Unveiling the Best Practices for Applying Speech Foundation Models to Speech Intelligibility Prediction for Hearing-Impaired People

Boxuan Cao; Changgeng Mo; Haoshuai Zhou; Linkai Li; Shan Xiang Wang

REVIEW 4 major objections 7 minor 26 references

Unveiling the Best Practices for Applying Speech Foundation Models to Speech Intelligibility Prediction for Hearing-Impaired People

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read For speech intelligibility prediction in hearing-impaired listeners, selecting a single encoder layer from a speech foundation model beats learned combinations of all layers, and temporal prediction heads plus ensembling improve accuracy…

desk verdict Useful empirical recipe for SIP-HI, but the headline single-layer claim lacks error bars and Table 3 has a concrete duplicate-row error; worth reviewing with revisions. read the letter →

arxiv 2505.08215 v1 pith:7HLZGU4S submitted 2025-05-13 cs.AI cs.SDeess.AS

Haoshuai Zhou , Boxuan Cao , Changgeng Mo , Linkai Li , Shan Xiang Wang This is my paper · ORCID

classification cs.AIcs.SDeess.AS

keywords speechintelligibilitypredictionhearing-impairedlistenersfoundationmodelsencoderlayerselectionheaddesigntemporalmodelingmodelensemblingClarityChallenge

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proposes a recipe for adapting frozen speech foundation models to predict how intelligible processed speech is to hearing-impaired listeners. It reports that the standard practice of feeding all encoder layers into a trained fusion module is usually worse than picking one layer, that prediction heads with temporal transformers outperform simpler pooling even when layer fusion is simple, and that ensembling three speech models beats every single model. If true, these are practical design rules for building speech intelligibility prediction systems with limited training data, where foundation-model transfer is currently the dominant approach.

What carries the argument

The central machinery is an adapter framework: a frozen speech foundation model encoder plus a lightweight trainable prediction head. Layer selection is studied by feeding either one encoder layer or all encoder layers into one of two heads, WA-TGP (weighted average with temporal global pooling) or DT (double transformer, with separate temporal and layer-wise transformers). A third head, WA-TT, uses a temporal transformer but weights layer features by a simple learned average, isolating the contribution of temporal modeling. The best single layer is found by sweeping layer depth per model, and ensembles are formed by learning a weighted average of the best configurations of three different speech foundation models.

What would settle it

Re-run the layer-selection, prediction-head, and ensemble experiments many times with different random seeds and different listener splits on the same dataset. If the spread of RMSE across seeds or splits is comparable to the observed gaps (for example, the best ensemble at 22.29 RMSE versus the worst single model at 25.03, or the layer-to-layer differences within a single model), then the claimed rankings of layers and ensembles would not be stable and the central recommendations would fail to replicate.

Watch

Extended reading notes

Core claim

The paper claims that the prevailing use-all-layers strategy for adapting speech foundation models to speech intelligibility prediction is suboptimal. Across five speech foundation models and two prediction head architectures, the best-performing configurations almost always rely on features from a single encoder layer rather than self-learned weighted combinations of all layers. It also claims that temporal modeling in the prediction head is more influential than layer fusion capability or embedding dimensions, and that ensembling three speech foundation models consistently outperforms the best individual model, with stronger individual models contributing more to the ensemble gain.

Load-bearing premise

The reported performance differences are treated as meaningful even though each configuration was trained only once on a small repartitioned development set, so random training variation or listener sampling noise could be as large as the observed gaps.

Editorial extensions

If this is right

Speech intelligibility prediction systems should sweep encoder layers per foundation model rather than defaulting to all-layer fusion, because the best layer index differs across models.
Prediction heads should include a temporal transformer before any layer fusion, since WA-TT and DT both beat the simpler WA-TGP head.
Ensembling three speech foundation models improves both RMSE and NCC over the best single model, so ensemble selection should prioritize stronger individual models.
Very large prediction-head embedding dimensions (1536) tend to hurt performance, so dimensions near the foundation model's own feature dimension (typically 1024) are safer.
Because optimal layer choice does not transfer across speech foundation models, each new model requires its own layer-wise evaluation for this task.

Reading between the lines

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

A likely consequence not tested here is that the all-layer attention mechanisms in previous top systems add parameters without helping accuracy; a single-layer plus temporal-transformer head may reach comparable or better scores with lower compute.
The strong correlation with ASR word error rate suggests SIP-HI performance could be roughly predicted from a foundation model's ASR quality, letting practitioners preselect models before running expensive layer sweeps.
Since optimal layer depth varies with architecture and training objective, a testable hypothesis is that later layers help ASR-oriented models while instruction-tuned multimodal models peak at earlier layers, possibly reflecting where phonetic detail is most explicit.
The rankings are based on single-run training on a small repartitioned development set, so repeated-seed and listener-resampling experiments are needed before treating the layer and ensemble orderings as stable.

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.

Desk Editor's Note

read the letter

Short version: this is a useful empirical recipe for SIP-HI, not a field-level breakthrough. The main findings are plausible, but the paper needs variance estimates and a correction to Table 3 before they're fully credible.

The genuinely new part is the systematic sweep: five current SFMs, three prediction heads, layer-by-layer feature selection, and three-model ensembles on the CPC dataset. Nobody had published that combination for SIP-HI. The single-layer result is a real counterpoint to the use-all-layers strategy from CPC2, and the temporal-head result is consistent with other speech tasks. The authors also deserve credit for reporting the Phi-4 exception and the WA-TGP instability instead of hiding them.

Where it gets soft: every headline number is a single run. The dataset is small (13,126 clips, 27 listeners) and the RMSE gaps between configurations are small: best ensemble 22.29 vs worst 23.15, best individual 23.56 vs worst 25.03. Without per-fold or per-seed error bars, the layer ranking and the ensemble ranking could easily shuffle. The stress-test note worries about selection bias from picking the best layer on the same test scores. That concern seems overstated if layer selection is done on validation, as the reader assumes; but the paper never clearly says which split was used for layer selection, so the reader cannot tell. More concretely, Table 3 lists (Canary, OWSM, Whisper) at both rank 7 and rank 10 and omits (Parakeet, OWSM, Whisper). That is a factual error in a load-bearing table, and it needs fixing before publication.

The attribute correlation analysis in Section 4 is post hoc with five data points; the 'fewer training data hours help' observation is not something I'd put in a conclusion. They should soften it or drop it.

Who this is for: practitioners tuning SFM adapters for hearing-aid intelligibility prediction, and to a lesser extent anyone building lightweight heads on frozen speech encoders. It is a benchmark recipe, not a theory paper.

Recommendation: send it to peer review, but ask for per-fold or per-seed variance, the exact data split, and a corrected Table 3. With those, it would be a solid reference for the SIP-HI community.

Referee Report

4 major / 7 minor

Summary. The paper reports an empirical study of how to adapt five speech foundation models (Canary, Parakeet, Whisper, OWSM, Phi-4) to speech intelligibility prediction for hearing-impaired listeners (SIP-HI) using the Clarity Prediction Challenge dataset. It compares encoder layer selection (single layer vs. learned fusion of all layers), three prediction head architectures (WA-TGP, WA-TT, DT), and all ten three-SFM ensembles. The main claims are that a single encoder layer outperforms self-learned all-layer fusion, that temporal modeling in the prediction head is critical, and that ensembles of stronger individual SFMs are more effective. It also explores correlations between SFM attributes and SIP-HI performance.

Significance. If the findings hold, the paper provides concrete, actionable design guidance for SIP-HI systems: freeze an SFM, select a single best encoder layer, use a temporal transformer head, and ensemble strong individual models. The study uses publicly available SFMs and a public dataset, and the experimental protocol is described in sufficient detail to be reproducible. The authors also honestly acknowledge the instability of WA-TGP and the Phi-4 exception to the layer-fusion result, which adds credibility. However, the evidence base is small—five SFMs and one repartitioned development set—and the reported differences are often smaller than the likely run-to-run or fold-to-fold variation, which limits the strength of the conclusions until the statistical robustness is demonstrated.

major comments (4)

[3.1, Figure 2, Table 1] The central claim that a single encoder layer outperforms self-learned fusion of all layers is not yet established because the layer-selection protocol is ambiguous and no variance information is reported. The manuscript does not state whether the best-performing layer was selected using validation scores or the same test-set averages used for the comparison; if the latter, the maximum over many correlated layer estimates is biased upward, making the single-layer advantage at least partly an artifact of selection. Please clarify the selection protocol, report the selected layer indices for each configuration, and provide per-fold or per-seed RMSE and NCC values for the layer sweep.
[3.3, Table 3] Table 3 contains a concrete data inconsistency: the combination (Canary, OWSM, Whisper) appears at both rank 7 (RMSE 22.79) and rank 10 (RMSE 23.15), while the combination (Parakeet, OWSM, Whisper) is missing. Because the ensemble ranking and the accompanying claim that stronger individual models lead to better ensembles depend directly on this table, the error must be corrected (or a proper enumeration of all ten combinations with unique rows must be provided) before the third main claim can be evaluated.
[2.1, 2.6, Tables 1–3] The paper states that it adopts the three-fold practice of [2] to reduce score variance, but all tables report single scalar scores with no per-fold or per-seed values. Given that the best ensemble RMSE is 22.29 and the worst is 23.15, while the best individual models range from 23.56 to 25.03, the rankings could change under fold-to-fold or seed-to-seed variation. Please release per-fold results and the exact data split, or provide error bars and significance tests, so that the reported orderings are meaningful.
[3.3, Tables 2 and 3] The claim that stronger individual SFMs are better ensemble candidates is partly circular because the individual rankings in Table 2 and the ensemble rankings in Table 3 are computed on the same test set, with ensemble weights learned from the same data. This does not invalidate the observation, but it weakens its value as a design rule. To support the recommendation, please show that the ranking of ensemble gain is robust when individual models are ranked on validation data or when ensemble evaluation uses a different test split.

minor comments (7)

[2.2] There is a typo in the model name: 'Whipser' should be 'Whisper'.
[2.3] The word 'adpopted' should be 'adopted'.
[Throughout] The abbreviation 'W A-TGP' contains an extra space; please use 'WA-TGP' consistently.
[Table 1] The caption does not explain the slash-separated values in the table; please state explicitly that they correspond to the four embedding dimensions (192, 384, 768, 1536).
[Figure 2] The solid and dotted lines may be difficult to distinguish in grayscale; consider adding distinct markers or line styles for single-layer vs. all-layer results.
[4] The correlation analysis is based on only five SFMs, and the text reports 'strong positive correlation' and 'strong negative relationship' without giving the rank correlation coefficients or their uncertainty; please report these values and temper the claims accordingly.
[2.2, Reference [24]] The paper uses the Phi-4-Multimodal model, but the cited reference [24] is the Phi-4-mini technical report; please ensure the citation matches the actual model version used.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity; the study is an empirical comparison on held-out data, with only a mild post-hoc correlation in the 'stronger models help' claim.

full rationale

The paper's claims are all empirical comparisons, not derivations from first principles. Section 2.6 states that for each configuration the checkpoint with the lowest RMSE on the validation set is selected for final evaluation on the test set, so layer and head choices are model selection rather than fitted predictions. Section 3.1 compares single-layer versus all-layer features on the same held-out test data; no equation defines one outcome in terms of the other. Section 3.2 compares three prediction-head architectures directly. Section 3.3 learns ensemble weights and evaluates all three-model combinations on the test set. The statement that stronger individual SFMs yield better ensembles is a post-hoc correlation between the test-set rankings in Tables 2 and 3, not a derivation that is forced by construction; this is a mild statistical limitation, not circularity. There are no load-bearing self-citations: references [12], [13], and [25] are external systems used as baselines or design sources, not as internal justifications for the paper's conclusions. The duplicate (Canary, OWSM, Whisper) row in Table 3 is an internal inconsistency that affects evidence quality, but it does not make the claim circular.

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

The paper introduces no new physical or model entities. Its dependence is on domain assumptions about the reliability of the CPC labels, the sufficiency of frozen SFM features, and the representativeness of a single-run, three-fold evaluation. The main free parameters are the selected layer, embedding dimension, ensemble weights, and training hyperparameters.

free parameters (4)

Best encoder layer index per (SFM, prediction head) = Not stated numerically; Canary and Parakeet favor middle layers, OWSM and Whisper favor later layers, Phi-4 favors…
Selected by lowest validation RMSE during the layer sweep in Section 3.1; it is the core hyperparameter behind Claim 1 and is model-specific, so a new SFM requires a fresh sweep.
Prediction head embedding dimension = Best values per head in Table 1, with 1536 usually worst
Swept over 192, 384, 768, and 1536 in Section 2.4; used to pick the reported best configuration, and the paper notes no clear optimal dimension except the 1536 penalty.
Ensemble combination weights = Softmax weights, distributions shown in Figure 3, numeric values not listed
Learned weighted average across the three SFM outputs in Section 2.5; central to the ensemble results, and the weights are fit to the validation set.
Training hyperparameters (learning rate, warm-up, epochs, loss) = LR 1e-4 for WA-TGP, 3e-5 for transformer heads, 50 epochs, Huber loss, cosine schedule
Chosen by hand in Section 2.6; not swept, but they affect all configurations. They are standard values, not tuned per model.

assumptions (4)

domain assumption CPC listening-test scores from 27 listeners are reliable enough to rank configurations separated by about 1 RMSE point.
The whole comparison uses these labels as ground truth in Section 2.1; measurement noise and listener variance are not modeled, so close rankings could be noise.
domain assumption Frozen SFM representations are sufficient; feature extractors need not be fine-tuned for SIP-HI.
The adapter framework in Section 2.3 fixes all five backbones and trains only the prediction head, so any limitation of frozen features is inherited by the conclusions.
domain assumption Three-fold repartition of the CPC development set is representative and has no listener leakage.
Section 2.1 describes the repartition and non-overlap but gives no fold seeds or fold-wise listener counts.
domain assumption Each configuration is run once; run-to-run variance from random initialization and data shuffling is negligible.
Section 2.6 gives no seeds or repeated runs, yet all table comparisons treat the reported scores as exact.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Unveiling the Best Practices for Applying Speech Foundation Models to Speech Intelligibility Prediction for Hearing-Impaired People." pith.science (2026). https://pith.science/paper/7HLZGU4S

@misc{pith2026250508215,
  author       = {Pith},
  title        = {Pith review of: Unveiling the Best Practices for Applying Speech Foundation Models to Speech Intelligibility Prediction for Hearing-Impaired People},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7HLZGU4S}},
  note         = {Machine review of arXiv:2505.08215}
}

read the original abstract

Speech foundation models (SFMs) have demonstrated strong performance across a variety of downstream tasks, including speech intelligibility prediction for hearing-impaired people (SIP-HI). However, optimizing SFMs for SIP-HI has been insufficiently explored. In this paper, we conduct a comprehensive study to identify key design factors affecting SIP-HI performance with 5 SFMs, focusing on encoder layer selection, prediction head architecture, and ensemble configurations. Our findings show that, contrary to traditional use-all-layers methods, selecting a single encoder layer yields better results. Additionally, temporal modeling is crucial for effective prediction heads. We also demonstrate that ensembling multiple SFMs improves performance, with stronger individual models providing greater benefit. Finally, we explore the relationship between key SFM attributes and their impact on SIP-HI performance. Our study offers practical insights into effectively adapting SFMs for speech intelligibility prediction for hearing-impaired populations.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 18 canonical work pages

[2]

The 2nd clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,

J. Barker, M. A. Akeroyd, W. Bailey, T. J. Cox, J. F. Culling, J. Firth, S. Graetzer, and G. Naylor, “The 2nd clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,” ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. 11 551–11 555, 2024

work page 2024
[1]

The 1st clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,

J. Barker, M. A. Akeroyd, T. J. Cox, J. F. Culling, J. Firth, S. Graetzer, H. Griffiths, L. Harris, G. Naylor, Z. Podwinska, E. Porter, and R. V . Mu˜noz, “The 1st clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,” in Interspeech, 2022

work page 2022
[3]

The hearing-aid speech perception index (haspi),

J. M. Kates and K. H. Arehart, “The hearing-aid speech perception index (haspi),” Speech Commun. , vol. 65, pp. 75–93, 2014

work page 2014
[4]

Effect of hearing aid technology level and individual characteristics on listener outcome measures

P. N. Plyler, J. Hausladen, M. Capps, and M. A. Cox, “Effect of hearing aid technology level and individual characteristics on listener outcome measures.” Journal of speech, language, and hearing research : JSLHR , pp. 1–13, 2021

work page 2021
[5]

Auditory inspired machine learning techniques can improve speech intelligibility and quality for hearing-impaired listeners

J. J. M. Monaghan, T. Goehring, X. Yang, F. Bolner, S. Wang, M. C. M. Wright, and S. Bleeck, “Auditory inspired machine learning techniques can improve speech intelligibility and quality for hearing-impaired listeners.” The Journal of the Acoustical Society of America , vol. 141 3, p. 1985, 2017

work page 1985
[6]

Measuring speech intelligibility and hearing-aid benefit using everyday conversational sentences in real-world environments,

K. Miles, T. Beechey, V . Best, and J. M. Buchholz, “Measuring speech intelligibility and hearing-aid benefit using everyday conversational sentences in real-world environments,” Frontiers in Neuroscience, vol. 16, 2022

work page 2022
[7]

On the opportunities and risks of foundation models,

R. Bommasani, D. A. Hudson, E. Adeli, R. Altman, S. Arora, S. von Arx, M. S. Bernstein, J. Bohg, A. Bosselut, E. Brunskill, E. Brynjolfsson, S. Buch, D. Card, R. Castellon, N. S. Chatterji, A. S. Chen, K. A. Creel, J. Davis, D. Demszky, C. Donahue, M. K. B. Doumbouya, E. Durmus, S. Ermon, J. Etchemendy, K. Ethayarajh, L. Fei-Fei, C. Finn, T. Gale, L. Gill...

arXiv 2021
[8]

wav2vec 2.0: A framework for self-supervised learning of speech representations,

A. Baevski, H. Zhou, A. rahman Mohamed, and M. Auli, “wav2vec 2.0: A framework for self-supervised learning of speech representations,” ArXiv, vol. abs/2006.11477, 2020

arXiv 2006

Show all 26 references

[9]

Hubert: Self-supervised speech representation learning by masked prediction of hidden units,

W.-N. Hsu, B. Bolte, Y .-H. H. Tsai, K. Lakhotia, R. Salakhutdinov, and A. rahman Mohamed, “Hubert: Self-supervised speech representation learning by masked prediction of hidden units,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 29, pp. 3451–3460, 2021

2021
[10]

Language models are few-shot learners,

T. B. Brown, B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert- V oss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. teusz Litwin, S. G...

2005 arXiv
[11]

Superb: Speech processing universal performance benchmark,

S.-W. Yang, P.-H. Chi, Y .-S. Chuang, C.-I. Lai, K. Lakhotia, Y . Y . Lin, A. T. Liu, J. Shi, X. Chang, G.-T. Lin, T. hsien Huang, W.-C. Tseng, K. tik Lee, D.-R. Liu, Z. Huang, S. Dong, S.-W. Li, S. Watanabe, A. rahman Mohamed, and H. yi Lee, “Superb: Speech processing univers...

2021
[12]

Speech foundation models on intelligibility prediction for hearing-impaired listeners,

S. Cuervo and R. Marxer, “Speech foundation models on intelligibility prediction for hearing-impaired listeners,” ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1421–1425, 2024

2024
[13]

Non-intrusive speech intelligibility prediction for hearing- impaired users using intermediate asr features and human memory models,

R. Mogridge, G. Close, R. Sutherland, T. Hain, J. Barker, S. Goetze, and A. Ragni, “Non-intrusive speech intelligibility prediction for hearing- impaired users using intermediate asr features and human memory models,” ICASSP 2024 - 2024 IEEE International Conference on Acousti...

2024
[14]

What do speech foundation models not learn about speech?

A. Waheed, H. Atwany, B. Raj, and R. Singh, “What do speech foundation models not learn about speech?” ArXiv, vol. abs/2410.12948, 2024

2024 arXiv
[15]

Analysis of xls-r for speech quality assessment,

B. Tamm, R. Vandenberghe, and H. V . Hamme, “Analysis of xls-r for speech quality assessment,” 2023 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA) , pp. 1–5, 2023

2023
[16]

Less is more: Accurate speech recognition & translation without web- scale data,

K. C. Puvvada, P. ˙Zelasko, H. Huang, O. Hrinchuk, N. R. Koluguri, K. Dhawan, S. Majumdar, E. Rastorgueva, Z. Chen, V . Lavrukhin et al. , “Less is more: Accurate speech recognition & translation without web- scale data,” arXiv preprint arXiv:2406.19674 , 2024

2024 arXiv
[17]

Fast conformer with linearly scalable attention for efficient speech recognition,

D. Rekesh, S. Kriman, S. Majumdar, V . Noroozi, H. Juang, O. Hrinchuk, A. Kumar, and B. Ginsburg, “Fast conformer with linearly scalable attention for efficient speech recognition,” 2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU) , pp. 1–8, 2023

2023
[18]

Attention is all you need,

A. Vaswani, N. M. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” in Neural Information Processing Systems , 2017

2017
[19]

Efficient sequence transduction by jointly predicting tokens and durations,

H. Xu, F. Jia, S. Majumdar, H. Huang, S. Watanabe, and B. Ginsburg, “Efficient sequence transduction by jointly predicting tokens and durations,” ArXiv, vol. abs/2304.06795, 2023

2023 arXiv
[20]

Robust speech recognition via large-scale weak supervision,

A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak supervision,” in International Conference on Machine Learning , 2022

2022
[21]

Reproducing whisper-style training using an open-source toolkit and publicly available data,

Y . Peng, J. Tian, B. Yan, D. Berrebbi, X. Chang, X. Li, J. Shi, S. Arora, W. Chen, R. Sharma, W. Zhang, Y . Sudo, M. Shakeel, J. weon Jung, S. Maiti, and S. Watanabe, “Reproducing whisper-style training using an open-source toolkit and publicly available data,” 2023 IEEE Auto...

2023
[22]

Owsm v3.1: Better and faster open whisper-style speech models based on e-branchformer,

Y . Peng, J. Tian, W. Chen, S. Arora, B. Yan, Y . Sudo, M. Shakeel, K. Choi, J. Shi, X. Chang, J. weon Jung, and S. Watanabe, “Owsm v3.1: Better and faster open whisper-style speech models based on e-branchformer,” ArXiv, vol. abs/2401.16658, 2024

2024 arXiv
[23]

E-branchformer: Branchformer with enhanced merging for speech recognition,

K. Kim, F. Wu, Y . Peng, J. Pan, P. Sridhar, K. J. Han, and S. Watanabe, “E-branchformer: Branchformer with enhanced merging for speech recognition,” 2022 IEEE Spoken Language Technology Workshop (SLT) , pp. 84–91, 2022

2022
[24]

Phi-4-mini technical report: Compact yet powerful multimodal language models via mixture-of-loras,

A. Abouelenin, A. Ashfaq, A. Atkinson, H. H. Awadalla, N. Bach, J. Bao, A. Benhaim, M. Cai, V . Chaudhary, C. Chen, D. Chen, D. Chen, J. Chen, W. Chen, Y .-C. Chen, Y . ling Chen, Q. Dai, X. Dai, R. Fan, M. Gao, M. Gao, A. Garg, A. Goswami, J. Hao, A. Hendy, Y . Hu, X. Jin, M....

2025 arXiv
[25]

Whisper-at: Noise- robust automatic speech recognizers are also strong general audio event taggers,

Y . Gong, S. Khurana, L. Karlinsky, and J. Glass, “Whisper-at: Noise- robust automatic speech recognizers are also strong general audio event taggers,” in Interspeech, 2023

2023
[26]

Open automatic speech recognition leaderboard,

V . Srivastav, S. Majumdar, N. Koluguri, A. Moumen, S. Gandhi et al. , “Open automatic speech recognition leaderboard,” https://huggingface.co/ spaces/hf-audio/open asr leaderboard, 2023

2023

Pith tools

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

[1] [2]

The 2nd clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,

J. Barker, M. A. Akeroyd, W. Bailey, T. J. Cox, J. F. Culling, J. Firth, S. Graetzer, and G. Naylor, “The 2nd clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,” ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. 11 551–11 555, 2024

work page 2024

[2] [1]

The 1st clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,

J. Barker, M. A. Akeroyd, T. J. Cox, J. F. Culling, J. Firth, S. Graetzer, H. Griffiths, L. Harris, G. Naylor, Z. Podwinska, E. Porter, and R. V . Mu˜noz, “The 1st clarity prediction challenge: A machine learning challenge for hearing aid intelligibility prediction,” in Interspeech, 2022

work page 2022

[3] [3]

The hearing-aid speech perception index (haspi),

J. M. Kates and K. H. Arehart, “The hearing-aid speech perception index (haspi),” Speech Commun. , vol. 65, pp. 75–93, 2014

work page 2014

[4] [4]

Effect of hearing aid technology level and individual characteristics on listener outcome measures

P. N. Plyler, J. Hausladen, M. Capps, and M. A. Cox, “Effect of hearing aid technology level and individual characteristics on listener outcome measures.” Journal of speech, language, and hearing research : JSLHR , pp. 1–13, 2021

work page 2021

[5] [5]

Auditory inspired machine learning techniques can improve speech intelligibility and quality for hearing-impaired listeners

J. J. M. Monaghan, T. Goehring, X. Yang, F. Bolner, S. Wang, M. C. M. Wright, and S. Bleeck, “Auditory inspired machine learning techniques can improve speech intelligibility and quality for hearing-impaired listeners.” The Journal of the Acoustical Society of America , vol. 141 3, p. 1985, 2017

work page 1985

[6] [6]

Measuring speech intelligibility and hearing-aid benefit using everyday conversational sentences in real-world environments,

K. Miles, T. Beechey, V . Best, and J. M. Buchholz, “Measuring speech intelligibility and hearing-aid benefit using everyday conversational sentences in real-world environments,” Frontiers in Neuroscience, vol. 16, 2022

work page 2022

[7] [7]

On the opportunities and risks of foundation models,

R. Bommasani, D. A. Hudson, E. Adeli, R. Altman, S. Arora, S. von Arx, M. S. Bernstein, J. Bohg, A. Bosselut, E. Brunskill, E. Brynjolfsson, S. Buch, D. Card, R. Castellon, N. S. Chatterji, A. S. Chen, K. A. Creel, J. Davis, D. Demszky, C. Donahue, M. K. B. Doumbouya, E. Durmus, S. Ermon, J. Etchemendy, K. Ethayarajh, L. Fei-Fei, C. Finn, T. Gale, L. Gill...

arXiv 2021

[8] [8]

wav2vec 2.0: A framework for self-supervised learning of speech representations,

A. Baevski, H. Zhou, A. rahman Mohamed, and M. Auli, “wav2vec 2.0: A framework for self-supervised learning of speech representations,” ArXiv, vol. abs/2006.11477, 2020

arXiv 2006