ASR-FAIRBENCH: Measuring and Benchmarking Equity Across Speech Recognition Systems

Anand Rai; Animesh Mukherjee; Satyam Rahangdale; Utkarsh Anand

REVIEW 5 major objections 5 minor 11 references

ASR-FAIRBENCH: Measuring and Benchmarking Equity Across Speech Recognition Systems

T0 review · 5 major / 5 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 The paper introduces a real-time leaderboard that scores speech recognition models on demographic fairness as well as accuracy, and shows that Whisper-medium leads with a Fairness-Adjusted ASR Score of 29.41.

desk verdict Useful fairness leaderboard idea, but the headline rankings are not supported by the manuscript's equations or reported data. read the letter →

arxiv 2505.11572 v1 pith:GIENMGNI submitted 2025-05-16 cs.SD cs.CLeess.AS

Anand Rai , Satyam Rahangdale , Utkarsh Anand , Animesh Mukherjee This is my paper · ORCID

classification cs.SDcs.CLeess.AS

keywords ASRfairnessspeechrecognitionleaderboardmixed-effectsPoissonregressionWordErrorRatedemographicdisparityFAASFair-Speechdataset

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 introduces ASR-FAIRBENCH, a real-time leaderboard that scores automatic speech recognition systems on both accuracy and demographic equity. Its central claim is that fairness can be quantified from a small stratified sample of the Fair-Speech dataset using mixed-effects Poisson regression, then folded with Word Error Rate into a single Fairness-Adjusted ASR Score (FAAS). If this is right, benchmark rankings change: Whisper-medium tops FAAS at 29.41, Whisper-tiny is fairer despite higher WER, and fairness gaps can outweigh small accuracy gains. The platform is offered as a public audit tool for model submissions.

What carries the argument

The central object is the Fairness-Adjusted ASR Score (FAAS), built from a mixed-effects Poisson regression of word error counts with demographic attributes as fixed effects and speaker-level random effects. The regression produces disparity ratios $e^{\beta_1}$; predicted WERs per group are min-max scaled to 0–100 and weighted by group prevalence; a likelihood-ratio test decides which disparities are penalized; and the log-ratio $\text{FAAS} = 10 \log_{10}(\text{Overall Score} / \text{WER})$ combines fairness and accuracy into one number. This lets a single score rank models by both competence and equity.

What would settle it

Re-run the full FAAS pipeline on the complete 26,471-utterance Fair-Speech corpus (or on several fresh stratified draws) and compare disparity-ratio confidence intervals and model rankings. If Whisper-medium's top FAAS of 29.41 or Whisper-tiny's fairness lead reverses, the leaderboard's reported rankings are sample artifacts rather than stable model properties.

Watch

Extended reading notes

Core claim

On the paper's own terms: a 10% stratified sample of 26,471 Fair-Speech utterances preserves per-attribute entropy while cutting evaluation data by 90%; a mixed-effects Poisson regression on word error counts estimates disparity ratios per demographic attribute; statistically significant disparities ($p < 0.05$) penalize category scores by $p/0.05$; and the weighted overall fairness score enters $\text{FAAS} = 10 \log_{10}(\text{Overall Score} / \text{WER})$. Applied to five models, this yields a leaderboard where Whisper-medium scores 29.41, Whisper-tiny shows better overall fairness despite higher WER, and fine-tuned Wav2Vec and HuBERT models lag in fairness despite efficiency.

Load-bearing premise

The load-bearing premise is that a 10% stratified sample whose per-attribute entropy matches the full dataset also preserves the joint demographic distribution and leaves enough speakers per group for the regression and significance tests to be reliable; if not, fairness rankings could be artifacts of sparse demographic cells.

Editorial extensions

If this is right

A model with lower WER can rank below a fairer model with slightly higher WER, so accuracy-only leaderboards can mislead purchasers and developers.
Researchers can submit any ASR model and receive an automated fairness audit within minutes, making equity measurement a routine part of model release.
The five-tier classification (from severely biased to exemplarily fair) gives attribute-specific diagnostics, showing which demographic category drives a model's penalty.
Whisper-family models currently lead FAAS; Whisper-tiny's better fairness at higher WER suggests model scale and fairness do not move together.
The $p < 0.05$ penalty schedule means only statistically supported disparities affect scores, preventing noise from moving rankings.

Reading between the lines

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

Marginal entropy matching does not guarantee joint-distribution preservation; the authors' sampling decision deserves a direct check by comparing full-data and sampled-data rankings.
The FAAS formulation could be ported to other speech tasks (speaker verification, speech translation) or to intersectional groups (gender × ethnicity), where disparities may be larger than any single attribute shows.
A public leaderboard with a single signed score creates an incentive to optimize for FAAS; whether that drives genuine equity gains or mere statistical tailoring is a testable outcome of adoption.
The penalty $p/0.05$ is one of many possible schedules; sensitivity analysis over penalty shapes would show whether top rankings are robust.

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

Two things to know before you read it. The FAAS leaderboard concept is genuinely useful: a single number that combines WER with a fairness penalty, backed by a live, reproducible benchmark for ASR equity. That fills a real niche. The second thing is that the headline numbers—Whisper-medium at 29.41, Whisper-tiny fairer despite higher WER—are not verifiable from the manuscript. There is no results table, no p-values, no confidence intervals, and Eq. (2), which defines predicted WER, is malformed as typeset. The stress-test note is correct: the fairness score chain depends on that equation, so the rankings are not checkable from the paper alone.

What the paper does well: the mixed-effects Poisson regression approach to estimating demographic disparity ratios is sound and is an existing technique from earlier bias-measurement work, properly cited. The 10% stratified sampling is a pragmatic choice, and the entropy comparisons in Table 1 show per-attribute distributions are roughly preserved, even though the joint distribution is not checked. Building a public leaderboard with code and a HuggingFace space is a concrete, useful contribution that goes beyond many fairness papers.

Soft spots, in order of severity. First, the missing empirical backbone: a benchmarking paper must report per-model WER, disparity ratios, p-values, and per-category scores. This one reports only final FAAS numbers in the conclusion. That's a load-bearing omission. Second, Eq. (2) is not reproducible: \bar X is undefined, the β_logRef term appears from nowhere, and the denominator 'e \bar X − 1' doesn't parse. Since Eq. (3), (6), and (7) all feed off it, the whole metric is ambiguous. Third, the p/0.05 penalty in Eq. (5) is arbitrary and uncalibrated; it changes scores substantially and has no justification. Fourth, the 10% sample validation ignores intersections of demographic attributes, leaving sparse cells that could destabilize the mixed-effects regression. These are not minor quibbles—they undercut the claim that the rankings are real.

Who is this for? People building ASR systems for diverse users and researchers working on fairness evaluation metrics. They will appreciate the platform and the direction, but they cannot trust the specific rankings until the details are fixed. The paper deserves a serious referee because the topic is important and the platform is a real artifact, but the current draft should probably be rejected at review unless the authors supply the missing numbers, fix the equations, and justify the penalty. That's a heavy revision, not a polish.

Referee Report

5 major / 5 minor

Summary. The paper presents ASR-FAIRBENCH, a real-time leaderboard that evaluates both accuracy (WER) and equity in speech recognition, using a 10% stratified sample of the Fair-Speech dataset. The authors propose a mixed-effects Poisson regression to estimate demographic disparities, derive a raw fairness score from predicted WERs, apply a p-value-based penalty, and combine fairness with WER into a Fairness-Adjusted ASR Score (FAAS). They claim Whisper-medium achieves the highest FAAS (29.41), that Whisper-tiny is fairer despite higher WER, and that fairness can outweigh slight WER improvements in some comparisons. The paper also describes a web platform with interactive leaderboards and visualizations.

Significance. The intended contribution is timely and relevant: a benchmark that explicitly accounts for demographic equity in ASR would fill a real gap, and the authors provide a live leaderboard and reproducible code. If the metric were rigorously defined and the empirical results fully reported, this could be a useful community resource. However, in its current form the central scientific claims are not verifiable: the predicted-WER equation is malformed, the fairness penalty is ad hoc, and no numerical results or statistical summaries support the reported rankings. The paper therefore currently functions more as a platform demo than as a self-contained research contribution.

major comments (5)

[Section 2.2.2, Eq. (2)] The formula for predicted WER, which underpins the raw fairness score, is malformed and undefined: the symbols \bar X, β_logRef, and the denominator e \bar X − 1 are not introduced, and the expression is inconsistent with the regression in Eq. (1). Since Eqs. (3), (6), and (7) all depend on Eq. (2), the FAAS score cannot be computed or reproduced from the paper as written. A pointer to a HuggingFace space does not replace a self-contained derivation.
[Section 4 and Figure 1] The central empirical claims—Whisper-medium FAAS of 29.41, Whisper-tiny's superior fairness despite higher WER, and the Wav2Vec-large vs. Hubert-large comparison—are stated only in the conclusion, with no results table, per-model fairness scores, confidence intervals, or p-values anywhere in the manuscript. The screenshot in Figure 1 is not a substitute for reporting the underlying numbers. Without this evidence, the reader cannot assess whether the observed disparities are real or an artifact of the methodology.
[Section 2.2.3, Eq. (5)] The multiplicative penalty factor p/0.05 is invoked without any justification or sensitivity analysis. Because the p-value depends on sample size, this penalty can make the fairness score arbitrarily small for large datasets even when the group disparity is practically negligible, and it will tend to zero for any statistically significant effect as N grows. The choice of 0.05 as the threshold and the specific functional form need a principled derivation or at least a robustness check; currently the penalty is ad hoc and could dominate the ranking.
[Section 2.1, Table 1] The validation of the 10% stratified subsample is limited to reporting per-attribute entropy values. Matching marginal entropy for each attribute does not guarantee preservation of the joint demographic distribution, nor does it ensure sufficient observations in every cell of the mixed-effects Poisson model. The paper provides no analysis of cell sizes, sparsity, or standard errors for the disparity ratios, so the stability of the β1 estimates and the LRT p-values is unknown. If any demographic group is sparsely represented, the resulting FAAS scores could be driven by noise rather than by genuine model behavior.
[Section 2.2.2 and 2.2.3] The fairness score is computed from the fitted values of the same mixed-effects Poisson regression used to estimate demographic disparities, making the reported 'fairness' a direct transformation of the model's own predictions. This is not logically circular, but it means any misspecification of the Poisson model—such as unmodeled overdispersion, omitted covariates, or an incorrect random-effects structure—directly transfers into the fairness score. The manuscript does not report model diagnostics, goodness-of-fit measures, or a comparison of the Poisson model with alternative count or continuous-error models, so the robustness of the fairness ranking is unestablished.

minor comments (5)

[Section 2.2.1] The WER definition should parenthesize the numerator as (S+D+I)/N to avoid ambiguity.
[Section 1] The introduction mentions a five-tier classification from 'severely biased' to 'exemplarily fair' for overall and attribute-specific assessments, but this scheme is never described in the methodology or used in the results. Either define the tier boundaries and report the resulting tiers, or remove the claim.
[Section 5, Ref. [5]] The URL in reference [5] is malformed: it begins with 'urlhttps://' and contains a duplicated path ('huggingface.co/spaces/huggingface.co/spaces').
[Section 2.2.2, Eq. (3)] The min-max scaling in the raw fairness score is not specified as to whether the minimum and maximum are taken across groups within a category, across categories, or across all models, which affects the comparability of category scores.
[General] The paper uses an inconsistent mix of notation, such as \WER vs. WER and βlogRef vs. β1; the authors should unify the notation and define every symbol at its first appearance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity in the ASR-FAIRBENCH derivation; the fairness score is a stated transformation of the fitted Poisson regression, and the disparity claim rests on a likelihood-ratio test that is not forced by construction.

full rationale

No significant circularity found. The paper's fairness score (Eq. 3) is an explicit min-max normalization of the predicted WER \WER_g from the mixed-effects Poisson regression (Eq. 2), which is in turn fitted to the benchmark data. This is a measurement summary of the fitted demographic effects, not an independent prediction of a separate quantity, and the paper does not claim to predict fairness from first principles. The existence of disparities is additionally supported by the likelihood-ratio test (Eq. 4), whose p-value could in principle be non-significant; therefore the central claim is not forced by construction. The only self-citation, reference [3], appears in a background sentence about demographic variation in ASR and is not load-bearing. The manuscript does contain an explicitly flagged missing support: at the end of Section 2.2.3, it states that 'The detailed formulation of the FAAS metric can be found in the HuggingFace Space,' while Eq. (2) is under-specified in the typeset text and the reported per-model fairness values and p-values are not tabulated. Those are reproducibility and evidence-quality defects, not circularity, so they do not raise the circularity score under the specified criteria.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The central contribution rests on a new metric (FAAS) that depends on several ad hoc choices: the p/0.05 penalty, optional category weights, min-max normalization, and log scaling. The statistical model assumes Poisson-distributed WERs and valid demographic labels, while the sampling argument uses only entropy values to claim representativeness. These free parameters and assumptions are not externally benchmarked or validated.

free parameters (4)

Statistical significance penalty factor p/0.05 = p/0.05, with p threshold 0.05
In Eq. (5), category scores are multiplied by p/0.05 when p<0.05. This arbitrary hand-chosen penalty materially changes rankings, and no justification is provided.
Category weights w_c = unspecified
The overall fairness score in Eq. (6) uses optional weights w_c, but the paper does not state default values or how they are chosen. The final rankings depend on these weights unless they are uniform.
Min-max normalization scale for raw fairness score = 0-100
Raw fairness scores are rescaled to 0-100 via min-max normalization in Eq. (3), which changes spacing between group scores and depends on the worst and best groups in the sample.
FAAS scaling constants 10 and log10 = 10*log10(OverallScore/WER)
The logarithmic scaling with factor 10 in Eq. (7) is chosen for interpretability; it is not derived from any optimality or fairness criterion.

assumptions (5)

domain assumption Word error rates are Poisson-distributed counts conditional on demographic attributes and covariates
Eq. (1) uses a mixed-effects Poisson regression, but no overdispersion, zero-inflation, or model-fit checks are reported in Section 2.2.1.
domain assumption Entropy preservation of a 10% stratified sample implies preservation of joint demographic distribution and adequate cell sizes
Table 1 shows per-attribute entropy values only; no joint distribution checks or minimum cell count checks are reported in Section 2.1.
standard math The likelihood ratio test statistic follows its nominal chi-square distribution under the fitted mixed model
Eq. (4) uses an LRT to compare full and reduced models, assuming standard asymptotic theory without small-sample corrections.
domain assumption Self-reported demographics in Fair-Speech are accurate and the 593 participants are representative of the intended population
The benchmark inherits Fair-Speech demographic labels without independent validation, as stated in Section 2.1.
ad hoc to paper The p/0.05 multiplicative penalty provides a valid fairness adjustment
Eq. (5) penalizes significant disparities by scaling category scores, but no statistical, ethical, or empirical justification is given for this specific penalty.

invented entities (1)

Fairness-Adjusted ASR Score (FAAS)
purpose: A single composite score combining overall fairness with WER for ranking ASR models
FAAS is defined by Eq. (7) and is not validated against any external fairness judgment, user satisfaction measure, or downstream outcome. Its behavior is fully determined by the authors' chosen formulas and free parameters.

how reviews work

0 comments

Cite this review

Pith. "Pith review of ASR-FAIRBENCH: Measuring and Benchmarking Equity Across Speech Recognition Systems." pith.science (2026). https://pith.science/paper/GIENMGNI

@misc{pith2026250511572,
  author       = {Pith},
  title        = {Pith review of: ASR-FAIRBENCH: Measuring and Benchmarking Equity Across Speech Recognition Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GIENMGNI}},
  note         = {Machine review of arXiv:2505.11572}
}

read the original abstract

Automatic Speech Recognition (ASR) systems have become ubiquitous in everyday applications, yet significant disparities in performance across diverse demographic groups persist. In this work, we introduce the ASR-FAIRBENCH leaderboard which is designed to assess both the accuracy and equity of ASR models in real-time. Leveraging the Meta's Fair-Speech dataset, which captures diverse demographic characteristics, we employ a mixed-effects Poisson regression model to derive an overall fairness score. This score is integrated with traditional metrics like Word Error Rate (WER) to compute the Fairness Adjusted ASR Score (FAAS), providing a comprehensive evaluation framework. Our approach reveals significant performance disparities in SOTA ASR models across demographic groups and offers a benchmark to drive the development of more inclusive ASR technologies.

Figures

Figures reproduced from arXiv: 2505.11572 by the authors.

**Figure 1.** Key UI features: ASR-FAIRBENCH leaderboard, result summary, and performance analysis for whisper-small. WER, Whisper-tiny demonstrates better overall fairness. Finetuned Wav2Vec and Hubert models show efficiency but lag in fairness. While FAAS generally follows WER trends, the comparison between fine-tuned Wav2Vec-large and Hubert-large reveals that fairness issues can outweigh slight WER improvements—emphasizin… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

11 extracted references · 7 canonical work pages

[1]

However, despite impressive strides in overall accuracy, these systems of- ten exhibit significant performance disparities across diverse demographic groups

Introduction Automatic Speech Recognition (ASR) systems have revolution- ized human-computer interaction, powering applications from virtual assistants to real-time transcription services. However, despite impressive strides in overall accuracy, these systems of- ten exhibit significant performance disparities across diverse demographic groups. Variations...

work page
[2]

Dataset A stratified 10% sample was extracted from the fair speech dataset to benchmark ASR models on the ASR-FAIRBENCH leaderboard

Methodology 2.1. Dataset A stratified 10% sample was extracted from the fair speech dataset to benchmark ASR models on the ASR-FAIRBENCH leaderboard. As shown in Table 1, the entropy values across all demographic attributes remain nearly identical between the original and sampled datasets, ensuring that the distribution is preserved while reducing the eva...

arXiv 2025
[3]

Built with React.js, leveraging an NVIDIA T4 GPU for inference, it of- fers an interactive platform for model submissions, performance analysis, and real-time leaderboard tracking

Application overview The ASR-FAIRBENCH leaderboard is a web-based platform designed for evaluating and ranking ASR models. Built with React.js, leveraging an NVIDIA T4 GPU for inference, it of- fers an interactive platform for model submissions, performance analysis, and real-time leaderboard tracking. The platform is fully reproducible from its repositor...

work page
[4]

Despite higher 1https://huggingface.co/spaces/satyamr196/ASR-FairBench 2https://github.com/SatyamR196/ASR-FairBench (a) ASR-F AIRBENCHleaderboard (b) Summarized results view

Conclusion Our ASR-FAIRBENCH leaderboard (Figure 1a) highlights Whisper models as leading in fairness-adjusted ASR scores, with Whisper-medium scoring highest at 29.41. Despite higher 1https://huggingface.co/spaces/satyamr196/ASR-FairBench 2https://github.com/SatyamR196/ASR-FairBench (a) ASR-F AIRBENCHleaderboard (b) Summarized results view. Figure 1: Key...

work page
[5]

Effects of talker dialect, gender & race on accuracy of bing speech and youtube automatic captions

R. Tatman and C. Kasten, “Effects of talker dialect, gender & race on accuracy of bing speech and youtube automatic captions.” in Interspeech, 2017

work page 2017
[6]

Quan- tifying bias in automatic speech recognition,

S. Feng, O. Kudina, B. M. Halpern, and O. Scharenborg, “Quan- tifying bias in automatic speech recognition,” arXiv preprint arXiv:2103.15122, 2021

arXiv 2021
[7]

A deep dive into the disparity of word error rates across thousands of nptel mooc videos,

A. K. Rai, S. D. Jaiswal, and A. Mukherjee, “A deep dive into the disparity of word error rates across thousands of nptel mooc videos,” in AAAI ICWSM, 2024

work page 2024
[8]

Open universal arabic asr leaderboard,

Y . Wang, A. Alhmoud, and M. Alqurishi, “Open universal arabic asr leaderboard,” arXiv preprint arXiv:2412.13788, 2024

arXiv 2024

Show all 11 references

[9]

Open automatic speech recognition leaderboard,

V . Srivastav, S. Majumdar, N. Koluguri, A. Moumen, S. Gandhi, H. F. Team, N. N. Team, and S. Team, “Open automatic speech recognition leaderboard,” urlhttps://huggingface.co/spaces/huggingface.co/spaces/open-asr- leaderboard/leaderboard, 2023

2023
[10]

Racial disparities in automated speech recognition,

A. Koenecke, A. Nam, E. Lake, J. Nudell, M. Quartey, Z. Menge- sha, C. Toups, J. R. Rickford, D. Jurafsky, and S. Goel, “Racial disparities in automated speech recognition,” Proceedings of the national academy of sciences , vol. 117, no. 14, pp. 7684–7689, 2020

2020
[11]

Towards measuring fairness in speech recognition: Fair-speech dataset,

I.-E. Veliche, Z. Huang, V . A. Kochaniyan, F. Peng, O. Kalinli, and M. L. Seltzer, “Towards measuring fairness in speech recognition: Fair-speech dataset,” arXiv preprint arXiv:2408.12734, 2024

2024 arXiv

Pith tools

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

[1] [1]

However, despite impressive strides in overall accuracy, these systems of- ten exhibit significant performance disparities across diverse demographic groups

Introduction Automatic Speech Recognition (ASR) systems have revolution- ized human-computer interaction, powering applications from virtual assistants to real-time transcription services. However, despite impressive strides in overall accuracy, these systems of- ten exhibit significant performance disparities across diverse demographic groups. Variations...

work page

[2] [2]

Dataset A stratified 10% sample was extracted from the fair speech dataset to benchmark ASR models on the ASR-FAIRBENCH leaderboard

Methodology 2.1. Dataset A stratified 10% sample was extracted from the fair speech dataset to benchmark ASR models on the ASR-FAIRBENCH leaderboard. As shown in Table 1, the entropy values across all demographic attributes remain nearly identical between the original and sampled datasets, ensuring that the distribution is preserved while reducing the eva...

arXiv 2025

[3] [3]

Built with React.js, leveraging an NVIDIA T4 GPU for inference, it of- fers an interactive platform for model submissions, performance analysis, and real-time leaderboard tracking

Application overview The ASR-FAIRBENCH leaderboard is a web-based platform designed for evaluating and ranking ASR models. Built with React.js, leveraging an NVIDIA T4 GPU for inference, it of- fers an interactive platform for model submissions, performance analysis, and real-time leaderboard tracking. The platform is fully reproducible from its repositor...

work page

[4] [4]

Despite higher 1https://huggingface.co/spaces/satyamr196/ASR-FairBench 2https://github.com/SatyamR196/ASR-FairBench (a) ASR-F AIRBENCHleaderboard (b) Summarized results view

Conclusion Our ASR-FAIRBENCH leaderboard (Figure 1a) highlights Whisper models as leading in fairness-adjusted ASR scores, with Whisper-medium scoring highest at 29.41. Despite higher 1https://huggingface.co/spaces/satyamr196/ASR-FairBench 2https://github.com/SatyamR196/ASR-FairBench (a) ASR-F AIRBENCHleaderboard (b) Summarized results view. Figure 1: Key...

work page

[5] [5]

Effects of talker dialect, gender & race on accuracy of bing speech and youtube automatic captions

R. Tatman and C. Kasten, “Effects of talker dialect, gender & race on accuracy of bing speech and youtube automatic captions.” in Interspeech, 2017

work page 2017

[6] [6]

Quan- tifying bias in automatic speech recognition,

S. Feng, O. Kudina, B. M. Halpern, and O. Scharenborg, “Quan- tifying bias in automatic speech recognition,” arXiv preprint arXiv:2103.15122, 2021

arXiv 2021

[7] [7]

A deep dive into the disparity of word error rates across thousands of nptel mooc videos,

A. K. Rai, S. D. Jaiswal, and A. Mukherjee, “A deep dive into the disparity of word error rates across thousands of nptel mooc videos,” in AAAI ICWSM, 2024

work page 2024

[8] [8]

Open universal arabic asr leaderboard,

Y . Wang, A. Alhmoud, and M. Alqurishi, “Open universal arabic asr leaderboard,” arXiv preprint arXiv:2412.13788, 2024

arXiv 2024