Pith. sign in

REVIEW 3 major objections 5 minor 35 references

Mel-McNet: A Mel-Scale Framework for Online Multichannel Speech Enhancement

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Mel-McNet claims that moving multichannel enhancement into the Mel-frequency domain cuts compute by nearly 60% while matching the original McNet's quality and ASR accuracy.

desk verdict Efficiency claim is solid and structurally supported, but the SOTA performance claim is confounded by the vocoder and ASR feature asymmetry; worth engaging for the FLOPs result, needs a control experiment. read the letter →

arxiv 2505.19576 v1 pith:WXYAQKTM submitted 2025-05-26 eess.AS cs.SDeess.SP

Yujie Yang , Bing Yang , Xiaofei Li This is my paper · ORCID
classification eess.AScs.SDeess.SP
keywords multichannelspeechenhancementMel-scaleprocessingonlineMcNetbackboneSTFT-to-MelcompressionMelpowerratiomaskneuralvocoderCHiME-3
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

Mel-McNet is a framework for online multichannel speech enhancement that does almost all of its work on a Mel-frequency scale instead of the usual linear-frequency STFT scale. The paper's claim is that a compact STFT-to-Mel module plus a modified McNet backbone can generate enhanced LogMel spectrograms directly, and that these can be handed to a vocoder for waveform reconstruction or to an ASR system for transcription. On the simulated CHiME-3 benchmark the framework is reported to cut computational complexity by nearly 60% relative to the McNet backbone while keeping enhancement quality and word error rate comparable, and to beat other online multichannel methods on DNSMOS and WER. If true, this makes perceptually motivated frequency compression a viable design choice for real-time, low-complexity multichannel enhancement on edge devices.

What carries the argument

The load-bearing object is the STFT-to-Mel module: two parallel branches, one for multichannel magnitude and one for phase, each built from a linear projection, frequency compression blocks made of along-frequency 1-D convolutions and handcrafted or learnable Mel filterbanks, and a causal time-smoothing convolution. An attention-like information-communication operation lets the phase branch modulate the magnitude branch, and the compressed features are stacked and smoothed to produce a T frames by 80 Mel bands by D embedding for the McNet backbone. The training target is the rectified Mel power ratio mask, Mel-PRM, which clips the ratio of clean to noisy Mel power at 1. The module's role is to compress the frequency axis before the expensive along-frequency and along-time LSTM processing in McNet, which is where the nearly 60% complexity reduction comes from.

What would settle it

Run the McNet backbone through the same pipeline as Mel-McNet: convert McNet's enhanced STFT output to LogMel, reconstruct with the same Vocos vocoder, and evaluate DNSMOS and WER on CHiME-3. If this McNet plus Vocos baseline matches Mel-McNet's numbers, the 60% complexity reduction would be confirmed but the quality and ASR claim would not be due to Mel-scale processing.

Watch

Extended reading notes

Core claim

The central claim is that the nonlinear Mel scale can carry the full multichannel enhancement computation without a performance penalty. The paper constructs Mel-McNet by inserting an STFT-to-Mel module before a modified McNet backbone, so that both spectral cues (magnitude) and spatial cues (phase) are compressed from F linear frequencies to 80 Mel bands before the four McNet modules process them. The network is trained to predict a rectified Mel power ratio mask (Mel-PRM), and at inference the enhanced LogMel power spectrogram is obtained by masking the noisy power and taking the logarithm. On the CHiME-3 evaluation the paper reports a 60% FLOPs reduction, a real-time factor of 0.799 on CPU, and DNSMOS/WER results that match or exceed the linear-frequency baselines; it attributes the small drops in WB-PESQ and STOI to the neural vocoder, not to the Mel-scale processing.

Load-bearing premise

The load-bearing premise is that comparing Mel-McNet, whose enhanced LogMel must be turned into a waveform by a neural vocoder, against baselines that produce waveforms directly is a fair test of Mel-scale processing; if the vocoder itself is responsible for the observed DNSMOS and WER behavior, the claimed benefit of Mel-scale processing is not cleanly isolated.

Editorial extensions

If this is right

  • If Mel-McNet works as reported, multichannel speech enhancement can be run in the Mel domain with a roughly 60% reduction in FLOPs and no loss in perceived quality or ASR accuracy.
  • The enhanced LogMel output can be consumed directly by an ASR model, so the enhancement front-end and recognition backend can share the same feature representation.
  • The CPU real-time factor of about 0.799 (0.755 for the network, 0.044 for the vocoder) suggests the framework is deployable in real time without a GPU.
  • Mel-scale compression appears to preserve spatial information well enough for the narrow-band and full-band spatial modules of McNet to still separate target speech from noise.

Reading between the lines

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

  • The reported gains are not yet isolated from the vocoder: since Mel-McNet outputs LogMel and reconstructs waveforms with Vocos while the linear-frequency baselines output waveforms directly, a McNet plus the same Vocos baseline would be needed to attribute the DNSMOS and WER results specifically to Mel-scale processing.
  • Because the CHiME-3 simulated training data is reverberation-free (time delay only), the framework's behavior under reverberation or with moving speakers is an open extension, not yet demonstrated.
  • The STFT-to-Mel front-end is modular and could be attached to other sub-band/full-band enhancement backbones; the paper names this as future work, and a direct test would be to replace only the input representation of an existing model.
  • The WER gains on CHiME-3 may partly reflect that ASR systems are trained on LogMel features, so enhancement in the same feature space avoids an extra linear-to-Mel conversion and its associated mismatch.
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 / 5 minor

Summary. The paper proposes Mel-McNet, an online multichannel speech enhancement framework that compresses multichannel STFT features into an 80-band Mel-scale representation via a new STFT-to-Mel module and then processes spectral and spatial information with a modified McNet backbone operating directly in the Mel domain. The enhanced LogMel spectrogram can be fed either to the Vocos neural vocoder for waveform reconstruction or to a pretrained ASR system for transcription. Experiments on the CHiME-3 dataset report a roughly 60% reduction in FLOPs relative to the McNet backbone while maintaining comparable enhancement and ASR performance, and the paper claims to outperform other SOTA online methods. An ablation study compares alternative compression schemes for the STFT-to-Mel module.

Significance. If the efficiency result holds, the paper makes a useful contribution by demonstrating that Mel-scale processing can be applied to multichannel spatial information without a full-band recovery step, and the released code plus the use of an external FLOPs counter are strengths. The 60% FLOPs reduction is structurally supported by the compression from F=257 to F'=80 frequency bins and is robust even when the vocoder's FLOPs are included. However, the secondary claim of outperforming SOTA methods is weakened by a confounded comparison: Mel-McNet uses a neural vocoder while all baselines output waveforms directly, and the ASR evaluation is asymmetric. The paper's value would be significantly strengthened by adding a McNet+Vocos control and by reporting ASR results on a common feature-extraction path.

major comments (3)
  1. [§3.2, Table 1] The claim that Mel-McNet 'outperforms other SOTA online methods' is not cleanly supported because the comparison changes two variables at once: the enhancement backbone and the output backend. Mel-McNet outputs an 80-band LogMel spectrogram that is converted to a waveform by the pretrained Vocos, whereas EaBNet, oSpatialnet-mamba, and McNet output time-domain waveforms directly. The 'Clean Mel+Vocos' row in Table 1 shows that Vocos is not a neutral transform for STOI (100.0 to 97.7) and WB-PESQ (4.06 to 4.05), even though DNSMOS is preserved. To attribute the DNSMOS improvement (3.43 vs 3.40) and the WER result (4.4) to Mel-scale processing, please add a 'McNet+Vocos' control: take McNet's enhanced waveform, extract its LogMel spectrogram, reconstruct with Vocos, and evaluate all metrics. Without this control, the observed advantages could be produced by the vocoder rather than by the STFT-to-Mel compression or the modified backbone.
  2. [§2.2 and §3.2, Table 1] The ASR evaluation is asymmetric. Mel-McNet's LogMel output is fed directly to the pretrained E-branchformer ASR, while all baselines must first synthesize a time-domain waveform and then extract LogMel features from that waveform. This direct path avoids possible mismatches in feature extraction and may favor Mel-McNet independent of enhancement quality. Please report a control in which all baselines' enhanced waveforms are converted to LogMel using the same feature pipeline and fed to the same ASR, or explicitly discuss this confound when interpreting the WER comparison.
  3. [§3.2, Table 1] The 'comparable performance' claim is not fully supported by STOI. Mel-McNet's STOI is 95.4% versus McNet's 97.7%, a 2.3-point drop, while WB-PESQ is tied at 3.99. The authors attribute the lower STOI to the vocoder, but the 'Clean Mel+Vocos' row shows that Vocos on clean LogMel yields 97.7% STOI, only 2.3 points below clean speech; it does not explain the additional 2.3-point drop to 95.4% in Mel-McNet. Please provide a decomposition (e.g., evaluate the enhanced LogMel before vocoding with a Mel-domain reference, or compare McNet+Vocos and Mel-McNet+Vocos on identical pipelines), or restrict the 'comparable' claim to the non-intrusive metrics and WER.
minor comments (5)
  1. [§1] The phrasing 'few works are implemented in a Mel-frequency domain' is awkward; consider 'few works operate in the Mel-frequency domain' or a similar wording.
  2. [§2.1.1, Eq. (1)] The notation in Eq. (1) uses the index i both for the block index in 'EMag_i' and 'EPha_i' and for the element-wise multiplication symbol; clarify the range of i (i = 1, ..., Q) and define the linear(·) layer's input/output dimensions.
  3. [§3.1] Please justify the choice of the fifth microphone as the reference channel, since the choice of reference can affect the reported enhancement and ASR metrics.
  4. [Table 1 caption] The caption states that 'Param. and FLOPs of speech enhancement network/neural vocoder are given in addition' for Mel-McNet, but the table does not explicitly separate the two in the caption text; adding a note such as '(enhancement network + vocoder)' would improve readability.
  5. [§3.2] The statement that EaBNet 'has lower computational complexity' while the ASR performance is 'heavily impacted' is presented without noting that EaBNet also has a different parameter count and architecture; the sentence would benefit from a brief context about the comparison being across different designs.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FLOPs reduction is externally measured and performance claims rest on retrained baselines, not on fitted constants or load-bearing self-citations.

full rationale

No circular derivation is present. The paper's central claim is that inserting an STFT-to-Mel module before the authors' prior McNet backbone reduces computational cost by about 60% while preserving enhancement/ASR performance. This is supported by an external FLOPs counter (torch.utils.flop_counter.FlopCounterMode) applied to the actual networks (43.75 vs 115.13 G/s), so the efficiency claim does not reduce to a fitted constant or to a definition. The enhancement and ASR results are evaluated against retrained external baselines (EaBNet, oSpatialnet-mamba) and a retrained McNet baseline on the CHiME-3 evaluation set, with WER from a pretrained ESPnet E-branchformer; none of the reported metrics are recomputed from the training loss or from parameters fitted to the test set. Self-citations to McNet [8] and to Mel-fullsubnet [12] are architecture reuse and methodological borrowing, but they are not load-bearing in the sense of the present paper relying on those papers' unpublished numbers: McNet is retrained as a baseline, and the vocoder choice is an independent external module. The vocoder-induced asymmetry in Table 1 (Mel-McNet outputs LogMel reconstructed by Vocos while baselines output waveforms) is a real evaluation-validity concern for the 'outperforms SOTA' wording, but it is not circularity: the vocoder is a fixed external component, and the paper's own 'clean Mel+Vocos' upper-bound row makes the confound explicit rather than concealing it. This is a fairness/interpretation issue, not an equivalence-by-construction issue. The derivation chain from STFT input, through the STFT-to-Mel module and McNet backbone, to the Mel-PRM mask and enhanced LogMel is self-contained and does not presuppose the conclusion.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper does not fit any physical constants or derive laws from first principles; the central claim is architectural. Free parameters are the standard hyperparameters of the model, all hand-chosen. The main assumptions are about the safety of Mel-scale compression for spatial cues, the validity of the mask/vocoder pipeline, and the representativeness of the simulated dataset.

free parameters (5)
  • Number of Mel bins F' = 80
    Hand-chosen; determines the compressed frequency resolution and directly drives the 60% FLOPs reduction. Not tuned by a validation search.
  • Hidden embedding dimension D = 64
    Hand-chosen; matches the original McNet configuration and affects model capacity.
  • Number of F-Conv1d blocks Q = 3
    Hand-chosen; Q frequency convolution blocks with kernel size 3 are used in the STFT-to-Mel module.
  • T-Conv1d kernel size = 6
    Hand-chosen; a temporally causal convolution with kernel 6 and 5 zero-padded past frames smooths the time axis.
  • STFT window and hop = 512/128 samples (32ms/8ms)
    Hand-chosen; shared with baselines for fair comparison, though it also affects the Mel filterbank mapping.
assumptions (4)
  • domain assumption Mel-scale compression preserves enough spatial information for multichannel enhancement
    Compressing from 257 linear-STFT bins to 80 Mel bins discards high-frequency resolution; inter-channel phase differences scale linearly with frequency, so this compression could remove spatial cues. The paper addresses this with a separate phase branch using learnable filterbanks, but the adequacy is only shown empirically.
  • domain assumption Power ratio mask (PRM) computed from clean and noisy Mel power is an appropriate training target
    The network predicts min(sqrt(SMel/XMel),1) with MSE loss; this is a standard masking assumption but is stated as the chosen target without derivation.
  • domain assumption The neural vocoder (Vocos) reconstructs time-domain speech from enhanced LogMel without introducing distortion that changes conclusions
    The paper shows 'clean Mel+Vocos' has lower WB-PESQ and STOI than clean speech, indicating the vocoder is not transparent; the comparison of Mel-McNet to baselines that do not use a vocoder relies on the vocoder's distortion being acceptable.
  • domain assumption The evaluation on simulated CHiME-3 reflects real-world performance
    Only the simulated CHiME-3 eval/test sets are used; no real recorded multichannel noise or moving speakers are tested.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mel-McNet: A Mel-Scale Framework for Online Multichannel Speech Enhancement." pith.science (2026). https://pith.science/paper/WXYAQKTM

@misc{pith2026250519576,
  author       = {Pith},
  title        = {Pith review of: Mel-McNet: A Mel-Scale Framework for Online Multichannel Speech Enhancement},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WXYAQKTM}},
  note         = {Machine review of arXiv:2505.19576}
}
read the original abstract

Online multichannel speech enhancement has been intensively studied recently. Though Mel-scale frequency is more matched with human auditory perception and computationally efficient than linear frequency, few works are implemented in a Mel-frequency domain. To this end, this work proposes a Mel-scale framework (namely Mel-McNet). It processes spectral and spatial information with two key components: an effective STFT-to-Mel module compressing multi-channel STFT features into Mel-frequency representations, and a modified McNet backbone directly operating in the Mel domain to generate enhanced LogMel spectra. The spectra can be directly fed to vocoders for waveform reconstruction or ASR systems for transcription. Experiments on CHiME-3 show that Mel-McNet can reduce computational complexity by 60% while maintaining comparable enhancement and ASR performance to the original McNet. Mel-McNet also outperforms other SOTA methods, verifying the potential of Mel-scale speech enhancement.

Figures

Figures reproduced from arXiv: 2505.19576 by the authors.

Figure 1
Figure 1. The proposed Mel-scale framework. (a) The system overview. (b) The diagram of Mel-McNet. (c) The architecture of STFT-to-Mel module. F, F ′ , T, and M represent the number of linear-scale STFT frequencies, nonlinear Mel-scale frequencies, time frames, and microphones, respectively. N1 and N2 are the number of adjacent frequencies. C is the number of context frames. D is the dimension of hidden embeddings. For the Mc… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 30 canonical work pages

  1. [1]

    Mel-McNet: A Mel-Scale Framework for Online Multichannel Speech Enhancement

    Introduction Speech enhancement aims to separate target speech from back- ground noise. It has gained increasing importance in emerg- ing smart devices, such as smart homes and embodied intelli- gence systems. These applications pose high requirements on not only real-time implementation and low computational com- plexity for deployment on edge devices, b...

    work page Pith review arXiv 2025

  2. [2]

    Method The proposed Mel-scale framework for multichannel speech en- hancement is illustrated in Fig.1 (a). The noisy speech is fed into Mel-McNet to obtain the enhanced LogMel spectrogram, which is followed by a vocoder to reconstruct the speech signal in the time domain and/or an ASR model for text transcription. 2.1. Mel-McNet As depicted in Fig.1(b), t...

    work page

  3. [3]

    Experiments 3.1. Experimental Setup Dataset: We evaluate both speech enhancement and recognition performance on the official evaluation and test sets (compris- ing 1,640 and 1,320 utterances, respectively) of the simulated CHiME-3 dataset [20]. The training set of speech enhancement model is con- structed by combining clean speech signals from the DNS Cha...

    work page

  4. [4]

    With an effective STFT-to- Mel module inserted before the McNet backbone, Mel-McNet can process both spectral and spatial information in the Mel domain

    Conclusion This work proposes a Mel-scale framework for online mul- tichannel speech enhancement. With an effective STFT-to- Mel module inserted before the McNet backbone, Mel-McNet can process both spectral and spatial information in the Mel domain. The enhanced LogMel spectrogram output by Mel- McNet can be passed to a vocoder/ASR model for reconstruct-...

    work page

  5. [5]

    Tf-gridnet: Integrating full- and sub-band modeling for speech separation,

    Z.-Q. Wang, S. Cornell, S. Choi, Y . Lee, B.-Y . Kim, and S. Watan- abe, “Tf-gridnet: Integrating full- and sub-band modeling for speech separation,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 31, pp. 3221–3236, 2023

    work page 2023

  6. [6]

    Supervised speech separation based on deep learning: An overview,

    D. Wang and J. Chen, “Supervised speech separation based on deep learning: An overview,” IEEE/ACM Transactions on Au- dio, Speech, and Language Processing, vol. 26, no. 10, pp. 1702– 1726, 2018

    2018

  7. [7]

    Unsupervised speech enhancement based on mul- tichannel nmf-informed beamforming for noise-robust automatic speech recognition,

    K. Shimada, Y . Bando, M. Mimura, K. Itoyama, K. Yoshii, and T. Kawahara, “Unsupervised speech enhancement based on mul- tichannel nmf-informed beamforming for noise-robust automatic speech recognition,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 27, no. 5, pp. 960–971, 2019

    work page 2019

  8. [8]

    Blstm supported gev beamformer front-end for the 3rd chime challenge,

    J. Heymann, L. Drude, A. Chinaev, and R. Haeb-Umbach, “Blstm supported gev beamformer front-end for the 3rd chime challenge,” in 2015 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU), 2015, pp. 444–451

    work page 2015

Show all 35 references
  1. [9]

    Spatialnet: Extensively learning spatial in- formation for multichannel joint speech separation, denoising and dereverberation,

    C. Quan and X. Li, “Spatialnet: Extensively learning spatial in- formation for multichannel joint speech separation, denoising and dereverberation,” IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 32, pp. 1310–1323, 2024

    2024

  2. [10]

    Tparn: Triple-path attentive recurrent network for time-domain multichannel speech enhancement,

    A. Pandey, B. Xu, A. Kumar, J. Donley, P. Calamia, and D. Wang, “Tparn: Triple-path attentive recurrent network for time-domain multichannel speech enhancement,” inICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Pro- cessing (ICASSP), 2022, pp....

    2022

  3. [11]

    Insights into deep non-linear filters for improved multi-channel speech enhancement,

    K. Tesch and T. Gerkmann, “Insights into deep non-linear filters for improved multi-channel speech enhancement,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 31, pp. 563–575, 2023

    2023

  4. [12]

    Multichannel long-term streaming neural speech enhancement for static and moving speakers,

    C. Quan and X. Li, “Multichannel long-term streaming neural speech enhancement for static and moving speakers,”IEEE Signal Processing Letters, vol. 31, pp. 2295–2299, 2024

    2024

  5. [13]

    Mcnet: Fuse multiple cues for multichannel speech enhancement,

    Y . Yang, C. Quan, and X. Li, “Mcnet: Fuse multiple cues for multichannel speech enhancement,” inICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Pro- cessing (ICASSP), 2023, pp. 1–5

    2023

  6. [14]

    Fas- net: Low-latency adaptive beamforming for multi-microphone au- dio processing,

    Y . Luo, C. Han, N. Mesgarani, E. Ceolini, and S.-C. Liu, “Fas- net: Low-latency adaptive beamforming for multi-microphone au- dio processing,” in2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2019, pp. 260–267

    2019

  7. [15]

    Embedding and beamform- ing: All-neural causal beamformer for multichannel speech en- hancement,

    A. Li, W. Liu, C. Zheng, and X. Li, “Embedding and beamform- ing: All-neural causal beamformer for multichannel speech en- hancement,” in ICASSP 2022 - 2022 IEEE International Confer- ence on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 6487–6491

    2022

  8. [16]

    Multichannel speech enhancement without beamforming,

    ——, “Multichannel speech enhancement without beamforming,” in ICASSP 2022 - 2022 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP) , 2022, pp. 6502– 6506

    2022

  9. [17]

    Mel-fullsubnet: Mel- spectrogram enhancement for improving both speech quality and asr,

    R. Zhou, X. Li, Y . Fang, and X. Li, “Mel-fullsubnet: Mel- spectrogram enhancement for improving both speech quality and asr,” arXiv preprint arXiv:2402.13511, 2024

    2024 arXiv

  10. [18]

    Deep neural mel-subband beamformer for in-car speech separation,

    V . Kothapally, Y . Xu, M. Yu, S.-X. Zhang, and D. Yu, “Deep neural mel-subband beamformer for in-car speech separation,” in ICASSP 2023 - 2023 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP), 2023, pp. 1–5

    2023

  11. [19]

    A multi-phase gammatone filterbank for speech separation via tasnet,

    D. Ditter and T. Gerkmann, “A multi-phase gammatone filterbank for speech separation via tasnet,” in ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Pro- cessing (ICASSP), 2020, pp. 36–40

    2020

  12. [20]

    The third ‘chime’ speech separation and recognition challenge: Dataset, task and baselines,

    J. Barker, R. Marxer, E. Vincent, and S. Watanabe, “The third ‘chime’ speech separation and recognition challenge: Dataset, task and baselines,” in ASRU, 2015, pp. 504–511

    2015

  13. [21]

    Phasen: A phase-and- harmonics-aware speech enhancement network,

    D. Yin, C. Luo, Z. Xiong, and W. Zeng, “Phasen: A phase-and- harmonics-aware speech enhancement network,” Proceedings of the AAAI Conference on Artificial Intelligence , vol. 34, no. 05, pp. 9458–9465, Apr. 2020. [Online]. Available: https://ojs.aaai.org/index.php/AAAI/article...

    2020

  14. [22]

    A scale for the measurement of the psychological magnitude pitch,

    S. S. Stevens, J. E. V olkmann, and E. B. Newman, “A scale for the measurement of the psychological magnitude pitch,” Journal of the Acoustical Society of America , vol. 8, pp. 185–190, 1937. [Online]. Available: https://api.semanticscholar.org/CorpusID: 122448736

    1937

  15. [23]

    V ocos: Closing the gap between time-domain and fourier-based neural vocoders for high-quality audio synthesis,

    H. Siuzdak, “V ocos: Closing the gap between time-domain and fourier-based neural vocoders for high-quality audio synthesis,” arXiv preprint arXiv:2306.00814, 2023

    2023 arXiv

  16. [24]

    Ul- tra dual-path compression for joint echo cancellation and noise suppression,

    H. Chen, J. Yu, Y . Luo, R. Gu, W. Li, Z. Lu, and C. Weng, “Ul- tra dual-path compression for joint echo cancellation and noise suppression,” in INTERSPEECH 2023, 2023, pp. 2523–2527

    2023

  17. [25]

    EARS: An anechoic fullband speech dataset benchmarked for speech enhancement and derever- beration,

    J. Richter, Y .-C. Wu, S. Krenn, S. Welker, B. Lay, S. Watan- abe, A. Richard, and T. Gerkmann, “EARS: An anechoic fullband speech dataset benchmarked for speech enhancement and derever- beration,” in Interspeech, 2024

    2024

  18. [26]

    About 200-hour high- quality speech data are selected for each language

    and DNS challenge [21] for English. About 200-hour high- quality speech data are selected for each language. The ASR model is trained on the mixture of ”tr05” and ”si284” of the Wall Street Journal (WSJ) corpus [27] dataset as provided in the ESPNet toolkit. Parameter Settings...

  19. [27]

    Icassp 2023 deep noise suppression challenge,

    H. Dubey, A. Aazami, V . Gopal, B. Naderi, S. Braun, R. Cutler, H. Gamper, M. Golestaneh, and R. Aichner, “Icassp 2023 deep noise suppression challenge,” in ICASSP, 2023

    2023

  20. [28]

    Aishell-1: An open- source mandarin speech corpus and a speech recognition base- line,

    H. Bu, J. Du, X. Na, B. Wu, and H. Zheng, “Aishell-1: An open- source mandarin speech corpus and a speech recognition base- line,” in 2017 20th conference of the oriental chapter of the inter- national coordinating committee on speech databases and speech I/O systems and asses...

    2017

  21. [29]

    AISHELL-2: Transforming Mandarin ASR Research Into Industrial Scale,

    J. Du, X. Na, X. Liu, and H. Bu, “AISHELL-2: Transforming Mandarin ASR Research Into Industrial Scale,”ArXiv, Aug. 2018

    2018

  22. [30]

    Thchs-30: A free chinese speech cor- pus,

    D. Wang and X. Zhang, “Thchs-30: A free chinese speech cor- pus,” arXiv preprint arXiv:1512.01882, 2015

    2015 arXiv

  23. [31]

    Cstr vctk corpus: English multi-speaker corpus for cstr voice cloning toolkit (version 0.92),

    J. Yamagishi, C. Veaux, and K. MacDonald, “Cstr vctk corpus: English multi-speaker corpus for cstr voice cloning toolkit (version 0.92),” 2019. [Online]. Available: https: //api.semanticscholar.org/CorpusID:213060286

    2019

  24. [32]

    The design for the wall street journal- based csr corpus,

    D. B. Paul and J. M. Baker, “The design for the wall street journal- based csr corpus,” in 2nd International Conference on Spoken Language Processing (ICSLP 1992), 1992, pp. 899–902

    1992

  25. [33]

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

    A. Rix, J. Beerends, M. Hollier, and A. Hekstra, “Perceptual eval- uation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs,” inICASSP, vol. 2, 2001, pp. 749–752 vol.2

    2001

  26. [34]

    An al- gorithm for intelligibility prediction of time–frequency weighted noisy speech,

    C. H. Taal, R. C. Hendriks, R. Heusdens, and J. Jensen, “An al- gorithm for intelligibility prediction of time–frequency weighted noisy speech,” IEEE Transactions on Audio, Speech, and Lan- guage Processing, vol. 19, no. 7, pp. 2125–2136, 2011

    2011

  27. [35]

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

    C. K. Reddy, V . Gopal, and R. Cutler, “Dnsmos p. 835: A non-intrusive perceptual objective speech quality metric to eval- uate noise suppressors,” in ICASSP 2022-2022 IEEE Interna- tional Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2022, pp. 886–890

    2022

Pith tools

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