Pith. sign in

REVIEW 4 major objections 6 minor 33 references

Stereo sound event localization and detection based on PSELDnet pretraining and BiMamba sequence modeling

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read By swapping the Conformer decoder for a bidirectional Mamba with asymmetric convolutions, this paper reports the best stereo SELD scores on the DCASE2025 Task 3 development set, using roughly a third of the Conformer's parameters.

desk verdict Plausible engineering tweak; the 'significantly better' claim is undercut by uncontrolled hyperparameters. read the letter →

arxiv 2506.13455 v1 pith:Z3NSOYJQ submitted 2025-06-16 eess.AS cs.SD

Wenmiao Gao , Yang Xiao This is my paper · ORCID
classification eess.AScs.SD
keywords stereosoundeventlocalizationanddetectionSELDBiMambaselectivestatespacemodelasymmetricconvolutionPSELDnetDCASE2025Task3pseudo-FOAconversion
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 proposes a stereo sound event localization and detection (SELD) system that replaces the Conformer decoder in a pretrained PSELDnet with a bidirectional Mamba (BiMamba) sequence model augmented by asymmetric convolutions. Its central claim is that this BiMambaAC model outperforms both the original Conformer-based PSELDnet and the CRNN baseline on the DCASE2025 Task 3 development set, while using fewer parameters and fewer MACs than the Conformer. If true, it shows that selective state-space models are a computationally cheaper alternative to attention for SELD on ordinary stereo audio, and that pretrained FOA-oriented models can be transferred to stereo via pseudo-FOA reconstruction.

What carries the argument

The central object is the BiMamba module: a bidirectional selective state-space model that processes a sequence forward and backward and fuses the hidden states, combined with asymmetric convolutions that alternate between the time and frequency dimensions. This decoder replaces the Conformer in the CNN14-Conformer PSELDnet. A second load-bearing step is the conversion of stereo left-right signals into pseudo first-order-ambisonic (FOA) channels, $W = (L+R)/2$ and $Y = (L-R)/2$ with $X = Z = 0$, which lets the pretrained FOA-based PSELDnet be fine-tuned on stereo data. The mechanism is meant to capture both directions of temporal context and to decouple time-frequency features, handling overlapping events better than the CRNN baseline.

What would settle it

Retrain the Conformer and BiMambaAC models under identical hyperparameters (same seed, learning rate, batch size, weight decay) or a small shared hyperparameter grid, and check whether BiMambaAC keeps its 1.4-point F1 lead over Conformer; a Conformer configuration that matches or exceeds 39.6% F1 would falsify the paper's central claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the CNN14-BiMamba hybrid — a pretrained CNN14 encoder with a BiMamba decoder plus asymmetric convolution — reaches a location-dependent F1 of 39.6%, a DOA error of 15.8°, and a relative distance error of 33% on the DCASE2025 Task 3 development set, with 76M parameters and 4.63G MACs. This beats the Conformer decoder (38.2% F1, 15.9° DOAE, 33% RDE, 210M parameters) and the CRNN baseline (22.8%, 24.5°, 41%, 0.7M), and is the best configuration the authors tested. The paper also claims that the asymmetric convolutions are responsible for part of the gain, since BiMamba without them scores 36.2% F1 with 178M parameters.

Load-bearing premise

The comparison assumes that the single training run per model, with different learning rates, batch sizes, and weight decay for each architecture, is a fair way to rank architectures; if those hyperparameters, rather than the BiMamba design, explain the score gap, the headline claim collapses.

Editorial extensions

If this is right

  • If the result holds, BiMamba can serve as a drop-in Conformer replacement in pretrained SELD networks, cutting parameters by roughly a factor of three without losing accuracy.
  • The pseudo-FOA reconstruction trick suggests that large FOA-pretrained models can be adapted to stereo-only consumer content rather than requiring new stereo pretraining.
  • Asymmetric convolutions appear to be a cheap way to shave parameter count and improve discrimination of overlapping events, which may transfer to other time-frequency audio tasks.
  • The reported 39.6% F1 on the development set provides a concrete reference point that future stereo SELD systems can compare against.

Reading between the lines

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

  • The same decoder swap could transfer to other audio tasks that currently pair a pretrained encoder with an attention decoder, though the paper does not test this.
  • Because the stereo-to-pseudo-FOA reconstruction discards the X and Z ambisonic components, an open question is whether a binaural or stereo-native pretraining objective would preserve spatial information that the current pipeline loses.
  • The azimuth folding to [-90°, 90°] exploits the fixed frontal perspective of this year's challenge; an obvious extension is testing whether the architecture handles full 360° azimuth when labels are not folded.
  • The channel-swap augmentation and pseudo-FOA trick could apply to other fixed-perspective SELD settings, such as smart displays or robotics, without collecting new spatial audio data.
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 / 6 minor

Summary. The paper proposes a stereo sound event localization and detection (SELD) system for the DCASE2025 Task 3 audio-only track. The authors convert stereo input into a pseudo-FOA representation, fine-tune pretrained PSELDnet models, and replace the Conformer decoder in the CNN14-Conformer architecture with a bidirectional Mamba (BiMamba) module combined with asymmetric convolutions, calling the resulting model BiMambaAC. On the development set, BiMambaAC is reported to achieve an F1 score of 39.6%, DOAE of 15.8 degrees, and RDE of 33% with 76M parameters and 4.63G MACs, compared with 38.2%, 15.9 degrees, 33%, 210M parameters, and 4.69G MACs for the Conformer variant, and 22.8%, 24.5 degrees, and 41% for the official baseline. The paper concludes that BiMambaAC is significantly better than the Conformer decoder while reducing computational complexity.

Significance. If the reported comparison were robust, the paper would provide useful evidence that state-space decoders can replace Transformer decoders in a pretrained SELD pipeline with fewer parameters and comparable or better accuracy. The use of public challenge data, a standardized evaluation protocol, and reporting of parameter counts and MACs are strengths. However, the central architecture comparison currently rests on a single run per model with deliberately different training hyperparameters, so the headline claim of significant improvement is not yet supported. The incremental nature of the technical change and the absence of statistical significance testing limit the contribution to a suggestive empirical result rather than a definitive architecture comparison.

major comments (4)
  1. [Section 4 and Table 1] The central comparative claim that BiMambaAC outperforms the Conformer decoder is supported only by a single run per model with uncontrolled training hyperparameters. Section 4 reports different learning rates, weight decays, and batch sizes for each model (BiMambaAC: lr 3e-5, wd 5e-6, bs 32; Conformer: lr 1e-4, wd 1e-4, bs 256). With a fixed 120-epoch budget, the batch size difference alone gives BiMambaAC roughly eight times more parameter updates per epoch, so the 1.4-point F1 gap (39.6% vs. 38.2%) and 0.1-degree DOAE gap (15.8 degrees vs. 15.9 degrees) in Table 1 cannot be attributed to the BiMamba architecture. Please run multiple seeds and either match the training recipe across models or perform a hyperparameter sweep; without this, the abstract's 'significantly better' is unsupported.
  2. [Section 5 and Conclusion] The conclusion that BiMambaAC achieves 'the best overall performance across all metrics' is not consistent with Table 1: HTS-AT has a lower RDE (30% vs. 33%), and BiMambaAC's RDE is identical to that of Conformer and BiMamba. The advantage over Conformer is limited to F1 and DOAE, with DOAE differing by only 0.1 degree. Please revise the abstract and conclusion to state precisely which metrics are improved and by how much.
  3. [Section 5, Table 1] The 'reduced computational complexity' claim is overstated as written. Table 1 shows BiMambaAC at 4.63G MACs versus 4.69G for Conformer, a 1.3% reduction, while the parameter count drops from 210M to 76M. If the claim refers to parameter count, it should say so explicitly; if it refers to MACs, the difference is marginal and should not be described as a major advantage, especially since HTS-AT already operates at 2.88G MACs.
  4. [Section 3.3 and Section 4] The paper does not state whether the BiMamba module and the asymmetric convolution layers are initialized from pretrained weights or randomly initialized, whereas the Conformer decoder inherits PSELDnet pretrained weights. If the BiMamba components are randomly initialized, the comparison mixes initialization and architecture effects. Please disclose the initialization scheme and, ideally, train the Conformer baseline from the same random initialization to isolate the contribution of the architecture change.
minor comments (6)
  1. [Section 5, paragraph 2] The Conformer DOAE is reported as 16.6 degrees in the text but 15.9 degrees in Table 1; please correct and unify the value.
  2. [Section 3.1, Eq. (4)] Setting X(n)=Z(n)=0 discards all non-horizontal spatial information; please state explicitly that the pseudo-FOA representation is azimuth-only and discuss the implications for the evaluation, which already excludes elevation.
  3. [Section 4] The sentence 'the random seed is set to be 42' implies a single seed; please state the number of seeds used per configuration, consistent with the request for multiple runs in the major comments.
  4. [Throughout] The capitalization of PSELDnet/PSELDNet is inconsistent; please standardize the spelling.
  5. [Table 1] The baseline row is crowded ('Baseline0.7M57M'); please format the table for readability.
  6. [References] Several references are listed as arXiv preprints; where peer-reviewed versions now exist, please update the citation details.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical comparison against an external benchmark, and the pretrained models are inputs rather than conclusions.

full rationale

The central claim is that fine-tuning PSELDnet-based models with a BiMamba decoder and asymmetric convolutions improves SELD performance on the DCASE2025 Task 3 development set relative to the challenge baseline and the original Conformer-decoder PSELDnet. This is an experimental evaluation, not a derivation. The pretrained PSELDnet weights are adopted from prior work and are used as starting points for fine-tuning, so citing them is a legitimate use of external building blocks, not a self-citation that supplies the conclusion. No quantity called a prediction is obtained by inverting a fitted parameter or by re-using the target metric as an input. The reported F1, DOAE, RDE, parameter counts, and MACs are measured on the development set's test split according to the challenge protocol. The use of different batch sizes, learning rates, and weight decay values across models is a potential experimental confound and a fairness concern, but it is not circular reasoning; it does not make any result true by definition. The paper's equations (SSM discretization, pseudo-FOA conversion, and activation functions) are standard technical components and none of them is defined in terms of the reported results. The only self-citations are incidental references to the authors' prior work on sound event detection and localization, and none of those citations is load-bearing for the paper's main comparison. Overall, no circular step is identifiable, so a score of 0 is appropriate.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on standard Mamba equations, an assumption about the pseudo-FOA reconstruction, and an assumption about ACS augmentation. No new physical entities or fitted constants are introduced; the main free parameters are the per-model training hyperparameters, which are not controlled across architectures.

free parameters (1)
  • Per-model training hyperparameters = BiMambaAC: lr 3e-5, weight decay 5e-6, batch size 32; Conformer: lr 1e-4, wd 1e-4, bs 256; etc.
    Each model is trained with a different learning rate, weight decay, and batch size (Section 4). These choices are hand-tuned and could affect the relative performance, so they act as free parameters in the comparison.
assumptions (3)
  • domain assumption Stereo-to-pseudo-FOA conversion (W=(L+R)/2, Y=(L-R)/2, X=0, Z=0) retains sufficient spatial information for SELD.
    Used in Section 3.1, Eq. (4). If X and Z carry meaningful spatial cues, discarding them could limit absolute performance, though it affects all compared models equally.
  • standard math Mamba state-space model discretization (Eqs. 1-3) is a valid basis for sequence modeling in this audio task.
    The paper restates the standard S4/Mamba update; this is standard background, not an ad hoc assumption.
  • domain assumption Audio Channel Swapping (ACS) augmentation produces valid DOA label flips under the front-back ambiguity folding.
    Section 3.2; the azimuth folding to [-90, 90] and ACS rely on geometric symmetry of the stereo setup.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Stereo sound event localization and detection based on PSELDnet pretraining and BiMamba sequence modeling." pith.science (2026). https://pith.science/paper/Z3NSOYJQ

@misc{pith2026250613455,
  author       = {Pith},
  title        = {Pith review of: Stereo sound event localization and detection based on PSELDnet pretraining and BiMamba sequence modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z3NSOYJQ}},
  note         = {Machine review of arXiv:2506.13455}
}
read the original abstract

Pre-training methods have achieved significant performance improvements in sound event localization and detection (SELD) tasks, but existing Transformer-based models suffer from high computational complexity. In this work, we propose a stereo sound event localization and detection system based on pre-trained PSELDnet and bidirectional Mamba sequence modeling. We replace the Conformer module with a BiMamba module and introduce asymmetric convolutions to more effectively model the spatiotemporal relationships between time and frequency dimensions. Experimental results demonstrate that the proposed method achieves significantly better performance than the baseline and the original PSELDnet with Conformer decoder architecture on the DCASE2025 Task 3 development dataset, while also reducing computational complexity. These findings highlight the effectiveness of the BiMamba architecture in addressing the challenges of the SELD task.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 29 canonical work pages

  1. [1]

    INTRODUCTION The objective of the sound event localization and detection (SELD) task is to identify sound events from predefined target classes, track their temporal dynamics, and estimate their respective spatial tra- jectories where they are active[1].This technology plays a vital role in various real-world applications, such as robotic auditory sensing...

    work page Pith review arXiv 2019

  2. [2]

    THEORY The Mamba architecture, based on the S4 (Structured State Space Sequence) model [25, 26], effectively combines the advantages of CNN and RNN, allowing the use of CNN’s parallel computation benefits during training and RNN’s temporal modeling capabilities during inference. The introduction of state-selection mechanisms allows the model to selectivel...

    work page

  3. [3]

    PROPOSED METHOD 3.1. Feature Extraction Since PSELDNet operates on FOA-format audio data by concate- nating 4-channel log-mel spectrograms with 3-channel intensity vectors, to meet the input feature requirements of the pre-trained network, we convert the stereo left-right ear signalsL(n)andR(n) back to the FOA componentsW(n)andY(n)according to the ACN/SN3...

    work page 2025

  4. [4]

    The audio is resampled to 24 kHz, and 64 mel filters are used for feature extraction

    TRAINING The proposed system is trained on the development dataset (stereo format) of STARSS23[27]. The audio is resampled to 24 kHz, and 64 mel filters are used for feature extraction. The short-time Fourier transform (STFT) is computed with a hop length of 20 ms (480 samples) and a window length of 40 ms. Each input feature consists of 250 frames, corre...

    work page

  5. [5]

    RESULTS Table 1 presents a quantitative comparison between the baseline system and various pre-trained models on the development set. The metrics reported include the number of model parameters, the location-dependent F1-score (F 20◦, higher is better), the Direction of Arrival Error (DOAE, lower is better), and the Relative Distance Error (RDE, lower is ...

    work page

  6. [6]

    CONCLUSION In this work, we propose a novel approach for stereo sound event lo- calization and detection (SELD) by combining PSELDnet pretrain- ing with BiMamba sequence modeling. Our experimental results demonstrate several key findings:First, the integration of pre-trained models significantly improves SELD performance compared to the Detection and Clas...

    work page 2025

  7. [7]

    Sound Event Localization and Detection of Overlapping Sources Us- ing Convolutional Recurrent Neural Networks,

    S. Adavanne, A. Politis, J. Nikunen, and T. Virtanen, “Sound Event Localization and Detection of Overlapping Sources Us- ing Convolutional Recurrent Neural Networks,”arXiv, June 2018

    work page 2018

  8. [8]

    Context- dependent sound event detection,

    T. Heittola, A. Mesaros, A. Eronen, and T. Virtanen, “Context- dependent sound event detection,”J. AUDIO SPEECH MU- SIC PROC., vol. 2013, no. 1, pp. 1–13, Dec. 2013

    work page 2013

Show all 33 references
  1. [9]

    Acoustic event detection in real life recordings,

    A. Mesaros, T. Heittola, A. Eronen, and T. Virtanen, “Acoustic event detection in real life recordings,” in2010 18th European Signal Processing Conference, 2010, pp. 1267–1271

    2010

  2. [10]

    FMSG-JLESS sub- mission for DCASE 2024 task4 on sound event detection with heterogeneous training dataset and potentially missing labels,

    Y . Xiao, H. Yin, J. Bai, and R. K. Das, “FMSG-JLESS sub- mission for DCASE 2024 task4 on sound event detection with heterogeneous training dataset and potentially missing labels,” DCASE 2024 Challenge, Tech. Rep., 2024

    2024

  3. [11]

    WildDESED: an LLM-powered dataset for wild domestic environment sound event detection system,

    Y . Xiao and R. K. Das, “WildDESED: an LLM-powered dataset for wild domestic environment sound event detection system,” inProc. Workshop on Detection and Classification of Acoustic Scenes and Events (DCASE), 2024, pp. 196–200

    2024

  4. [12]

    UCIL: An Unsupervised Class Incremental Learn- ing Approach for Sound Event Detection,

    ——, “UCIL: An Unsupervised Class Incremental Learn- ing Approach for Sound Event Detection,”arXiv:2407.03657, 2024

    2024 arXiv

  5. [13]

    Mixstyle based do- main generalization for sound event detection with heteroge- neous training data,

    Y . Xiao, H. Yin, J. Bai, and R. K. Das, “Mixstyle based do- main generalization for sound event detection with heteroge- neous training data,”arXiv preprint arXiv:2407.03654, 2024

    2024 arXiv

  6. [14]

    Multiple emitter location and signal parameter estimation,

    R. Schmidt, “Multiple emitter location and signal parameter estimation,”IEEE Trans. Antennas Propag., vol. 34, no. 3, pp. 276–280, Mar. 1986

    1986

  7. [15]

    Where’s that voice coming? continual learning for sound source localization,

    Y . Xiao and R. K. Das, “Where’s that voice coming? continual learning for sound source localization,” 2025. [Online]. Available: https://arxiv.org/abs/2407.03661

    2025 arXiv

  8. [16]

    An Improved Event-Independent Network for Poly- phonic Sound Event Localization and Detection,

    Y . Cao, T. Iqbal, Q. Kong, F. An, W. Wang, and M. D. Plumb- ley, “An Improved Event-Independent Network for Poly- phonic Sound Event Localization and Detection,” inICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2020...

    2021

  9. [17]

    ACCDOA: Activity-Coupled Cartesian Direc- tion of Arrival Representation for Sound Event Localization and Detection,

    K. Shimada, Y . Koyama, N. Takahashi, S. Takahashi, and Y . Mitsufuji, “ACCDOA: Activity-Coupled Cartesian Direc- tion of Arrival Representation for Sound Event Localization and Detection,”arXiv, Oct. 2020

    2020

  10. [18]

    Multi-ACCDOA: Localizing and Detecting Overlapping Sounds from the Same Class with Auxiliary Duplicating Permutation Invariant Training,

    K. Shimada, Y . Koyama, S. Takahashi, N. Takahashi, E. Tsunoo, and Y . Mitsufuji, “Multi-ACCDOA: Localizing and Detecting Overlapping Sounds from the Same Class with Auxiliary Duplicating Permutation Invariant Training,”arXiv, Oct. 2021

    2021

  11. [19]

    PSELDNets: Pre-trained Neural Net- works on Large-scale Synthetic Datasets for Sound Event Lo- calization and Detection,

    J. Hu, Y . Cao, M. Wu, F. Kang, F. Yang, W. Wang, M. D. Plumbley, and J. Yang, “PSELDNets: Pre-trained Neural Net- works on Large-scale Synthetic Datasets for Sound Event Lo- calization and Detection,”arXiv, Nov. 2024

    2024

  12. [20]

    PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition,

    Q. Kong, Y . Cao, T. Iqbal, Y . Wang, W. Wang, and M. D. Plumbley, “PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition,”arXiv, Dec. 2019

    2019

  13. [21]

    Ef- ficient Training of Audio Transformers with Patchout,

    K. Koutini, J. Schl ¨uter, H. Eghbal-zadeh, and G. Widmer, “Ef- ficient Training of Audio Transformers with Patchout,”arXiv, Oct. 2021

    2021

  14. [22]

    HTS-AT: A Hierarchical Token-Semantic Audio Transformer for Sound Classification and Detection,

    K. Chen, X. Du, B. Zhu, Z. Ma, T. Berg-Kirkpatrick, and S. Dubnov, “HTS-AT: A Hierarchical Token-Semantic Audio Transformer for Sound Classification and Detection,”arXiv, Feb. 2022

    2022

  15. [23]

    STARSS23: An Audio- Visual Dataset of Spatial Recordings of Real Scenes with Spa- tiotemporal Annotations of Sound Events,

    K. Shimada, A. Politis, P. Sudarsanam, D. Krause, K. Uchida, S. Adavanne, A. Hakala, Y . Koyama, N. Takahashi, S. Taka- hashi, T. Virtanen, and Y . Mitsufuji, “STARSS23: An Audio- Visual Dataset of Spatial Recordings of Real Scenes with Spa- tiotemporal Annotations of Sound Ev...

    2023

  16. [24]

    Mamba: Linear-Time Sequence Modeling with Selective State Spaces,

    A. Gu and T. Dao, “Mamba: Linear-Time Sequence Modeling with Selective State Spaces,”arXiv, Dec. 2023

    2023

  17. [25]

    SPMamba: State-space model is all you need in speech separation,

    K. Li, G. Chen, R. Yang, and X. Hu, “SPMamba: State-space model is all you need in speech separation,”arXiv, Apr. 2024

    2024

  18. [26]

    Mamba in Speech: To- wards an Alternative to Self-Attention,

    X. Zhang, Q. Zhang, H. Liu, T. Xiao, X. Qian, B. Ahmed, E. Ambikairajah, H. Li, and J. Epps, “Mamba in Speech: To- wards an Alternative to Self-Attention,”arXiv, May 2024

    2024

  19. [27]

    SELD-Mamba: Selective State-Space Model for Sound Event Localization and Detection with Source Distance Es- timation,

    D. Mu, Z. Zhang, H. Yue, Z. Wang, J. Tang, and J. Yin, “SELD-Mamba: Selective State-Space Model for Sound Event Localization and Detection with Source Distance Es- timation,”arXiv, Aug. 2024

    2024

  20. [28]

    Nd-BiMamba2: A Unified Bidirectional Architecture for Multi-Dimensional Data Processing,

    H. Liu, “Nd-BiMamba2: A Unified Bidirectional Architecture for Multi-Dimensional Data Processing,”arXiv, Nov. 2024

    2024

  21. [29]

    Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality,

    T. Dao and A. Gu, “Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality,”arXiv, May 2024

    2024

  22. [30]

    A Four-Stage Data Augmentation Approach to ResNet- Conformer Based Acoustic Modeling for Sound Event Local- ization and Detection,

    Q. Wang, J. Du, H.-X. Wu, J. Pan, F. Ma, and C.-H. Lee, “A Four-Stage Data Augmentation Approach to ResNet- Conformer Based Acoustic Modeling for Sound Event Local- ization and Detection,”arXiv, Jan. 2021

    2021

  23. [31]

    Efficiently Modeling Long Se- quences with Structured State Spaces,

    A. Gu, K. Goel, and C. R ´e, “Efficiently Modeling Long Se- quences with Structured State Spaces,”arXiv, Oct. 2021

    2021

  24. [32]

    Tf-mamba: A time-frequency network for sound source localization,

    Y . Xiao and R. K. Das, “Tf-mamba: A time-frequency network for sound source localization,” 2025. [Online]. Available: https://arxiv.org/abs/2409.05034

    2025 arXiv

  25. [33]

    DCASE2025 Task3 Stereo SELD Dataset,

    K. Shimada, A. Politis, I. R. Roman, P. Sudarsanam, D. D ´ıaz- Guerra Aparicio, R. Pandey, K. Uchida, Y . Koyama, N. Taka- hashi, T. Shibuya, S. Takahashi, T. Virtanen, and Y . Mitsufuji, “DCASE2025 Task3 Stereo SELD Dataset,”Zenodo, Apr. 2025

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.