Pith. sign in

REVIEW 3 major objections 6 minor 35 references

Efficient and Microphone-Fault-Tolerant 3D Sound Source Localization

T0 review · 3 major / 6 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 The paper claims a sparse cross-attention 3D sound localizer that uses frozen audio embeddings and weighted TDOA features beats prior methods on LuViRA music3/speech3 while needing fewer microphones and estimating faulty mic positions.

desk verdict The architecture is promising and the benchmark numbers are plausible, but the stated training loss has no source-coordinate term for the single-source case, so the main results are not supported by the paper as written. read the letter →

arxiv 2505.20961 v1 pith:SZNA6CLU submitted 2025-05-27 cs.SD cs.AIcs.LGeess.AS

Yiyuan Yang , Shitong Xu , Niki Trigoni , Andrew Markham This is my paper
classification cs.SDcs.AIcs.LGeess.AS
keywords soundsourcelocalization3Dsparsecross-attentionmicrophonefaulttolerancemaskedautoencoderTDOAfeaturesadaptivesignalcoherence
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 tries to show that an end-to-end, learning-based 3D sound source localization system can be accurate, cheap, and robust at the same time. It proposes a model that fuses pretrained audio embeddings, microphone coordinates, and weighted time-difference-of-arrival features through sparse cross-attention, and trains it with masked reconstruction plus auxiliary losses. The claimed payoff is that the model beats existing localization methods on the LuViRA music3 and speech3 recordings while using fewer microphones, keeps working when some microphone positions are unknown or wrong, and can even estimate those faulty positions. A sympathetic reader would care because real deployments cannot always guarantee calibrated, dense, fully working arrays.

What carries the argument

The object that carries the argument is a three-stream transformer-style network. The acoustic stream passes each microphone's raw audio through a frozen pretrained encoder (BEATs) to get audio embeddings; the coordinate stream encodes microphone positions with an MLP and extracts TDOA features via NGCC-PHAT, weighted by an adaptive signal coherence metric (ASCM) that down-weights unreliable pairs; the joint stream uses cross-attention to align audio with coordinates, then sparse cross-attention (retaining only the top-T attention weights per query) to fuse TDOA embeddings, a masked autoencoder to reconstruct masked audio and coordinate embeddings, and a position decoder to output source locations. The sparse attention is what keeps the model cheap as the number of microphones grows.

What would settle it

Look at the training code for the K=1 case: if Eq. (8)'s source-localization term is the only localization loss, its gradient with respect to the predicted source position is zero, so the reported single-source MAE cannot be charged to that term. A direct experiment is to train the model on speech3 with and without an added absolute-coordinate term such as $|\hat{r}-r|$ and compare the single-source MAE curves.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that microphone-array localization does not need every microphone to be calibrated and reliable: the model jointly learns a spatial geometry from audio and coordinates, and in the paper's experiments it reports the best single-source MAE on LuViRA music3/speech3 (13.9 cm and 21.3 cm, with 96.8% and 88.4% within 30 cm), cuts the required array size (28.7 cm MAE with seven microphones, close to wav2pos's 28.4 cm with nine), localizes faulty microphone positions (down to 29.5 cm MAE in scene B with nine known microphones), and gives preliminary multi-source results of 45.21 cm and 69.86 cm MAE for two and three overlapping sources using the same eleven microphones.

Load-bearing premise

The reported single-source results depend on a training signal the paper does not actually supply: its source-localization loss (Eq. 8) sums over pairs of sources, so with a single source the sum is empty, and no alternative absolute-coordinate loss is stated.

Editorial extensions

If this is right

  • With seven microphones the model reaches 28.7 cm MAE on speech3, close to the 28.4 cm that wav2pos needs nine microphones to achieve, so deployment cost can drop.
  • Faulty microphones no longer force catastrophic failure: the model estimates an unknown microphone position to about 29.5 cm MAE in scene B with nine known microphones.
  • Preliminary multi-source results of 45.21 cm and 69.86 cm MAE for two and three simultaneous sources suggest the same hardware can handle overlapping sound events.
  • Inference takes 0.4 seconds with a 6.8M-parameter model, and the frozen audio encoder removes the need to train the acoustic front end.

Reading between the lines

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

  • If the missing single-source loss is resolved in the code, the reported gains may turn out to come mostly from the frozen BEATs embeddings and masked-reconstruction pretraining rather than from explicit geometric supervision; an ablation that adds a plain absolute-coordinate loss would show this.
  • The same architecture can be read as a self-calibrating array: by treating every microphone position as latent, the model could estimate an entire unknown array geometry from audio alone, which could be tested on ad-hoc recordings with moving microphones.
  • Because sparse cross-attention keeps only the top-T weights per query, the method should scale to much larger microphone counts than tested; a stress test with 20-50 microphones would show whether accuracy saturates or keeps improving.
  • The adaptive coherence weighting is a plausible source of robustness in reverberant rooms; a synthetic room-impulse-response sweep with varying reverberation time would separate its contribution from that of the pretrained encoder.
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

3 major / 6 minor

Summary. The paper proposes a 3D sound source localization framework that combines a pretrained audio encoder (BEATs), NGCC-PHAT features with adaptive signal coherence metrics, sparse cross-attention, and a masked autoencoder-style reconstruction. It claims state-of-the-art single-source localization on the LuViRA music3 and speech3 datasets, improved performance with fewer microphones, and the ability to estimate faulty microphone positions. The training objective is stated as a weighted sum of reconstruction loss, microphone-coordinate loss, and a pair-wise source localization loss. The experiments compare against multilateration, DI-NN, GNN, and wav2pos baselines.

Significance. If the results hold, the framework would offer a meaningful improvement in 3D sound source localization efficiency and robustness, particularly the capability to operate with fewer microphones and to estimate faulty microphone positions. The main strengths are the empirical comparisons on the public LuViRA dataset, the reported gains in MAE and acc@30cm, and the ablation-style study of microphone reduction and fault tolerance. However, the central claims rest on a training objective that is internally inconsistent for the single-source experiments that constitute the main results, and the fault-tolerance evaluation does not directly measure source localization error under microphone faults. The paper also omits key implementation details and does not release code, limiting reproducibility. The contribution is potentially valuable but not currently established by the manuscript as written.

major comments (3)
  1. [Section 3.2.3, Eq. (8)] The source localization loss in Eq. (8) is defined as a sum over source pairs (i,j)∈K of the absolute difference of squared pairwise distances. For K=1, which is exactly the setting of Tables 1 and 2, this sum is empty, so the position decoder receives no gradient from any source-coordinate objective. The text preceding Eq. (8) states that the loss combines reconstruction, microphone-coordinate, and source-localization losses, but no absolute-coordinate loss is given anywhere. The reported single-source results therefore cannot be produced by the objective as written. This is a load-bearing internal inconsistency in the central empirical claim; the authors must either state the correct loss (e.g., an L1/L2 loss on source coordinates) or release the training code to demonstrate the actual objective used.
  2. [Section 4.3, Table 3] The fault-tolerance claims in the abstract and introduction are supported only by microphone-position MAE, not by source localization error under faulty or unknown microphone positions. Table 3 reports MAE for estimating unknown microphone locations in scenes A and B, but the paper never reports source localization MAE in those fault scenarios. Since the claimed benefit is that the system remains accurate for source localization when microphones fail or have unknown positions, the evidence presented is indirect and incomplete. The authors should report source localization error under faulty-microphone conditions (both with known and estimated microphone positions) to substantiate the fault-tolerance claim.
  3. [Section 4.1 and Section 3.2] Reproducibility is severely limited by missing hyperparameters and design choices that are central to the method. The ASCM coherence exponent α in Eq. (4), the number of learned NGCC-PHAT filters P, the sparse cross-attention top-T, the masking ratio used in training, and the loss weights λ_sound, λ_m-loc, and λ_s-loc are not specified. Furthermore, no code or trained models are provided, and the baseline comparison relies on 'following their original hyperparameters' without reporting whether the same data splits, evaluation protocol, and training budgets were used. Without these details, the benchmark comparisons in Tables 1 and 2 cannot be independently verified or reproduced.
minor comments (6)
  1. [Abstract and Introduction] The abstract calls the multi-source experiments 'preliminary', but the introduction and conclusion claim scalability for multi-source localization as a key contribution. The paper should clarify whether multi-source localization is a demonstrated result or a preliminary observation.
  2. [Equation (2)] The superscripts in r^{PHAT}_{ij} are rendered incorrectly in the text (appearing as 'rPHA T'), which makes the equation hard to read. This is likely a LaTeX formatting issue that should be fixed.
  3. [Section 3.2.3, Eq. (8)] The notation for loss weights is inconsistent: the text uses λsound, λm-loc, and λs-loc, while Eq. (8) uses λ_sound, λ_m-loc, and λ_s-loc. The authors should unify the notation.
  4. [Table 2] The entry for Multilat* at M=5 is listed as N/A, but no explanation is given. If the method cannot be run in that configuration, the reason should be stated.
  5. [Section 4.2] The multi-source localization results (MAE 45.21 cm for two sources and 69.86 cm for three sources) are reported in a single sentence with no comparison to baselines, no variance over runs, and no information about how the number of sources K is determined during inference. This is insufficient to support the scalability claim.
  6. [Figure 2] Figure 2 is dense and the labels for the 'four highlighted components' mentioned in the caption are not clearly identifiable. The figure should be simplified or annotated more explicitly to help the reader map the components to the text.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: central claims are empirical benchmark comparisons against external baselines, not derivations that reduce to their inputs.

full rationale

The paper's main claims are experimental results on the external LuViRA dataset, compared against baselines (Multilat, DI-NN, GNN, wav2pos) that are trained and evaluated under the same protocol. There is no derivation chain in which a predicted quantity is defined in terms of the quantity it claims to predict, and no fitted parameter is renamed as a prediction. The only self-citation is reference [27] (He and Markham), which is a prior work by one co-author; it is not load-bearing for any claim in the paper. The model's components (BEATs pretrained encoder, NGCC-PHAT TDOA features, cross-attention, masked autoencoder, position decoder) are architectural choices, not circular inputs. One internal inconsistency exists: Eq. (8) defines the source-localization loss as a sum over source pairs, which is empty for the single-source experiments reported in Tables 1 and 2, so the stated training objective does not explicitly supervise single-source absolute position prediction. This is a correctness/consistency concern about the reported experiments, not a circularity: the MAE and accuracy numbers are not forced by construction, and the issue could be fixed by a corrected loss equation or a clarified training objective. Because the paper's central results are external empirical comparisons and no step reduces by definition to its inputs, the circularity score is 0.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The method is an empirical deep-learning pipeline; the central claim depends on standard signal-model assumptions, transferability of frozen pretrained audio features, and the correctness of the training objective as implemented. No free physical parameters are introduced, but several architectural hyperparameters are omitted.

free parameters (5)
  • alpha (ASCM coherence exponent) = unspecified
    Appears in Eq. (4) as the exponent on the coherence weights c_ij; no value, range, or learning rule is given, and it controls how much low-coherence microphone pairs are downweighted.
  • lambda_sound, lambda_m-loc, lambda_s-loc = unspecified
    Weighting of the three loss terms in Eq. (8); values are not reported and directly control the optimization balance.
  • sparse cross-attention top-T = unspecified
    Number of retained attention weights per query is never specified, so the claimed computational savings and the fusion behavior are not reproducible.
  • masking ratio = unspecified
    The masked encoder randomly masks tokens during training, but the masking probability is not given.
  • NGCC-PHAT filter count P = unspecified
    Eq. (2) sums P learned filters to form the refined correlation; P is not reported.
assumptions (3)
  • domain assumption Room impulse response mixture model in Eq. (1) is accurate for the LuViRA recordings.
    The environment is modeled as linear convolution of clean source signals plus i.i.d. Gaussian noise; real recordings include reverberation, noise correlation, and possibly model mismatch.
  • domain assumption Frozen BEATs embeddings carry enough spatial information for SSL.
    BEATs is pretrained for general audio tasks, not SSL; the paper provides no analysis of what spatial information survives and does not fine-tune the encoder.
  • ad hoc to paper The non-source losses in Eq. (8) are sufficient to anchor absolute source positions.
    The source localization loss is pairwise only; the paper implicitly assumes audio reconstruction and microphone-coordinate reconstruction force absolute source coordinates, but this is not demonstrated and is likely false for K=1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient and Microphone-Fault-Tolerant 3D Sound Source Localization." pith.science (2026). https://pith.science/paper/SZNA6CLU

@misc{pith2026250520961,
  author       = {Pith},
  title        = {Pith review of: Efficient and Microphone-Fault-Tolerant 3D Sound Source Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SZNA6CLU}},
  note         = {Machine review of arXiv:2505.20961}
}
read the original abstract

Sound source localization (SSL) is a critical technology for determining the position of sound sources in complex environments. However, existing methods face challenges such as high computational costs and precise calibration requirements, limiting their deployment in dynamic or resource-constrained environments. This paper introduces a novel 3D SSL framework, which uses sparse cross-attention, pretraining, and adaptive signal coherence metrics, to achieve accurate and computationally efficient localization with fewer input microphones. The framework is also fault-tolerant to unreliable or even unknown microphone position inputs, ensuring its applicability in real-world scenarios. Preliminary experiments demonstrate its scalability for multi-source localization without requiring additional hardware. This work advances SSL by balancing the model's performance and efficiency and improving its robustness for real-world scenarios.

Figures

Figures reproduced from arXiv: 2505.20961 by the authors.

Figure 1
Figure 1. Overview of the sound source localization task. It aims at predicting the sound source position [XS, YS, ZS] condi￾tioned on the recordings from calibrated microphones at known positions [XRi, YRi, ZRi]. If Faulty Microphones exist, our model is also capable of predicting their position [XF , YF , ZF ] conditioned on their recording. In practical applications, 3D sound source localization faces several critical chal… view at source ↗
Figure 2
Figure 2. The workflow of the proposed framework. It consists of three streams: Acoustic stream (Section 3.2.1), Coordinate stream (Section 3.2.2), and Joint stream (Section 3.2.3). Within these, we specifically highlight four components that can enhance the efficiency and the parts that need to be trained or frozen during the training process. First, for the original microphone spatial coordinates {rm} M m=1, we apply a simp… view at source ↗
Figure 3
Figure 3. Result visualization under different settings with ran￾dom initialization of sound source and microphone positions. achieving 96.8% and 88.4% accuracy, respectively. Our model has 6.8M parameters, with an average training time of 21.5 hours and an inference time of 0.4 seconds based on an A10 GPU. Be￾sides, [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

35 extracted references · 34 canonical work pages

  1. [1]

    Efficient and Microphone-Fault-Tolerant 3D Sound Source Localization

    Introduction Sound source localization (SSL) involves accurately determining the direction and distance of the sound sources in complex and noisy environments [9]. It is crucial for enhancing situational awareness, facilitating human-computer interaction, and optimiz- ing the performance of various signal-processing applications. For instance, prior works...

    work page Pith review arXiv 2025

  2. [2]

    Traditional Sound Source Localization Traditional approaches to sound source localization are based on establishing geometric and statistical frameworks

    Related Work 2.1. Traditional Sound Source Localization Traditional approaches to sound source localization are based on establishing geometric and statistical frameworks. In the sim- plest scenario, when the number of independent distance mea- surements equals the spatial dimension, trilateration provides an elegant closed-form solution [14, 15]. However...

    work page

  3. [3]

    Problem Definition Consider a 3D acoustic environment with M microphones at known coordinates rm ∈ R3, m = 1,

    Methodology 3.1. Problem Definition Consider a 3D acoustic environment with M microphones at known coordinates rm ∈ R3, m = 1, . . . , Mand K simulta- neous sound sources located at unknown positions rk ∈ R3, k = 1, . . . , K(K ≪ M). During a short time frame of N sam- ples, each source emits a distinct signal sk ∈ RN , while every microphone records a su...

    work page

  4. [4]

    Experiment 4.1. Experimental Setup Datasets: We use two recordings, music3 and speech3, from LuViRA [1] audio-only dataset for evaluation and other 3 music and 3 speech recordings for training and validation. The sound source is randomly sampled in a room of size 7 × 8 × 2 m. Eval- uation metrics: The performance of localization is assessed using Mean Abs...

    work page

  5. [5]

    Conclusion This paper presented a novel 3D sound source localization frame- work to address key challenges in efficiency and robustness against microphone faults. By integrating sparse cross-attentions, adaptive signal coherence metrics, and pretrained audio encoders, the proposed method achieves accurate localization with fewer microphones while maintain...

    work page

  6. [6]

    The luvira dataset: Synchronized vision, radio, and audio sensors for indoor localization,

    I. Yaman, G. Tian, M. Larsson, P. Persson, M. Sandra, A. D ¨urr, E. Tegler, N. Challa, H. Garde, F. Tufvesson et al., “The luvira dataset: Synchronized vision, radio, and audio sensors for indoor localization,” in 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024, pp. 11 920–11 926

    work page 2024

  7. [7]

    Extension of time-difference-of-arrival self calibration solutions using ro- bust multilateration,

    K. ˚Astr¨om, M. Larsson, G. Flood, and M. Oskarsson, “Extension of time-difference-of-arrival self calibration solutions using ro- bust multilateration,” in 2021 29th European Signal Processing Conference (EUSIPCO). IEEE, 2021, pp. 870–874

    work page 2021

  8. [8]

    Dual input neural net- works for positional sound source localization,

    E. Grinstein, V . W. Neo, and P. A. Naylor, “Dual input neural net- works for positional sound source localization,” EURASIP Journal on Audio, Speech, and Music Processing, vol. 2023, no. 1, p. 32, 2023

    work page 2023

Show all 35 references
  1. [9]

    Graph neural net- works for sound source localization on distributed microphone networks,

    E. Grinstein, M. Brookes, and P. A. Naylor, “Graph neural net- works for sound source localization on distributed microphone networks,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2023, pp. 1–5

    2023

  2. [10]

    wav2pos: Sound source localization using masked au- toencoders,

    A. Berg, J. Gulin, M. O’Connor, C. Zhou, K. ˚Astr¨om, and M. Os- karsson, “wav2pos: Sound source localization using masked au- toencoders,” in 2024 14th International Conference on Indoor Positioning and Indoor Navigation (IPIN). IEEE, 2024, pp. 1–8

    2024

  3. [11]

    Extending gcc-phat using shift equivariant neural networks,

    A. Berg, M. O’Connor, K. ˚Astr¨om, and M. Oskarsson, “Extending gcc-phat using shift equivariant neural networks,”arXiv preprint arXiv:2208.04654, 2022

    2022 arXiv

  4. [12]

    Beats: Audio pre-training with acoustic tokenizers,

    S. Chen, Y . Wu, C. Wang, S. Liu, D. Tompkins, Z. Chen, and F. Wei, “Beats: Audio pre-training with acoustic tokenizers,”arXiv preprint arXiv:2212.09058, 2022

    2022 arXiv

  5. [13]

    A survey of sound source localization with deep learning methods,

    P.-A. Grumiaux, S. Kiti ´c, L. Girin, and A. Gu ´erin, “A survey of sound source localization with deep learning methods,” The Journal of the Acoustical Society of America, vol. 152, no. 1, pp. 107–151, 2022

    2022

  6. [14]

    A. N. Popper, R. R. Fay, and A. N. Popper,Sound source localiza- tion. Springer, 2005, vol. 25

    2005

  7. [15]

    A review on sound source localiza- tion systems,

    D. Desai and N. Mehendale, “A review on sound source localiza- tion systems,” Archives of Computational Methods in Engineering, vol. 29, no. 7, pp. 4631–4642, 2022

    2022

  8. [16]

    End-to-end two-dimensional sound source localization with ad-hoc microphone arrays,

    Y . Gong, S. Liu, and X.-L. Zhang, “End-to-end two-dimensional sound source localization with ad-hoc microphone arrays,” in2022 Asia-Pacific Signal and Information Processing Association An- nual Summit and Conference (APSIPA ASC). IEEE, 2022, pp. 1944–1949

    2022

  9. [17]

    A novel 3d multilateration sensor using distributed ultrasonic beacons for indoor navigation,

    R. Kapoor, S. Ramasamy, A. Gardi, C. Bieber, L. Silverberg, and R. Sabatini, “A novel 3d multilateration sensor using distributed ultrasonic beacons for indoor navigation,” Sensors, vol. 16, no. 10, p. 1637, 2016

    2016

  10. [18]

    Sound source localization and classification for emergency vehicle siren detection using resource constrained systems,

    R. Munirathinam and S. Vitek, “Sound source localization and classification for emergency vehicle siren detection using resource constrained systems,” in 2024 34th International Conference Ra- dioelektronika (RADIOELEKTRONIKA). IEEE, 2024, pp. 1–5

    2024

  11. [19]

    Revisiting trilateration for robot localiza- tion,

    F. Thomas and L. Ros, “Revisiting trilateration for robot localiza- tion,” IEEE Transactions on robotics, vol. 21, no. 1, pp. 93–101, 2005

    2005

  12. [20]

    Efficient solution and performance analysis of 3-d position estimation by trilateration,

    D. E. Manolakis, “Efficient solution and performance analysis of 3-d position estimation by trilateration,” IEEE Transactions on Aerospace and Electronic systems, vol. 32, no. 4, pp. 1239–1248, 1996

    1996

  13. [21]

    A simple globally convergent algorithm for the nonsmooth nonconvex sin- gle source localization problem,

    D. R. Luke, S. Sabach, M. Teboulle, and K. Zatlawey, “A simple globally convergent algorithm for the nonsmooth nonconvex sin- gle source localization problem,” Journal of Global Optimization, vol. 69, pp. 889–909, 2017

    2017

  14. [22]

    Solvit: A reference-free source localization technique using majorization minimization,

    R. Jyothi and P. Babu, “Solvit: A reference-free source localization technique using majorization minimization,” IEEE/ACM Trans- actions on Audio, Speech, and Language Processing, vol. 28, pp. 2661–2673, 2020

    2020

  15. [23]

    Closed-form algorithms in mobile positioning: Myths and misconceptions,

    N. Sirola, “Closed-form algorithms in mobile positioning: Myths and misconceptions,” in 2010 7th Workshop on Positioning, Navi- gation and Communication. Ieee, 2010, pp. 38–44

    2010

  16. [24]

    Opti- mal trilateration is an eigenvalue problem,

    M. Larsson, V . Larsson, K. Astrom, and M. Oskarsson, “Opti- mal trilateration is an eigenvalue problem,” in ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019, pp. 5586–5590

    2019

  17. [25]

    The generalized correlation method for estimation of time delay,

    C. Knapp and G. Carter, “The generalized correlation method for estimation of time delay,” IEEE transactions on acoustics, speech, and signal processing, vol. 24, no. 4, pp. 320–327, 1976

    1976

  18. [26]

    Time delay estimation for speaker localization using cnn-based parametrized gcc-phat features

    D. Salvati, C. Drioli, G. L. Foresti et al., “Time delay estimation for speaker localization using cnn-based parametrized gcc-phat features.” in Interspeech, 2021, pp. 1479–1483

    2021

  19. [27]

    Syncnet: Correlating objective for time delay estimation in audio signals,

    A. Raina and V . Arora, “Syncnet: Correlating objective for time delay estimation in audio signals,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Process- ing (ICASSP). IEEE, 2023, pp. 1–5

    2023

  20. [28]

    Robust sound source tracking using srp-phat and 3d convolutional neural net- works,

    D. Diaz-Guerra, A. Miguel, and J. R. Beltran, “Robust sound source tracking using srp-phat and 3d convolutional neural net- works,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 29, pp. 300–311, 2021

    2021

  21. [29]

    Sr-srp: Super-resolution based srp- phat for sound source localization and tracking,

    J.-H. Cho and J.-H. Chang, “Sr-srp: Super-resolution based srp- phat for sound source localization and tracking,” in Proc. Inter- speech 2023, 2023, pp. 3769–3773

    2023

  22. [30]

    Fn-ssl: Full-band and narrow- band fusion for sound source localization,

    Y . Wang, B. Yang, and X. Li, “Fn-ssl: Full-band and narrow- band fusion for sound source localization,” arXiv preprint arXiv:2305.19610, 2023

    2023 arXiv

  23. [31]

    Binaural sound localization in noisy environments using frequency-based audio vision transformer (favit),

    W. Phokhinanan, N. Obin, and S. Argentieri, “Binaural sound localization in noisy environments using frequency-based audio vision transformer (favit),” in INTERSPEECH. ISCA, 2023, pp. 3704–3708

    2023

  24. [32]

    Sounddoa: Learn sound source direction of arrival and semantics from sound raw waveforms,

    Y . He and A. Markham, “Sounddoa: Learn sound source direction of arrival and semantics from sound raw waveforms,” in Inter- speech, 2022

    2022

  25. [33]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al., “Pytorch: An imperative style, high-performance deep learning library,” inAd- vances in Neural Information Processing Systems, vol. 32, 2019

    2019

  26. [34]

    Sound source localization us- ing deep learning models,

    N. Yalta, K. Nakadai, and T. Ogata, “Sound source localization us- ing deep learning models,” Journal of Robotics and Mechatronics, vol. 29, no. 1, pp. 37–48, 2017

    2017

  27. [35]

    Deep learning-based method for multiple sound source localization with high resolution and accuracy,

    S. Y . Lee, J. Chang, and S. Lee, “Deep learning-based method for multiple sound source localization with high resolution and accuracy,”Mechanical Systems and Signal Processing, vol. 161, p. 107959, 2021

    2021

Pith tools

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