Pith. sign in

REVIEW 4 major objections 5 minor 43 references

Distributed Asynchronous Device Speech Enhancement via Windowed Cross-Attention

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A windowed cross-attention module lets speech enhancement networks align and aggregate asynchronous microphone streams, outperforming the same-time-index TAC module on both tested backbones.

desk verdict A sensible attention module and a fair latency comparison, but the clock-drift claim is not exercised by the simulation—the contribution is narrower than the title suggests. read the letter →

arxiv 2507.16104 v1 pith:NS7OEWOS submitted 2025-07-21 eess.AS

Gene-Ping Yang , Sebastian Braun This is my paper · ORCID
classification eess.AS
keywords asynchronousmicrophonearraysclockdriftcross-attentionspeechenhancementmulti-talkerad-hocarraydevicelatencytransform-average-concatenate
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

Most multi-microphone speech enhancement assumes the microphone streams are time-synchronized; in a real meeting each device records on its own clock, so the streams arrive with unknown latency and gradually drift apart. This paper proposes a windowed cross-attention (WCA) module that aligns features across microphones by letting each time frame attend to a small neighborhood of frames in every other microphone, instead of averaging the same time index the way the standard transform-average-concatenate (TAC) module does. Inserting WCA into two existing enhancement networks and training toward the closest microphone to each active speaker yields better perceptual quality than TAC under simulated latency and clock drift, with the best configuration reaching an overall DNSMOS quality score (OVRL) of 2.41 versus 1.92 for TAC. The practical stake is that meeting capture could use every device in the room without manual muting, explicit synchronization, or a fixed array geometry.

What carries the argument

The central object is the Windowed Cross-Attention (WCA) layer, a cross-attention computation restricted in time. For each frame $i$ of microphone $m$, the query $Q_m[i]$ attends only to keys and values drawn from frames $j$ of every other microphone $n$ with $|i-j| \le L$; with $L=4$ encoded frames this is a 90 ms window in the experiments. The output is $A_m[i] = \sum_n \mathrm{softmax}\big(Q_m[i] \cdot K_n^u[i]^\top / \sqrt{d}\big) \, V_n^u[i]$, where $K_n^u[i]$ and $V_n^u[i]$ are the unfolded local windows of microphone $n$. Restricting attention in this way reduces the memory cost from $O(M^2 T^2)$ to $O(M^2 T L)$, allows gradual drift to be tracked within the window, and makes the module invariant to microphone permutation and to the number of microphones. The second load-bearing piece is the training target: the target is not a fixed reference microphone but the sum of direct-path signals from the microphone closest to each active speaker, which the paper argues is the better objective for multi-talker scenes.

What would settle it

Record a multi-minute real meeting with consumer devices whose clocks drift far enough that the relative offset between two streams exceeds 90 ms, or with bursts of packet loss, and compare WCA against TAC; if the WCA advantage disappears or reverses once the misalignment exceeds the fixed window, the claimed drift robustness is not general.

Watch

Extended reading notes

Core claim

The paper's central claim is that a temporal cross-attention restricted to a fixed local window can do the synchronization that explicit alignment methods struggle with. Under per-device latency up to 40 ms and clock drift modeled as per-device sample-rate offsets, models equipped with windowed cross-attention (WCA) beat transform-average-concatenate (TAC) models across nearly all configurations, and the largest gains come from training toward the closest microphone to each active speaker. On the CRUSE backbone with that target, WCA reaches a DNSMOS overall score (OVRL) of 2.41 and an XLSR-MOS score of 2.35, against 1.92 and 2.15 for TAC; WCA also outperforms the single-channel baseline with PickNet selection. The paper interprets this as evidence that WCA is a reusable, permutation- and number-invariant module for asynchronous multi-microphone enhancement, and that the closest-microphone-per-speaker target is the better training objective for multi-talker scenes.

Load-bearing premise

The central claim stands on the assumption that real-world clock drift and latency stay small enough to fall inside the fixed 90 ms attention window over the audio duration, with no large time-varying jitter or packet loss beyond what was simulated.

Editorial extensions

If this is right

  • Deploying WCA in an enhancement model means meeting capture can use all nearby devices simultaneously instead of muting all but one, because the model learns to align the asynchronous streams itself.
  • A model trained once with WCA is not tied to a fixed number or arrangement of microphones, since the attention is invariant to microphone permutation and count.
  • The closest-microphone-per-speaker training target improves perceived quality under multi-talker overlap, so systems that know which device is closest to each speaker can train toward a cleaner target than a fixed reference channel.
  • Faster convergence with WCA suggests the module also eases optimization relative to TAC, not just final quality.

Reading between the lines

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

  • The fixed 90 ms window sets an upper bound on the inter-device offset the model can absorb; for long recordings or heavy jitter, a wider window, an external drift-compensation front-end, or adaptive window sizing would be needed, none of which the paper tests.
  • The closest-microphone-per-speaker target is oracle-informed: forming it requires knowing which microphone is nearest to each active speaker at every moment, so a practical deployment would need a separate localization or speaker-diarization front-end to realize the same gains.
  • The reported metrics are non-intrusive perceptual scores that tolerate small time offsets, so the results demonstrate perceived-quality improvement but not necessarily sample-accurate synchronization; a downstream automatic speech recognition evaluation would be a stricter test of alignment quality.
  • The same windowed cross-attention mechanism could in principle align other unsynchronized sensor streams with bounded skew, such as distributed hearing-aid or wearable microphones, because the module only assumes a bounded temporal offset and permutation invariance.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a windowed cross-attention (WCA) module for multi-microphone speech enhancement in distributed asynchronous settings. WCA replaces the time-index-aligned TAC aggregation with a fixed temporal cross-attention window that can align features across devices with unknown latency and clock drift. The module is integrated into two backbones, iFaSNet and CRUSE, and evaluated on simulated noisy reverberant mixtures with multiple speakers, per-microphone time shifts, and per-device sample-rate offsets. Three training targets are compared: a fixed random microphone, the minimum-latency microphone, and the microphone closest to each active speaker. The reported results (DNSMOS, XLSR-MOS, and cepstral distance) show WCA generally outperforming TAC, with the closest-microphone target giving the largest gains. The paper also presents an ablation varying drift strength, delay magnitude, and speaker overlap, and concludes that WCA is robust to clock drift and generalizes to real asynchronous meeting conditions.

Significance. If the claims were fully supported, the WCA module would be a useful contribution: it is permutation- and microphone-count-invariant, has bounded memory cost through the fixed window, and can be dropped into existing enhancement backbones. The comparison with TAC under matched architectures is a fair and informative control, and the use of two independent backbones plus external learned MOS predictors strengthens the main architectural comparison. The closest-microphone training target for multi-talker mixtures is also a sensible idea that yields consistent improvements. However, the headline robustness-to-clock-drift claim is not actually exercised by the simulated conditions, because the accumulated drift over 10 s clips is sub-millisecond and the entire simulated latency range lies inside the attention window. The paper also lacks error bars or significance tests, and the best target strategy is selected on the same test set used for the final headline numbers. These issues leave the general claims about real asynchronous meetings unsupported, though the core WCA-versus-TAC comparison for small fixed offsets remains plausible.

major comments (4)
  1. [§5.1, Table 2] The clock-drift simulation is too weak to test the claimed robustness. Per-device sample rates are drawn from N(16000, 0.5) Hz, which is a 31 ppm offset; over a 10 s clip this accumulates at most roughly 0.3 ms of relative drift, and even the ablation with std 2 Hz accumulates only about 1.25 ms. These values are two orders of magnitude smaller than the configured 90 ms attention window, so Table 2 shows only that WCA handles small fixed offsets, not that it tracks time-varying drift. The statement in §5.4 that the model 'demonstrates robustness to clock drift' is not supported by the reported conditions; experiments with larger sample-rate offsets, longer utterances, or time-varying drift are needed.
  2. [§5.1, §5.2] The simulated latency range is fully covered by the attention window, so the window boundary is never stressed. Delays are sampled between -40 ms and 40 ms, giving a maximum pairwise offset of 80 ms, while L=4 is described as a 90 ms cross-attention window. The >100 ms network latency cited as motivation in §1 is therefore never tested. The conclusion that WCA handles large device latency requires either an experiment with offsets exceeding the window size or a principled argument that the window can be enlarged without performance loss.
  3. [Tables 1 and 2] No error bars, confidence intervals, or repeated-seed results are reported for any of the DNSMOS, XLSR-MOS, or CD numbers. Many of the reported advantages are small (for example, OVRL differences of 0.1-0.2), and it is not clear whether these are stable across training runs or test-set subsamples. The claim of 'consistent' improvement should be backed by variance estimates or a significance test, or at least by an explicit statement that the tables report a single run.
  4. [§5.4, Table 1] Two claims about Table 1 need correction or stronger support. First, iFaSNet with the random-microphone target does not consistently favor WCA: OVRL is 2.27 for WCA versus 2.28 for TAC, and SIG is 2.67 versus 2.68, so the text that says 'only a slight improvement over TAC' is not accurate for those metrics. Second, the closest-microphone target is declared the best among the three strategies after observing test-set results, and the headline OVRL/XLSR numbers are then reported for that same test set; this target-selection step should be performed on a validation set or otherwise accounted for to avoid optimistic selection bias.
minor comments (5)
  1. [Abstract and §4.3] The abstract calls the closest-microphone target an 'optimal training target,' but the paper only shows it is empirically best among three strategies on one test set; this wording overstates the finding unless a theoretical argument or broader evaluation is provided.
  2. [§1, §2, §3] There are several typos and grammatical slips, including 'awkard' in §1, 'microhpone' in §2, and 'synchonization' in §3; these should be corrected.
  3. [Eq. (9)] The softmax normalization in the windowed attention is not fully specified: it is unclear whether the softmax is taken over the 2L+1 window within each microphone pair n, or jointly across all n and the window. This matters for the definition of the attention weights and should be stated unambiguously.
  4. [§5.2] The mapping from L=4 to a '90 ms' window is not derived. With a 20 ms STFT window and 50% overlap, the hop size is 10 ms, and a window spanning i-L to i+L covers 80 ms of offsets plus frame boundary effects; the paper should state the exact relationship between L, hop length, and the physical offset range in milliseconds.
  5. [§5.4, Conclusion] The claim that the method 'generalized well to real-world conditions' is supported only by an unquantified website of audio examples. If a real-recording evaluation exists, it should be reported with metrics; otherwise the claim should be tempered to 'anecdotally demonstrated on audio examples.'

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: WCA is empirically compared against TAC, and the target-selection and drift-simulation choices are evaluation-coverage issues, not by-construction equivalences.

full rationale

The claimed derivation chain is not circular. WCA is defined by Eq. (9) as a windowed cross-attention over keys and values within +-L frames; it is not defined in terms of TAC's output, and it does not reduce to TAC by construction. TAC, Eq. (4), is a per-frame average with no temporal window, while WCA uses learned attention over a 90 ms window with L=4. The WCA-versus-TAC comparison is empirical on two independent backbones, iFaSNet and CRUSE, with metrics from external predictors DNSMOS and XLSR-MOS; no equation forces WCA to score higher. The 'closest microphone per speaker' target, Eq. (10), is a training objective, not a fitted parameter; selecting it as 'optimal' from the same test set is a model-selection concern that weakens that particular claim, but it does not make the WCA-versus-TAC comparison circular. The clock-drift robustness statement rests on a simulation whose drift magnitudes, N(16000, 0.5) Hz and 2 Hz in the ablation, accumulate to well under 1.3 ms over 10 s, far smaller than the 90 ms window; this is an adequacy-of-evaluation issue, not a circularity. Self-citations [34], [35], [39], and [40] supply the CRUSE backbone and loss, which are published, externally usable systems and are not invoked as an unverified uniqueness theorem. Therefore no circular step can be exhibited under the definitional, fitted-input, or self-citation criteria.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central additions are an attention window L and a training target; both rely on domain assumptions about the asynchrony model and the oracle target. No new physical entities are introduced.

free parameters (2)
  • Cross-attention window size L = 4 frames (90 ms)
    Hand-set in Section 5.2 to cover the simulated maximum device latency of 40 ms. It caps the misalignment WCA can absorb; larger real-world drift or latency would require larger L.
  • Clock drift simulation standard deviation = 0.5 Hz (mean 16 kHz); ablation up to 2 Hz
    Chosen in Section 5.1 to generate asynchrony. At these values the accumulated drift over 10 s is sub-millisecond to about 1.25 ms, so clock drift is effectively not stressed, which shapes the robustness claim.
assumptions (4)
  • domain assumption The observation model x_m(t) = sum_k h_{m,k}(t) * s_k(t) + n_m(t), with observation x_m(gamma_m t - tau_m), is a valid model for asynchronous device streams, assuming constant latency and drift over a clip.
    Section 2, Eqs. (1)-(2). Real networks may have jitter, packet loss, and time-varying latency, which are not modeled.
  • domain assumption The closest-microphone-per-speaker training target (Eq. 10) is computable at training time, which requires knowing the active speaker-to-microphone geometry.
    Section 4.3. This oracle target is used for the best-performing models; the paper does not show how to obtain it without positional knowledge.
  • ad hoc to paper Cross-attention restricted to a fixed window L can implicitly learn the temporal alignment between streams.
    Section 4.2, Eq. (9). This is the core design hypothesis; no proof or analysis of alignment behavior is given, only empirical results.
  • domain assumption DNSMOS P.835 and XLSR-MOS are valid perceptual proxies for evaluation of asynchronous outputs.
    Section 5.3. These learned metrics are not sensitive to absolute delay, so a clean but delayed output can score well; no human listening test is included.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Distributed Asynchronous Device Speech Enhancement via Windowed Cross-Attention." pith.science (2026). https://pith.science/paper/NS7OEWOS

@misc{pith2026250716104,
  author       = {Pith},
  title        = {Pith review of: Distributed Asynchronous Device Speech Enhancement via Windowed Cross-Attention},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NS7OEWOS}},
  note         = {Machine review of arXiv:2507.16104}
}
read the original abstract

The increasing number of microphone-equipped personal devices offers great flexibility and potential using them as ad-hoc microphone arrays in dynamic meeting environments. However, most existing approaches are designed for time-synchronized microphone setups, a condition that may not hold in real-world meeting scenarios, where time latency and clock drift vary across devices. Under such conditions, we found transform-average-concatenate (TAC), a popular module for neural multi-microphone processing, insufficient in handling time-asynchronous microphones. In response, we propose a windowed cross-attention module capable of dynamically aligning features between all microphones. This module is invariant to both the permutation and the number of microphones and can be easily integrated into existing models. Furthermore, we propose an optimal training target for multi-talker environments. We evaluated our approach in a multi-microphone noisy reverberant setup with unknown time latency and clock drift of each microphone. Experimental results show that our method outperforms TAC on both iFaSNet and CRUSE models, offering faster convergence and improved learning, demonstrating the efficacy of the windowed cross-attention module for asynchronous microphone setups.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [1]

    Auto-localization in ad-hoc microphone arrays,

    N. D. Gaubitch, W. B. Kleijn, and R. Heusdens, “Auto-localization in ad-hoc microphone arrays,” in ICASSP, 2013

    work page 2013

  2. [2]

    Self-localization of ad-hoc arrays using time difference of arrivals,

    L. Wang, T.-K. Hon, J. D. Reiss, and A. Cavallaro, “Self-localization of ad-hoc arrays using time difference of arrivals,” IEEE Transactions on Signal Processing, vol. 64, 2015

    work page 2015

  3. [3]

    Blind synchronization in wireless acoustic sensor networks,

    D. Cherkassky and S. Gannot, “Blind synchronization in wireless acoustic sensor networks,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , 2017

    work page 2017

  4. [4]

    Blind sampling rate offset estimation for wireless acoustic sensor networks through weighted least- squares coherence drift estimation,

    M. H. Bahari, A. Bertrand, and M. Moonen, “Blind sampling rate offset estimation for wireless acoustic sensor networks through weighted least- squares coherence drift estimation,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , 2017

    work page 2017

  5. [5]

    Distributed sampling rate offset estimation over acoustic sensor networks based on asynchronous network newton optimization,

    D. Hu, H. Zhang, F. Bao, and R. Wang, “Distributed sampling rate offset estimation over acoustic sensor networks based on asynchronous network newton optimization,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2022

    work page 2022

  6. [6]

    Multi-stage coherence drift based sampling rate synchro- nization for acoustic beamforming,

    J. Schmalenstroeer, J. Heymann, L. Drude, C. Boeddecker, and R. Haeb- Umbach, “Multi-stage coherence drift based sampling rate synchro- nization for acoustic beamforming,” in 2017 IEEE 19th International Workshop on Multimedia Signal Processing (MMSP) . IEEE, 2017

    work page 2017

  7. [7]

    Fasnet: Low- latency adaptive beamforming for multi-microphone audio processing,

    Y . Luo, C. Han, N. Mesgarani, E. Ceolini, and S.-C. Liu, “Fasnet: Low- latency adaptive beamforming for multi-microphone audio processing,” in 2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2019

    work page 2019

  8. [8]

    End-to-end microphone permutation and number invariant multi-channel speech separation,

    Y . Luo, Z. Chen, N. Mesgarani, and T. Yoshioka, “End-to-end microphone permutation and number invariant multi-channel speech separation,” in ICASSP, 2020

    work page 2020

Show all 43 references
  1. [9]

    Implicit filter-and-sum network for end-to-end multi-channel speech separation,

    Y . Luo and N. Mesgarani, “Implicit filter-and-sum network for end-to-end multi-channel speech separation,” in Interspeech, 2021

    2021

  2. [10]

    Multi-channel deep clustering: Discriminative spectral and spatial embeddings for speaker- independent speech separation,

    Z.-Q. Wang, J. Le Roux, and J. R. Hershey, “Multi-channel deep clustering: Discriminative spectral and spatial embeddings for speaker- independent speech separation,” in ICASSP, 2018

    2018

  3. [11]

    Enhanced deep speech separation in clustered ad hoc distributed microphone environments,

    J. Kim, S. Kindt, N. Madhu, and H.-G. Kang, “Enhanced deep speech separation in clustered ad hoc distributed microphone environments,” in Interspeech, 2024

    2024

  4. [12]

    Deep ad-hoc beamforming,

    X.-L. Zhang, “Deep ad-hoc beamforming,” Computer Speech & Language, vol. 68, 2021

    2021

  5. [13]

    Vararray: Array-geometry-agnostic continuous speech separation,

    T. Yoshioka, X. Wang, D. Wang, M. Tang, Z. Zhu, Z. Chen, and N. Kanda, “Vararray: Array-geometry-agnostic continuous speech separation,” in ICASSP, 2022

    2022

  6. [14]

    Time- domain ad-hoc array speech enhancement using a triple-path network,

    A. Pandey, B. Xu, A. Kumar, J. Donley, P. Calamia, and D. Wang, “Time- domain ad-hoc array speech enhancement using a triple-path network,” in Interspeech, 2022

    2022

  7. [15]

    Speech enhancement with ad-hoc microphone array using single source activity,

    R. Sakanashi, N. Ono, S. Miyabe, T. Yamada, and S. Makino, “Speech enhancement with ad-hoc microphone array using single source activity,” in 2013 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference , 2013

    2013

  8. [16]

    A framework for speech enhancement with ad hoc microphone arrays,

    V . M. Tavakoli, J. R. Jensen, M. G. Christensen, and J. Benesty, “A framework for speech enhancement with ad hoc microphone arrays,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 24, 2016

    2016

  9. [17]

    High-resolution frequency-wavenumber spectrum analysis,

    J. Capon, “High-resolution frequency-wavenumber spectrum analysis,” Proceedings of the IEEE , vol. 57, 1969

    1969

  10. [18]

    An alternative approach to linearly constrained adaptive beamforming,

    L. Griffiths and C. Jim, “An alternative approach to linearly constrained adaptive beamforming,” IEEE Transactions on antennas and propagation, vol. 30, 1982

    1982

  11. [19]

    On optimal frequency-domain multichannel linear filtering for noise reduction,

    M. Souden, J. Benesty, and S. Affes, “On optimal frequency-domain multichannel linear filtering for noise reduction,” IEEE Transactions on audio, speech, and language processing , vol. 18, 2009

    2009

  12. [20]

    A consoli- dated perspective on multimicrophone speech enhancement and source separation,

    S. Gannot, E. Vincent, S. Markovich-Golan, and A. Ozerov, “A consoli- dated perspective on multimicrophone speech enhancement and source separation,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 25, 2017

    2017

  13. [21]

    Comparison of reference microphone selection algorithms for distributed microphone array based speech enhancement in meeting recognition scenarios,

    S. Araki, N. Ono, K. Kinoshita, and M. Delcroix, “Comparison of reference microphone selection algorithms for distributed microphone array based speech enhancement in meeting recognition scenarios,” in 2018 16th International Workshop on Acoustic Signal Enhancement (IWAENC), 2018

    2018

  14. [22]

    Channel selection based on multichannel cross-correlation coefficients for distant speech recognition,

    K. Kumatani, J. McDonough, J. F. Lehman, and B. Raj, “Channel selection based on multichannel cross-correlation coefficients for distant speech recognition,” in 2011 Joint Workshop on Hands-free Speech Communication and Microphone Arrays . IEEE, 2011

    2011

  15. [23]

    Channel selection measures for multi- microphone speech recognition,

    M. Wolf and C. Nadeu, “Channel selection measures for multi- microphone speech recognition,” Speech Communication, vol. 57, 2014

    2014

  16. [24]

    A study on refer- ence microphone selection for multi-microphone speech enhancement,

    J. Zhang, H. Chen, L.-R. Dai, and R. C. Hendriks, “A study on refer- ence microphone selection for multi-microphone speech enhancement,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 29, 2020

    2020

  17. [25]

    Picknet: Real-time channel selection for ad hoc microphone arrays,

    T. Yoshioka, X. Wang, and D. Wang, “Picknet: Real-time channel selection for ad hoc microphone arrays,” in ICASSP, 2022

    2022

  18. [26]

    Continuous speech separation with ad hoc microphone arrays,

    D. Wang, T. Yoshioka, Z. Chen, X. Wang, T. Zhou, and Z. Meng, “Continuous speech separation with ad hoc microphone arrays,” in 2021 29th European Signal Processing Conference (EUSIPCO) . IEEE, 2021

    2021

  19. [27]

    Microphone subset selection for mvdr beamformer based noise reduction,

    J. Zhang, S. P. Chepuri, R. C. Hendriks, and R. Heusdens, “Microphone subset selection for mvdr beamformer based noise reduction,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , 2018

    2018

  20. [28]

    Sensor selection for relative acoustic transfer function steered linearly-constrained beamformers,

    J. Zhang, J. Du, and L.-R. Dai, “Sensor selection for relative acoustic transfer function steered linearly-constrained beamformers,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , 2021

    2021

  21. [29]

    Blind compensation of interchannel sampling frequency mismatch for ad hoc microphone array based on maximum likelihood estimation,

    S. Miyabe, N. Ono, and S. Makino, “Blind compensation of interchannel sampling frequency mismatch for ad hoc microphone array based on maximum likelihood estimation,” Signal Processing, vol. 107, 2015

    2015

  22. [30]

    Meeting transcription using asynchronous distant microphones

    T. Yoshioka, D. Dimitriadis, A. Stolcke, W. Hinthorn, Z. Chen, M. Zeng, and X. Huang, “Meeting transcription using asynchronous distant microphones.” in Interspeech, 2019

    2019

  23. [31]

    Samba: Speech enhance- ment with asynchronous ad-hoc microphone arrays,

    N. Furnon, R. Serizel, S. Essid, and I. Illina, “Samba: Speech enhance- ment with asynchronous ad-hoc microphone arrays,” arXiv preprint arXiv:2307.16582, 2023

    2023 arXiv

  24. [32]

    Graph attention based multi- channel u-net for speech dereverberation with ad-hoc microphone arrays,

    H. Guo, Y . Chen, X.-L. Zhang, and X. Li, “Graph attention based multi- channel u-net for speech dereverberation with ad-hoc microphone arrays,” in Interspeech, 2024

    2024

  25. [33]

    Deepvqe: Real time deep voice quality enhancement for joint acoustic echo cancellation, noise suppression and dereverberation,

    N. C. Ristea, E. Indenbom, A. Saabas, T. P ¨arnamaa, J. Guzhvin, and R. Cutler, “Deepvqe: Real time deep voice quality enhancement for joint acoustic echo cancellation, noise suppression and dereverberation,” in Interspeech, 2023

    2023

  26. [34]

    Towards efficient models for real-time deep noise suppression,

    S. Braun, H. Gamper, C. K. Reddy, and I. Tashev, “Towards efficient models for real-time deep noise suppression,” in ICASSP, 2021

    2021

  27. [35]

    Towards real-time single-channel speech separation in noisy and reverberant environments,

    J. Neri and S. Braun, “Towards real-time single-channel speech separation in noisy and reverberant environments,” in ICASSP, 2023

    2023

  28. [36]

    Icassp 2023 deep noise suppression challenge,

    H. Dubey, A. Aazami, V . Gopal, B. Naderi, S. Braun, R. Cutler, A. Ju, M. Zohourian, M. Tang, M. Golestaneh, and R. Aichner, “Icassp 2023 deep noise suppression challenge,” IEEE Open Journal of Signal Processing, vol. 5, 2024

    2023

  29. [37]

    Image method for efficiently simulating small-room acoustics,

    J. B. Allen and D. A. Berkley, “Image method for efficiently simulating small-room acoustics,” The Journal of the Acoustical Society of America , vol. 65, 1979

    1979

  30. [38]

    Dual-path rnn: efficient long sequence modeling for time-domain single-channel speech separation,

    Y . Luo, Z. Chen, and T. Yoshioka, “Dual-path rnn: efficient long sequence modeling for time-domain single-channel speech separation,” in ICASSP, 2020

    2020

  31. [39]

    Effect of noise suppression losses on speech distortion and ASR performance,

    S. Braun and H. Gamper, “Effect of noise suppression losses on speech distortion and ASR performance,” in ICASSP, 2022

    2022

  32. [40]

    A consolidated view of loss functions for supervised deep learning-based speech enhancement,

    S. Braun and I. Tashev, “A consolidated view of loss functions for supervised deep learning-based speech enhancement,” in 2021 44th International Conference on Telecommunications and Signal Processing (TSP), 2021

    2021

  33. [41]

    Generative pre-training for speech with flow matching,

    A. H. Liu, M. Le, A. Vyas, B. Shi, A. Tjandra, and W.-N. Hsu, “Generative pre-training for speech with flow matching,” in ICLR, 2024

    2024

  34. [42]

    Dnsmos p. 835: A non-intrusive perceptual objective speech quality metric to evaluate noise suppressors,

    C. K. Reddy, V . Gopal, and R. Cutler, “Dnsmos p. 835: A non-intrusive perceptual objective speech quality metric to evaluate noise suppressors,” in ICASSP, 2022

    2022

  35. [43]

    Distillation and pruning for scalable self- supervised representation-based speech quality assessment,

    B. Stahl and H. Gamper, “Distillation and pruning for scalable self- supervised representation-based speech quality assessment,” ICASSP, 2025

    2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.