Pith. sign in

REVIEW 4 major objections 6 minor 43 references

Room Impulse Response as a Prompt for Acoustic Echo Cancellation

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Feeding a measured room impulse response to an acoustic echo canceller as a prompt improves generalization to unseen rooms.

desk verdict Using a measured RIR as a prompt is a genuine idea with clean-RIR gains, but the paper never tests the noisy-RIR scenario it motivates, so the headline result is conditional on an unvalidated denoising step. read the letter →

arxiv 2505.19480 v1 pith:SWE5LWML submitted 2025-05-26 cs.SD eess.AS

Fei Zhao , Shulin He , Xueliang Zhang This is my paper · ORCID
classification cs.SDeess.AS
keywords acousticechocancellationroomimpulseresponseRIRpromptmodelgeneralizationdeeplearningspeechenhancementfusion
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

Data-driven acoustic echo cancellation models trained on synthetic data tend to degrade in real rooms, where the echo path is not directly observable. This paper claims that feeding the model a measured room impulse response (RIR) as an auxiliary prompt restores much of that lost performance. Among four fusion designs, the strongest is simple: denoise the RIR, truncate it, convolve it with the far-end signal to synthesize a 'prompt echo,' and pass that alongside the microphone and far-end inputs. On mismatched synthetic RIRs and real recorded RIRs, the prompt raises PESQ, SDR, and ERLE for both the ICCRN and MTFAA backbones, with a modest increase in computation. If this holds, devices that can re-measure their acoustic environment can keep neural echo cancellers accurate as rooms change.

What carries the argument

The central object is the RIR prompt, a measured room impulse response converted into an auxiliary network input. Fusion method (d), the effective variant, consists of a simple masking-based denoiser, a truncation to 3200 samples, and a convolution of the denoised truncated RIR with the far-end signal to synthesize the prompt echo v'(n). This prompt echo is concatenated with the far-end and microphone signals in the time-frequency domain before the AEC backbone processes them. The mechanism is that v'(n) approximates the true acoustic echo v(n), giving the network a direct, informed target for subtraction rather than forcing it to infer the echo path from the far-end signal alone. The four fusion designs test different ways of packaging the RIR: learned embeddings, temporal averaging, cropped STFT frames, and the convolution-based prompt echo.

What would settle it

Measure AEC quality with method (d) while systematically degrading the measured RIR (add noise, truncate further, use an RIR from a different room, or simulate a moving speaker). If PESQ/SDR on the real-RIR test set do not improve over the no-prompt baseline once the RIR deviates by a modest amount, the central claim that measured RIRs are useful prompts is falsified. Conversely, ablating the denoiser and comparing results with ground-truth clean RIRs would quantify how much of the gain depends on denoising.

Watch

Extended reading notes

Core claim

The paper's central claim is that a room impulse response, used as a training-time and inference-time prompt, improves AEC generalization. Concretely, fusion method (d) preprocesses the measured RIR with a lightweight masking denoiser, truncates it to 3200 samples, and convolves it with the far-end signal x(n) to form a prompt echo v'(n). The AEC network then takes the far-end, microphone, and prompt echo as inputs to predict the near-end signal. In experiments on a 1000-sample real-RIR test set, this raises ICCRN PESQ from 2.05 to 2.19 and SDR from 3.61 to 4.64, and MTFAA SDR from 7.75 to 10.16, while adding less than one Giga-MAC to ICCRN's 1.93 G-MAC baseline. The authors argue the prompt echo lets the network exploit RIR information directly, and that even a noisy RIR, unusable for classical adaptive filters, still helps a learned model.

Load-bearing premise

The method assumes that at inference time the device can measure and denoise its room impulse response accurately enough that convolving it with the far-end signal yields a prompt echo close to the actual echo; if the measured RIR is noisy, outdated, or distorted by moving people or nonlinear speakers, the gains may disappear.

Editorial extensions

If this is right

  • If a device can measure and denoise its own RIR, a neural AEC can be adapted to a new room without retraining on that room's data.
  • Method (d) works across backbones: both ICCRN and MTFAA gain on mismatched and real RIR sets, suggesting the prompt is architecture-agnostic.
  • The gains are largest where generalization is hardest: on real recorded RIRs and on mismatched microphone-loudspeaker distances, the conditions synthetic-only models fail.
  • The added cost is small: ICCRN parameter count grows from 0.463M to 0.611M and MACs from 1.93G to 2.77G with method (d).

Reading between the lines

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

  • A natural extension is to test how robust the gains are to RIR estimation error: as the measured RIR deviates from the true one, the prompt echo diverges from the real echo, and there should be a noise level beyond which method (d) underperforms the no-prompt baseline.
  • The prompt-echo idea may transfer to other room-dependent speech tasks, such as dereverberation or target speaker extraction, where a measured RIR could be convolved with a reference signal to condition the network on the environment.
  • The method assumes a static echo path between pulse measurements; combining it with continuous RIR tracking or online adaptation could cover moving speakers and rotating devices.
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. This paper proposes to improve the generalization of learned acoustic echo cancellation (AEC) systems by supplying the network with a room impulse response (RIR) as an auxiliary "prompt." Four fusion strategies are described: (a) a learned time-frequency embedding, (b) a temporally averaged STFT frame, (c) a cropped STFT prompt, and (d) denoising/truncating the RIR and convolving it with the far-end signal to create a prompt echo v'(n). Experiments with ICCRN and MTFAA backbones on matched synthetic, mismatched synthetic, and recorded real-RIR test sets report consistent gains for fusion (d), with smaller or mixed gains for the other fusions, at a modest increase in computation and parameters.

Significance. If the reported results hold under realistic RIR estimation errors, the work offers a practical, low-cost way to inject environment knowledge into learned AEC systems, and the evaluation across three test sets, including real RIRs, is a strength. The release of code and audio samples is also a positive feature. However, the current evaluation is built from clean, oracle RIRs: in fusion (d) the prompt echo is constructed from the exact RIR used to generate the true echo, so the comparison is leakage-prone. The paper's central real-world premise, namely that the RIR is measured in a noisy environment, is explicitly stated but never tested. The contribution is promising but not yet established.

major comments (4)
  1. [3.2(d), 4.1, Table 1] Fusion (d) constructs v'(n) by convolving the far-end signal with the denoised/truncated RIR. In the experiments of Section 4.1, the same image-method RIR is used to generate both the true echo v(n) and the prompt, so the model is given a near-exact copy of the echo it must cancel. The reported gains therefore do not isolate the benefit of RIR prompting. Please add ablations in which the prompt RIR is corrupted by noise, truncation, delay or gain errors, or is deliberately stale (collected at a different loudspeaker position), and report AEC metrics as a function of RIR error; also compare against a baseline that receives the oracle echo v(n) as an auxiliary input to quantify the ceiling.
  2. [5 (last paragraph), Figure 4] The paper motivates the denoising module by showing a noisy RIR spectrum, but no experiment actually uses noisy or perturbed RIR prompts; all Table 1 results use effectively clean RIRs. The claim that the method works in the noisy-pulse protocol described in Section 3.1 is therefore unsupported. Please report quantitative denoising accuracy (e.g., RIR SNR or echo-path error) and downstream AEC metrics under the noisy-pulse scenario, and describe the denoising module's architecture, training target, and training data if it is learned.
  3. [Table 1, Mismatch Synthetic RIR] For the ICCRN baseline on the Mismatch Synthetic RIR set, fusion methods (a), (b), and (c) do not improve over the baseline (PESQ 1.94/1.95/1.96 vs 1.98; SDR 3.70/4.03/3.84 vs 4.28). Only fusion (d) improves. The general statement that RIR prompting improves generalization is therefore not supported by three of the four variants; the conclusion should be restricted to fusion (d) or accompanied by an error analysis explaining why the other fusions fail on mismatched conditions.
  4. [5, Table 1] No error bars, confidence intervals, or significance tests are reported, and several differences are small relative to typical per-utterance variability in AEC evaluation, e.g., Match Synthetic PESQ 2.96 to 3.03 or Real RIR ERLE 3.73 to 4.79 for ICCRN. Please report per-utterance statistics or results over multiple seeds to substantiate the abstract's claim of significant improvement.
minor comments (6)
  1. [Equation (2)] The definition of the Stretched SI-SNR loss is unclear; the phrase "doubling the period of SISNR" does not explain how Eq. (2) is derived from the standard SI-SNR, and the formula as written does not equal the conventional -10 log10(1 - cos^2(beta)). Please provide a derivation or a precise citation.
  2. [4.2, Figure 2] The text says "For L1 and L3 in Figure 2, we set them as 8000 and 3200 sampling points," but Figure 2 does not label which length corresponds to which panel; please state explicitly which truncation length is used for the convolution prompt in fusion (d).
  3. [3.2(b)] The sentence "the initial frame is discarded" is confusing; please clarify which feature is concatenated with the far-end and microphone signals and what exactly is discarded after processing.
  4. [Table 1] The column layout is difficult to parse; please clarify whether, for each test-RIR type, the three metrics apply to both the DT and ST FE scenarios or whether PESQ/SDR are reported for DT and ERLE for ST FE, and label the columns accordingly.
  5. [Figure 4] The denoised RIR spectrum appears visually different from the clean one; add a quantitative measure, such as spectral distance or RIR SNR, to support the claim that the masking module preserves RIR information.
  6. [Reference [33]] Reference [33] points to a URL for the ICASSP 2023 AEC Challenge; please cite the specific dataset release and version so that the training and test mixtures are precisely identified.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: RIR is an externally measured conditioning input; the untested noisy-RIR deployment scenario is a validity risk, not a circular step.

full rationale

The central claim -- that appending an RIR-derived prompt improves AEC generalization -- does not derive its target from its own output or from a fitted parameter. In fusion (d), the prompt echo is constructed as v'(n) = RIR_d1 * x(n) (Section 3.2, Figure 2d), where RIR_d1 is a denoised, truncated version of an externally measured or simulated room impulse response. The RIR is not estimated from the microphone signal by the trained AEC network, nor is it optimized against the evaluation metrics; it is fixed by the room simulation (Section 4.1, Image method) or by the recorded-RIR database [35]. Thus the improvement cannot be said to reduce, by construction, to the fitted input. The self-citations (ICCRN [26] as baseline, [6] and [31] as background/loss references) are contextual and are not load-bearing for the RIR-prompt claim; removing them would not change the argument. The most serious concern is validity rather than circularity: because the RIR used to synthesize the echo is also provided as the prompt, the model receives near-oracle access to the echo path, and Table 1 measures the benefit of that conditioning. Section 5 asserts that real collection yields noisy RIRs and that denoising is an additional task, but no experiment injects noise or perturbation into the RIR prompt, so the claimed generalization to the paper's own deployment protocol is unverified. That is an omitted validation, not a self-referential derivation. Under the instruction to reserve circularity findings for demonstrated input-output equivalences, this paper scores at the no-significant-circularity end of the scale, with a small allowance for the presence of minor self-citations.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central result is empirical and rests primarily on trained network parameters. The explicit hand-set values listed above are minor; the load-bearing assumption is that an accurate RIR is available at deployment and that the synthesized echo estimate is close to the true echo.

free parameters (3)
  • spectral compression factor p = 0.5
    Hand-set in the RI+Mag loss (Eq. 3-4) following [32]; not fitted to the test sets.
  • RIR prompt length L1 = 8000 samples
    Chosen by hand in Figure 2 for the RIR signals; no sensitivity analysis is provided.
  • RIR prompt length L3 = 3200 samples
    Chosen by hand for truncation in fusion (d); affects the prompt echo quality.
assumptions (4)
  • domain assumption A usable RIR can be measured at inference time by emitting a pulse and capturing it, and the noisy measurement can be denoised well enough.
    Central to the prompt idea; stated in Section 3.1 and Section 5, where noisy and denoised RIR spectra are compared.
  • domain assumption Simulated Image-method RIRs and the Jeub database RIRs are representative of real deployment environments.
    Used to claim generalization in Section 4.1 and Section 5.
  • domain assumption The nonlinear echo model from [15] adequately captures real device nonlinearities.
    Nonlinear settings account for 90% of training data; Section 4.1.
  • ad hoc to paper The S-SISNR loss in Eq. (2) is a valid scale-invariant objective; its formula is assumed correct.
    Presented without derivation in Section 3.3; the trigonometric form is asserted.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Room Impulse Response as a Prompt for Acoustic Echo Cancellation." pith.science (2026). https://pith.science/paper/SWE5LWML

@misc{pith2026250519480,
  author       = {Pith},
  title        = {Pith review of: Room Impulse Response as a Prompt for Acoustic Echo Cancellation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SWE5LWML}},
  note         = {Machine review of arXiv:2505.19480}
}
read the original abstract

Data-driven acoustic echo cancellation (AEC) methods, predominantly trained on synthetic or constrained real-world datasets, encounter performance declines in unseen echo scenarios, especially in real environments where echo paths are not directly observable. Our proposed method counters this limitation by integrating room impulse response (RIR) as a pivotal training prompt, aiming to improve the generalization of AEC models in such unforeseen conditions. We also explore four RIR prompt fusion methods. Comprehensive evaluations, including both simulated RIR under unknown conditions and recorded RIR in real, demonstrate that the proposed approach significantly improves performance compared to baseline models. These results substantiate the effectiveness of our RIR-guided approach in strengthening the model's generalization capabilities.

Figures

Figures reproduced from arXiv: 2505.19480 by the authors.

Figure 1
Figure 1. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Four different fusion methods for RIR prompt [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of proposed method. posed method. The experimental setup is shown in Section 4. In Section 5, we demonstrate the performance of the proposed system. We conclude the paper in Section 6. 2. Problem Formulation The single-channel AEC is shown in [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Spectrum of clean RIR, noisy RIR, and denoised RIR. eter count of the MTFAA model, which potentially allows for a more nuanced capture of the acoustic characteristics present in the mismatched test set. In [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 33 canonical work pages

  1. [1]

    Room Impulse Response as a Prompt for Acoustic Echo Cancellation

    Introduction Acoustic Echo Cancellation (AEC) is a critical research area in speech signal processing, playing a vital role in hands-free communication systems [1–3]. The primary goal of AEC is to eliminate echo from the microphone signal using a far-end reference signal [4–6]. Traditional approaches predominantly rely on adaptive filtering [7–10] to mode...

    work page Pith review arXiv 2025

  2. [2]

    Problem Formulation The single-channel AEC is shown in Figure 1, the system typ- ically has access to two critical input signals: the near-end mi- crophone signal y(n) and the far-end signal x(n), n represents the time sampling point. The formulation of the microphone signal in the time domain is presented as follows: y(n) =s(n) +v(n) +d(n) (1) where the ...

    work page

  3. [3]

    Overview In this section, we provide an overview of our proposed method, as illustrated in Figure 3

    Proposed Method 3.1. Overview In this section, we provide an overview of our proposed method, as illustrated in Figure 3. The network incorporates the noisy RIR signal (denoted as RIR n) as a complementary input to the AEC model. The RIR data is first preprocessed by a dedicated fusion module and then merged with the regular input of the AEC model for fur...

    work page

  4. [4]

    Experimental Setups 4.1. Datasets The near-end and far-end signals employed in our experiments are sourced from the ICASSP 2023 AEC challenge’s synthetic datasets, representing near-end and far-end scenarios, respec- tively [33]. The RIR is generated using the Image method [34]. We simulate different rooms of size l × w × h m3 for train- ing mixtures, whe...

    work page 2023

  5. [5]

    Higher scores indicate better performance

    Experimental Results To evaluate the performance of the proposed method, we uti- lize the echo return loss enhancement (ERLE) [37], the per- ceptual evaluation of speech quality (PESQ) [3] and signal-to- distortion ratio (SDR) [38] as the metrics that measure the echo suppression for single-talk, near-end speech quality and near- end speech fidelity for d...

    work page

  6. [6]

    We investigate four distinct methods for integrating RIR prompts

    Conclusions In this paper, we propose a novel approach that utilizes room impulse response (RIR) as a prompt to enhance the general- ization of the acoustic echo cancellation (AEC) model. We investigate four distinct methods for integrating RIR prompts. The effectiveness of our approach is validated across three test sets: synthetic RIR with matching cond...

    work page

  7. [7]

    An adaptive echo canceller,

    M. Sondhi, “An adaptive echo canceller,” Bell System technical journal, vol. 46, no. 3, pp. 497–511, 1967

    work page 1967

  8. [8]

    Advances in network and acoustic echo cancellation,

    J. Benesty, T. G ¨ansler, D. R. Morgan, M. M. Sondhi, S. L. Gay et al. , “Advances in network and acoustic echo cancellation,” 2001

    2001

Show all 43 references
  1. [9]

    Acoustic echo control,

    G. Enzner, H. Buchner, A. Favrot, and F. Kuech, “Acoustic echo control,” in Academic press library in signal processing . Else- vier, 2014, vol. 4, pp. 807–877

    2014

  2. [10]

    A variable step-size affine projection algorithm designed for acoustic echo cancella- tion,

    C. Paleologu, J. Benesty, and S. Ciochina, “A variable step-size affine projection algorithm designed for acoustic echo cancella- tion,” IEEE Transactions on Audio, Speech, and Language Pro- cessing, vol. 16, no. 8, pp. 1466–1478, 2008

    2008

  3. [12]

    Deep echo path modeling for acoustic echo cancellation

    F. Zhao, C. Zhang, S. He, J. Liu, and X. Zhang, “Deep echo path modeling for acoustic echo cancellation.” International Speech Communication Association, 2024

    2024

  4. [13]

    An overview on optimized nlms algorithms for acoustic echo cancel- lation,

    C. Paleologu, S. Ciochin ˘a, J. Benesty, and S. L. Grant, “An overview on optimized nlms algorithms for acoustic echo cancel- lation,” EURASIP Journal on Advances in Signal Processing, vol. 2015, pp. 1–19, 2015

    2015

  5. [14]

    S. S. Haykin, Adaptive filter theory . Pearson Education India, 2005

    2005

  6. [15]

    Analysis of the frequency domain adaptive filter,

    N. Bershad and P. Feintuch, “Analysis of the frequency domain adaptive filter,” vol. 67, no. 12, pp. 1658–1659, 1979

    1979

  7. [16]

    State-space architecture of the partitioned-block-based acoustic echo controller,

    F. Kuech, E. Mabande, and G. Enzner, “State-space architecture of the partitioned-block-based acoustic echo controller,” 2014, pp. 1295–1299

    2014

  8. [17]

    Nonlinear acoustic echo cancellation based on volterra filters,

    A. Gu ´erin, G. Faucon, and R. Le Bouquin-Jeann `es, “Nonlinear acoustic echo cancellation based on volterra filters,” IEEE Trans- actions on Speech and Audio Processing, vol. 11, no. 6, pp. 672– 683, 2004

    2004

  9. [18]

    A neural network-based nonlinear acoustic echo canceller,

    M. M. Halimeh, C. Huemmer, and W. Kellermann, “A neural network-based nonlinear acoustic echo canceller,” IEEE Signal Processing Letters, vol. 26, no. 12, pp. 1827–1831, 2019

    2019

  10. [19]

    Nonlinear acoustic echo cancellation using low- complexity low-rank recursive least-squares algorithms,

    V . Patel, S. S. Bhattacharjee, J. R. Jensen, M. G. Christensen, and J. Benesty, “Nonlinear acoustic echo cancellation using low- complexity low-rank recursive least-squares algorithms,” Signal Processing, vol. 225, p. 109623, 2024

    2024

  11. [20]

    Nonlinear acoustic echo cancellation based on pipelined hermite filters,

    K.-L. Yin, M. M. Halimeh, Y .-F. Pu, L. Lu, and W. Kellermann, “Nonlinear acoustic echo cancellation based on pipelined hermite filters,” Signal Processing, vol. 220, p. 109470, 2024

    2024

  12. [21]

    A complex spectral mapping with inplace convolution recurrent neural networks for acoustic echo cancellation,

    C. Zhang, J. Liu, and X. Zhang, “A complex spectral mapping with inplace convolution recurrent neural networks for acoustic echo cancellation,” in ICASSP, 2022, pp. 751–755

    2022

  13. [22]

    Neural multi-channel and multi-microphone acoustic echo cancellation,

    C. Zhang, J. Liu, H. Li, and X. Zhang, “Neural multi-channel and multi-microphone acoustic echo cancellation,” IEEE ACM Trans. Audio Speech Lang. Process., vol. 31, pp. 2181–2192, 2023

    2023

  14. [23]

    Multi-scale temporal fre- quency convolutional network with axial attention for speech en- hancement,

    G. Zhang, L. Yu, C. Wang, and J. Wei, “Multi-scale temporal fre- quency convolutional network with axial attention for speech en- hancement,” in ICASSP, 2022, pp. 9122–9126

    2022

  15. [24]

    A neural acoustic echo canceller optimized using an automatic speech recognizer and large scale synthetic data,

    N. Howard, A. Park, T. Z. Shabestary, A. Gruenstein, and R. Prab- havalkar, “A neural acoustic echo canceller optimized using an automatic speech recognizer and large scale synthetic data,” in ICASSP, 2021, pp. 7128–7132

    2021

  16. [25]

    A deep hierarchical fusion network for fullband acoustic echo cancellation,

    H. Zhao, N. Li, R. Han, L. Chen, X. Zheng, C. Zhang, L. Guo, and B. Yu, “A deep hierarchical fusion network for fullband acoustic echo cancellation,” in ICASSP, 2022, pp. 9112–9116

    2022

  17. [26]

    A conformer- based waveform-domain neural acoustic echo canceller optimized for asr accuracy

    S. Panchapagesan, A. Narayanan, T. Z. Shabestary, S. Shao, N. Howard, A. Park, J. Walker, and A. Gruenstein, “A conformer- based waveform-domain neural acoustic echo canceller optimized for asr accuracy.” International Speech Communication Associ- ation, 2022, pp. 2538–2542

    2022

  18. [27]

    Icassp 2021 acous- tic echo cancellation challenge: Datasets, testing framework, and results,

    K. Sridhar, R. Cutler, A. Saabas, T. Parnamaa, M. Loide, H. Gam- per, S. Braun, R. Aichner, and S. Srinivasan, “Icassp 2021 acous- tic echo cancellation challenge: Datasets, testing framework, and results,” in ICASSP. IEEE, 2021, pp. 151–155

    2021

  19. [28]

    Conversational artificial intelligence in the aec industry: A review of present status, challenges and oppor- tunities,

    A. B. Saka, L. O. Oyedele, L. A. Akanbi, S. A. Ganiyu, D. W. Chan, and S. A. Bello, “Conversational artificial intelligence in the aec industry: A review of present status, challenges and oppor- tunities,” Advanced Engineering Informatics, vol. 55, p. 101869, 2023

    2023

  20. [29]

    Kalmannet: Data-driven kalman filtering,

    G. Revach, N. Shlezinger, R. J. Van Sloun, and Y . C. Eldar, “Kalmannet: Data-driven kalman filtering,” in ICASSP, 2021, pp. 3905–3909

    2021

  21. [30]

    Low- complexity acoustic echo cancellation with neural kalman filter- ing,

    D. Yang, F. Jiang, W. Wu, X. Fang, and M. Cao, “Low- complexity acoustic echo cancellation with neural kalman filter- ing,” in ICASSP, 2023, pp. 1–5

    2023

  22. [31]

    Kalmannet: A learnable kalman filter for acoustic echo cancellation,

    Y . Zhang, M. Yu, H. Zhang, D. Yu, and D. Wang, “Kalmannet: A learnable kalman filter for acoustic echo cancellation,”CoRR, vol. abs/2301.12363, 2023

    2023 arXiv

  23. [32]

    Iccrn: Inplace cepstral convolutional re- current neural network for monaural speech enhancement,

    J. Liu and X. Zhang, “Iccrn: Inplace cepstral convolutional re- current neural network for monaural speech enhancement,” in ICASSP, 2023, pp. 1–5

    2023

  24. [33]

    Room impulse response generator,

    E. A. Habets, “Room impulse response generator,” Technische Universiteit Eindhoven, Tech. Rep, vol. 2, no. 2.4, p. 1, 2006

    2006

  25. [34]

    Funnel deep complex u- net for phase-aware speech enhancement,

    Y . Sun, L. Yang, H. Zhu, and J. Hao, “Funnel deep complex u- net for phase-aware speech enhancement,” in Interspeech 2021, H. Hermansky, H. Cernock ´y, L. Burget, L. Lamel, O. Scharen- borg, and P. Motl´ıcek, Eds. ISCA, 2021, pp. 161–165

    2021

  26. [35]

    Conv-tasnet: Surpassing ideal time- frequency magnitude masking for speech separation,

    Y . Luo and N. Mesgarani, “Conv-tasnet: Surpassing ideal time- frequency magnitude masking for speech separation,”IEEE ACM Trans. Audio Speech Lang. Process. , vol. 27, no. 8, pp. 1256– 1266, 2019

    2019

  27. [36]

    Hierarchical speaker representation for target speaker extrac- tion,

    S. He, H. Zhang, W. Rao, K. Zhang, Y . Ju, Y . Yang, and X. Zhang, “Hierarchical speaker representation for target speaker extrac- tion,” in ICASSP, 2024, pp. 10 361–10 365

    2024

  28. [37]

    Attention-enhanced short-time wiener solution for acoustic echo cancellation,

    F. Zhao and X. Zhang, “Attention-enhanced short-time wiener solution for acoustic echo cancellation,” arXiv preprint arXiv:2412.18851, 2024

    2024 arXiv

  29. [38]

    On the importance of power compression and phase estimation in monaural speech dereverber- ation,

    A. Li, C. Zheng, R. Peng, and X. Li, “On the importance of power compression and phase estimation in monaural speech dereverber- ation,” JASA express letters, vol. 1, no. 1, 2021

    2021

  30. [39]

    (2023) icassp acoustic echo can- cellation challenge

    Microsoft., “(2023) icassp acoustic echo can- cellation challenge.” [Online]. Available: https: //www.microsoft.com/en-us/research/academic-program/ acoustic-echo-cancellation-challenge-icassp-2023/

    2023

  31. [40]

    Image method for efficiently sim- ulating small-room acoustics,

    J. B. Allen and D. A. Berkley, “Image method for efficiently sim- ulating small-room acoustics,” The Journal of the Acoustical So- ciety of America, vol. 65, no. 4, pp. 943–950, 1979

    1979

  32. [41]

    A binaural room impulse re- sponse database for the evaluation of dereverberation algorithms,

    M. Jeub, M. Sch ¨afer, and P. Vary, “A binaural room impulse re- sponse database for the evaluation of dereverberation algorithms,” in 16th International Conference on Digital Signal Processing, DSP 2009. IEEE, 2009, pp. 1–5

    2009

  33. [42]

    Adam: A method for stochastic op- timization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic op- timization,” in 3rd International Conference on Learning Repre- sentations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Con- ference Track Proceedings, Y . Bengio and Y . LeCun, Eds., 2015

    2015

  34. [43]

    Perceptual evaluation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs,

    A. W. Rix, J. G. Beerends, M. P. Hollier, and A. P. Hekstra, “Perceptual evaluation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs,” in ICASSP, vol. 2, 2001, pp. 749–752

    2001

  35. [44]

    Performance measure- ment in blind audio source separation,

    E. Vincent, R. Gribonval, and C. F ´evotte, “Performance measure- ment in blind audio source separation,” IEEE transactions on au- dio, speech, and language processing , vol. 14, no. 4, pp. 1462– 1469, 2006

    2006

Pith tools

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