper Mossformer2 and GSS separated signals are complementary with the noisy signal for ASR.
    Heuristic assertion in Section 2.2 that noisy speech preserves details, Mossformer2 handles non-stationary noise, and GSS retains structure; not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Overlap-Adaptive Hybrid Speaker Diarization and ASR-Aware Observation Addition for MISP 2025 Challenge." pith.science (2026). https://pith.science/paper/AMBIQ7ZI

@misc{pith2026250522013,
  author       = {Pith},
  title        = {Pith review of: Overlap-Adaptive Hybrid Speaker Diarization and ASR-Aware Observation Addition for MISP 2025 Challenge},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AMBIQ7ZI}},
  note         = {Machine review of arXiv:2505.22013}
}
read the original abstract

This paper presents the system developed to address the MISP 2025 Challenge. For the diarization system, we proposed a hybrid approach combining a WavLM end-to-end segmentation method with a traditional multi-module clustering technique to adaptively select the appropriate model for handling varying degrees of overlapping speech. For the automatic speech recognition (ASR) system, we proposed an ASR-aware observation addition method that compensates for the performance limitations of Guided Source Separation (GSS) under low signal-to-noise ratio conditions. Finally, we integrated the speaker diarization and ASR systems in a cascaded architecture to address Track 3. Our system achieved character error rates (CER) of 9.48% on Track 2 and concatenated minimum permutation character error rate (cpCER) of 11.56% on Track 3, ultimately securing first place in both tracks and thereby demonstrating the effectiveness of the proposed methods in real-world meeting scenarios.

Figures

Figures reproduced from arXiv: 2505.22013 by the authors.

Figure 1
Figure 1. The overall workflow of the combined diarization and ASR system submitted for MISP2025 challenge. multi-module approach, we employ the segmentation results to obtain the initial VAD results. Additionally, we use an MP￾SENet-based enhancement model for front-end dereverberation processing[13]. Although the speech enhancement model ef￾fectively suppresses noise and reverberation in meetings, it may introduce artifacts… view at source ↗
Figure 2
Figure 2. Our proposed ASR-aware observation-adding system (a) Overall framework (b)Bridging module structure sample in the training set, the CER is calculated after processing with different OA coefficients, resulting in 10 × ( 1 k + 2) sets of CER values. These WER values will guide the subsequent training of the bridging module, as described below: LOA = − log σ  cs(logits, σ(cers)) τ  (3) where logits represents the unn… view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Training-Free Intelligibility-Guided Observation Addition for Noisy ASR

    eess.AS 2026-02 conditional novelty 6.0 of 10

    Mixing noisy and enhanced speech with weights derived from the recognizer's confidence on each signal reduces ASR word error rate without any additional training.

Reference graph

Works this paper leans on

37 extracted references · 25 canonical work pages · cited by 1 Pith paper

  1. [22]

    Bayesian hmm clustering of x-vector sequences (vbx) in speaker diarization: the- ory, 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: the- ory, implementation and analysis on standard tasks,” Computer Speech & Language, vol. 71, p. 101254, 2022

    work page 2022

  2. [24]

    Gpu-accelerated guided source separation for meeting transcription,

    D. Raj, D. Povey, and S. Khudanpur, “Gpu-accelerated guided source separation for meeting transcription,” arXiv preprint arXiv:2212.05271, 2022

    arXiv 2022

  3. [1]

    This challenge provides an open- source, state-of-the-art, large-scale multimodal meeting dataset that encompasses meetings of various sizes and rich topic transitions[2]

    Introduction The MISP 2025 Challenge 1 aims to advance the application and development of multimodal speech processing technolo- gies in meeting contexts[1]. This challenge provides an open- source, state-of-the-art, large-scale multimodal meeting dataset that encompasses meetings of various sizes and rich topic transitions[2]. The competition comprises t...

    work page 2025

  4. [2]

    System Description Figure 1 illustrates the overall workflow of the combined di- arization and ASR system submitted for MISP2025 challenge. 2.1. Speaker Diarization In the diarization task, we employ a hybrid system that inte- grates traditional multi-module methods with end-to-end seg- mentation approaches. Specifically, for the end-to-end segmen- tation...

    work page Pith review arXiv 2025

  5. [3]

    Dataset 3.1. Speaker Diarization For the dereverberation model used in the diarization task, we trained it using near-field data from MISP2025 and generated room impulse responses using the Pyroomacoustics [27] library. Room dimensions were randomly set: length 3–6 meters, width 5–12 meters, and height fixed at 3 meters to simulate a real- istic conferenc...

    work page

  6. [4]

    Experimental Setup For the end-to-end segmentation method, the clustering thresh- old was set to 0.65

    Experiments 4.1. Experimental Setup For the end-to-end segmentation method, the clustering thresh- old was set to 0.65. For the traditional multi-module method, the weight coefficient of the dereverberation model’s output au- dio was set to 0.7, and that of the original audio to 0.3, before performing the mixing operation. In the VBx clustering stage, the...

    work page

  7. [5]

    For the diarization task, we combined traditional multi-module methods with a WavLM-based end- to-end segmentation model

    Conclusion This paper described our diarization and ASR systems for the MISP 2025 Challenge. For the diarization task, we combined traditional multi-module methods with a WavLM-based end- to-end segmentation model. For the ASR task, the proposed ASR-Aware OA method effectively addressed speech enhance- ment challenges in low-SNR environments, achieving a ...

    work page 2025

  8. [6]

    Acknowledgements This work is supported by National Engineering Research Cen- ter of Multi-dimensional Identification and Trusted Authentica- tion Technology (No.IDNERC202404)

    work page

Show all 37 references
  1. [7]

    The multimodal information based speech processing (misp) 2025 challenge: Audio-visual diarization 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 diarization and recognition,” 2025. [Online]. Available: https://arxiv.org/abs/2505.13971

    2025 arXiv

  2. [8]

    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,” inProceed- ings of the 63st Annual Meeting of the Association for Compu- tational Linguistics (Volum...

    2025

  3. [9]

    Multi-stage diarization refinement for the chime-7 dasr scenario,

    C. B. Boeddeker, T. Cord-Landwehr, T. v. Neumann, and R. Haeb- Umbach, “Multi-stage diarization refinement for the chime-7 dasr scenario,” in Proc. CHiME 2023, 2023, pp. 51–56

    2023

  4. [10]

    Ntt multi-speaker asr system for the dasr task of chime-8 challenge,

    N. Kamo, N. Tawara, A. Ando, T. Kano, H. Sato, R. Ikeshita, T. Moriya, S. Horiguchi, K. Matsuura, A. Ogawa et al. , “Ntt multi-speaker asr system for the dasr task of chime-8 challenge,” arXiv preprint arXiv:2409.05554, 2024

    2024 arXiv

  5. [11]

    The whu- alibaba audio-visual speaker diarization system for the misp 2022 challenge,

    M. Cheng, H. Wang, Z. Wang, Q. Fu, and M. Li, “The whu- alibaba audio-visual speaker diarization system for the misp 2022 challenge,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2023, pp. 1–2

    2022

  6. [12]

    Multi-speaker end-to-end multi-modal speaker diarization system for the misp 2022 chal- lenge,

    T. Liu, Z. Chen, Y . Qian, and K. Yu, “Multi-speaker end-to-end multi-modal speaker diarization system for the misp 2022 chal- lenge,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–2

    2022

  7. [13]

    The mul- timodal information based speech processing (misp) 2022 chal- lenge: Audio-visual diarization and recognition,

    Z. Wang, S. Wu, H. Chen, M.-K. He, J. Du, C.-H. Lee, J. Chen, S. Watanabe, S. Siniscalchi, O. Scharenborg et al. , “The mul- timodal information based speech processing (misp) 2022 chal- lenge: Audio-visual diarization and recognition,” in ICASSP 2023-2023 IEEE International C...

    2022

  8. [14]

    The flyspeech audio-visual speaker diarization system for misp challenge 2022,

    L. Zhang, H. Zhao, Y . Li, B. Pang, Y . Wang, H. Wang, W. Rao, Q. Wang, and L. Xie, “The flyspeech audio-visual speaker diarization system for misp challenge 2022,” arXiv preprint arXiv:2307.15400, 2023

    2022 arXiv

  9. [15]

    Ntt speaker diarization system for chime-7: multi-domain, multi-microphone end-to-end and vector clustering diarization,

    N. Tawara, M. Delcroix, A. Ando, and A. Ogawa, “Ntt speaker diarization system for chime-7: multi-domain, multi-microphone end-to-end and vector clustering diarization,” in ICASSP 2024- 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEE...

    2024

  10. [16]

    The fosafer system for the icassp2024 in-car multi-channel automatic speech recognition challenge,

    S. Huang, Y . Du, Y . Wang, J. Deng, and R. Zheng, “The fosafer system for the icassp2024 in-car multi-channel automatic speech recognition challenge,” in2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW). IEEE, 2024, pp. 5–6

    2024

  11. [17]

    Leveraging self-supervised learning for speaker diarization,

    J. Han, F. Landini, J. Rohdin, A. Silnova, M. Diez, and L. Bur- get, “Leveraging self-supervised learning for speaker diarization,” arXiv preprint arXiv:2409.09408, 2024

    2024 arXiv

  12. [18]

    Powerset multi-class cross entropy loss for neural speaker diarization,

    A. Plaquet and H. Bredin, “Powerset multi-class cross entropy loss for neural speaker diarization,” Proc. Interspeech, pp. 3222– 3226, 2023

    2023

  13. [19]

    MP-SENet: A speech enhance- ment model with parallel denoising of magnitude and phase spec- tra,

    Y .-X. Lu, Y . Ai, and Z.-H. Ling, “MP-SENet: A speech enhance- ment model with parallel denoising of magnitude and phase spec- tra,” in Proc. Interspeech, 2023, pp. 3834–3838

    2023

  14. [20]

    Simam: A sim- ple, parameter-free attention module for convolutional neural networks,

    L. Yang, R.-Y . Zhang, L. Li, and X. Xie, “Simam: A sim- ple, parameter-free attention module for convolutional neural networks,” in International conference on machine learning . PMLR, 2021, pp. 11 863–11 874

    2021

  15. [21]

    Wespeaker: A research and production oriented speaker embedding learning toolkit,

    H. Wang, C. Liang, S. Wang, Z. Chen, B. Zhang, X. Xiang, Y . Deng, and Y . Qian, “Wespeaker: A research and production oriented speaker embedding learning toolkit,” in ICASSP 2023- 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 20...

    2023

  16. [23]

    Dover-lap: A method for com- bining overlap-aware diarization outputs,

    D. Raj, L. P. Garcia-Perera, Z. Huang, S. Watanabe, D. Povey, A. Stolcke, and S. Khudanpur, “Dover-lap: A method for com- bining overlap-aware diarization outputs,” in 2021 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2021, pp. 881– 888

    2021

  17. [25]

    An audio-quality-based multi-strategy ap- proach for target speaker extraction in the misp 2023 challenge,

    R. Han, X. Yan, W. Xu, P. Guo, J. Sun, H. Wang, Q. Lu, N. Jiang, and L. Xie, “An audio-quality-based multi-strategy ap- proach for target speaker extraction in the misp 2023 challenge,” arXiv preprint arXiv:2401.03697, 2024

    2023 arXiv

  18. [26]

    The fos- afer system for the chime-8 mmcsg challenge,

    S. Huang, D. Zhang, Y . Wang, J. Deng, and R. Zheng, “The fos- afer system for the chime-8 mmcsg challenge,” in CHiME Work- shop on Speech Processing in Everyday Environments, 2024

    2024

  19. [27]

    Enhanced asr for stuttering speech: Combining adversarial and signal-based data augmentation,

    S. Huang, D. Zhang, J. Deng, and R. Zheng, “Enhanced asr for stuttering speech: Combining adversarial and signal-based data augmentation,” in 2024 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2024, pp. 393–400

    2024

  20. [28]

    Reducing the gap between pretrained speech en- hancement and recognition models using a real speech-trained bridging module,

    Z. Cui, C. Cui, T. Wang, M. He, H. Shi, M. Ge, C. Gong, L. Wang, and J. Dang, “Reducing the gap between pretrained speech en- hancement and recognition models using a real speech-trained bridging module,” arXiv preprint arXiv:2501.02452, 2025

    2025 arXiv

  21. [29]

    Mossformer2: Com- bining transformer and rnn-free recurrent network for enhanced time-domain monaural speech separation,

    S. Zhao, Y . Ma, C. Ni, C. Zhang, H. Wang, T. H. Nguyen, K. Zhou, J. Q. Yip, D. Ng, and B. Ma, “Mossformer2: Com- bining transformer and rnn-free recurrent network for enhanced time-domain monaural speech separation,” in ICASSP 2024-2024 IEEE International Conference on Acoust...

    2024

  22. [30]

    Noise robust speech emotion recognition with signal-to-noise ratio adapting speech en- hancement,

    Y .-W. Chen, J. Hirschberg, and Y . Tsao, “Noise robust speech emotion recognition with signal-to-noise ratio adapting speech en- hancement,” arXiv preprint arXiv:2309.01164, 2023

    2023 arXiv

  23. [31]

    Attentive statis- tics pooling for deep speaker embedding,

    K. Okabe, T. Koshinaka, and K. Shinoda, “Attentive statis- tics pooling for deep speaker embedding,” arXiv preprint arXiv:1803.10963, 2018

    2018 arXiv

  24. [32]

    Integrating self-supervised pre-training with adversarial learning for synthe- sized song detection,

    Y . Wang, Y . Du, D. Zhang, R. Zheng, and J. Deng, “Integrating self-supervised pre-training with adversarial learning for synthe- sized song detection,” in2024 IEEE Spoken Language Technology Workshop (SLT). IEEE, 2024, pp. 795–802

    2024

  25. [33]

    Pyroomacoustics: A python package for audio room simulation and array processing algorithms,

    R. Scheibler, E. Bezzam, and I. Dokmani ´c, “Pyroomacoustics: A python package for audio room simulation and array processing algorithms,” in 2018 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2018, pp. 351– 355

    2018

  26. [34]

    V oxceleb2: Deep speaker recognition,

    J. S. Chung, A. Nagrani, and A. Zisserman, “V oxceleb2: Deep speaker recognition,” arXiv preprint arXiv:1806.05622, 2018

    2018 arXiv

  27. [35]

    Paraformer: Fast and accurate parallel transformer for non-autoregressive end-to- end speech recognition,

    Z. Gao, S. Zhang, I. McLoughlin, and Z. Yan, “Paraformer: Fast and accurate parallel transformer for non-autoregressive end-to- end speech recognition,” arXiv preprint arXiv:2206.08317, 2022

    2022 arXiv

  28. [36]

    Funaudiollm: V oice understanding and gen- eration foundation models for natural interaction between humans and llms,

    K. An, Q. Chen, C. Deng, Z. Du, C. Gao, Z. Gao, Y . Gu, T. He, H. Hu, K. Hu et al., “Funaudiollm: V oice understanding and gen- eration foundation models for natural interaction between humans and llms,” arXiv preprint arXiv:2407.04051, 2024

    2024 arXiv

  29. [37]

    A post-processing system to yield reduced word er- ror rates: Recognizer output voting error reduction (rover),

    J. G. Fiscus, “A post-processing system to yield reduced word er- ror rates: Recognizer output voting error reduction (rover),” in 1997 IEEE Workshop on Automatic Speech Recognition and Un- derstanding Proceedings. IEEE, 1997, pp. 347–354

    1997

Pith tools

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