Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Evaluating Logit-Based GOP Scores for Mispronunciation Detection

T0 review · 4 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 Logit-based GOP scores improve mispronunciation detection, with the maximum-logit variant correlating best with human perception.

desk verdict A useful, honest empirical comparison of logit-based GOP variants, but the headline MaxLogit–human correlation result is presented without uncertainty quantification and is selected on the test set. read the letter →

arxiv 2506.12067 v2 pith:J4IZTHPH submitted 2025-06-02 eess.AS cs.AIcs.SD

Aditya Kamlesh Parikh , Cristian Tejedor-Garcia , Catia Cucchiarini , Helmer Strik This is my paper
classification eess.AScs.AIcs.SD
keywords goodnessofpronunciationGOPlogit-basedmispronunciationdetectionassessmentCTCphonemerecognitionsoftmaxposteriorprobabilitiesL2Englishspeech
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

This paper tests whether goodness-of-pronunciation (GOP) scores computed from raw logits, instead of softmax posteriors, improve mispronunciation detection and agreement with human raters. On two L2 English datasets — Dutch children and Mandarin speakers — logit-based scores generally outperform probability-based GOP in classification, but the best score differs by dataset. The maximum-logit GOP correlates most strongly with human phoneme accuracy ratings on SpeechOcean762, while margin-based GOP performs best on the Dutch children's corpus. If the results hold, pronunciation-training systems could use a fast, non-trainable score that captures perceived pronunciation quality better than softmax-based GOP.

What carries the argument

The machinery is a family of four GOP scores defined directly on raw logits: GOPMaxLogit (the maximum target-phoneme logit over aligned frames), GOPMargin (the mean difference between the target logit and the strongest competing logit), GOPVarLogit (the variance of target logits over the segment), and GOPCombined (a weighted sum of GOPMargin and the softmax GOPDNN). These scores are extracted after CTC-segmentation forced alignment with a fine-tuned wav2vec2 phoneme recognizer, so they need no extra training or parameters. The load-bearing mechanism is that raw logits preserve the model's unnormalized confidence, avoiding the overconfidence and gradient saturation of softmax normalization, which yields better separation between correct and mispronounced phonemes.

What would settle it

On an independently annotated corpus of naturally mispronounced L2 speech, compute the Pearson correlations of GOPMaxLogit and GOPDNN with human phoneme ratings; if GOPMaxLogit does not exceed GOPDNN's correlation, the paper's main perceptual-alignment claim would be overturned.

Watch

Extended reading notes

Core claim

The paper's central claim is that raw logits carry more discriminative information than softmax probabilities for scoring pronunciation, and that logit-based GOP scores are a viable non-trainable alternative for mispronunciation detection. Across the two corpora, logit-based scores beat probability-based GOP in classification, and GOPMaxLogit gives the strongest correlation with human phoneme accuracy judgments on SpeechOcean762, with Pearson correlations of 0.442 for low-confidence and 0.456 for high-confidence phoneme ratings. On the MPC corpus, GOPMargin wins on accuracy and MCC, illustrating that dataset acoustic variability determines which logit score is best. The paper also finds that softmax GOP over-detects mispronunciations with high recall but low precision, and correlates poorly with human ratings, while a hybrid of margin and softmax scores balances the two families.

Load-bearing premise

The MPC results hinge on the assumption that simulated pronunciation errors made by rule-based phoneme substitutions (such as /D/ to /d/ and diphthong simplification) behave like real L2 mispronunciations; if simulated errors are easier or harder to detect, the reported advantages of logit-based scores on MPC may not generalize to natural speech.

Editorial extensions

If this is right

  • Any CTC-based acoustic model can supply logit-based GOP scores without retraining or added parameters, making the approach fast and model-agnostic.
  • Logit-based scores separate correct from mispronounced phonemes more cleanly than GOPDNN, which over-detects errors, so they should improve precision in pronunciation-training feedback.
  • Maximum-logit GOP tracks human phoneme accuracy ratings more closely than probability-based GOP, making it a better target for perceptually meaningful pronunciation scores.
  • The best logit score is dataset-dependent, so deployment across learner populations should select or combine scores rather than assume one fixed metric.

Reading between the lines

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

  • The dataset dependence suggests a natural next step: a phoneme-specific or uncertainty-weighted combination of margin and max-logit scores may beat any single variant on both classification and human correlation.
  • Because MPC mispronunciations were simulated by rule-based phoneme substitutions, the reported improvements there should be revalidated on naturally occurring L2 errors, which may be harder to separate.
  • The gap between classification accuracy and human correlation implies that pronunciation-assessment systems should be evaluated on both; optimizing for classification alone can yield scores that agree with expert perception less well than a slightly weaker classifier.
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

4 major / 5 minor

Summary. The paper compares probability-based GOP (GOPDNN) with four logit-based GOP variants—GOPMaxLogit, GOPMargin, GOPVarLogit, and a hybrid GOPCombined—for phoneme-level mispronunciation detection and pronunciation assessment. Experiments are run on two L2 English datasets: MPC (Dutch children; simulated phoneme-level errors) and SpeechOcean762 (Mandarin speakers; human phoneme accuracy annotations). Classification is evaluated with accuracy, precision, recall, F1, MCC, and ROC AUC, and correlation with human ratings is measured via PCC and MSE after a second-order polynomial mapping. The main claims are that logit-based methods improve classification on MPC, GOPMaxLogit most strongly aligns with human perception on SpeechOcean762, and GOPCombined balances probability and logit information.

Significance. If the empirical claims hold, the contribution is practically valuable: a simple, non-trainable, model-agnostic family of GOP scores computable from any CTC-based acoustic model, with open-source code and evaluation on a public benchmark (SpeechOcean762) plus an in-house dataset. The paper also raises an interesting point about softmax overconfidence in GOP. However, the central empirical claims currently rest on test-set threshold optimization, winner selection on the test set without uncertainty quantification, and simulated errors in MPC. These issues must be resolved before the abstract-level claims about classification superiority and perceptual alignment are fully supported.

major comments (4)
  1. [Section 2.5 / Tables 1-2] The classification thresholds are selected by maximizing MCC on the same test set used to report all classification results (Section 2.5: 'we optimized the GOP threshold by selecting the percentile that maximized MCC'). This procedure can inflate MCC for every method and, more importantly, can differentially favor methods whose score distributions happen to provide a convenient percentile. Because the abstract claim 'Logit-based methods outperform probability-based GOP in classification' is grounded in Tables 1 and 2, the threshold-selection procedure is load-bearing. Please use a held-out development set for threshold selection, or otherwise report the chosen threshold and evaluate at a fixed or pre-registered threshold, and add an estimate of optimism (e.g., cross-validated MCC).
  2. [Section 3 / Table 2] The central perceptual-alignment result—GOPMaxLogit achieves the highest PCC with human phoneme accuracy ratings (0.442 low confidence, 0.456 high confidence)—is a winner-take-all selection made on the same test set used to compute the correlations. No confidence intervals, significance tests, or multiple-comparison corrections are reported, so the observed gaps over GOPDNN (0.278/0.295) and GOPVarLogit (0.341/0.357) could be within sampling noise. Please provide bootstrap confidence intervals for each PCC and, given the correlations are computed on the same utterances, a test for the difference between correlated correlation coefficients (e.g., a Williams test) or a held-out model-selection step.
  3. [Section 2.4 / Table 1] Because MPC lacks annotated mispronunciations, the authors 'introduced simulated pronunciation errors by modifying phoneme sequences,' including substitutions such as /D/ → /d/, /T/ → /s/, and diphthong simplification. The classification comparison on MPC—where logit-based GOPMargin achieves the largest MCC—therefore depends on the assumption that these synthetic substitutions are representative of natural L2 errors. If such substitutions introduce stronger or weaker acoustic cues than natural mispronunciations, the reported improvements may not generalize. Please validate the simulated errors (e.g., through human listening checks or comparison with naturally occurring L2 errors) or explicitly reframe the MPC results as a synthetic stress test rather than evidence for real-world classification gains.
  4. [Section 2.2.4 / Eq. (6)] GOPCombined has a free parameter α ∈ [0,1], but the value used to produce the results in Tables 1 and 2 is never reported, nor is the selection procedure for α described. Since the abstract and discussion highlight GOPCombined as a balancing hybrid, the reader cannot assess whether its reported performance comes from the proposed combination or from an α tuned in hindsight on the test data. Please report α for each dataset and provide either a sensitivity analysis across α or a principled selection rule (e.g., development-set optimization).
minor comments (5)
  1. [Section 4] The discussion states that GOPMaxLogit 'achiev[es] the highest PCC scores (Table 1)', but the PCC results appear in Table 2; this cross-reference should be corrected.
  2. [Section 2.2 / Eq. (3)] The notation l_t^{(p)} in Eq. (3) is not defined precisely; please state explicitly that it is the raw pre-softmax logit for phoneme p at frame t, and align this notation with the usage in Eqs. (4)–(6).
  3. [Section 2.5] The threshold-optimization description is vague about the search space: please specify the percentile grid and the step size used when 'selecting the percentile that maximized MCC.'
  4. [Abstract / Section 2.2.4] The abstract claims that the findings 'suggest that hybrid GOP methods incorporating uncertainty modeling and phoneme-specific weighting improve pronunciation assessment,' but the paper does not implement phoneme-specific weighting; either soften this claim or add the corresponding analysis.
  5. [Global] The dataset name is inconsistently capitalized as 'Speechocean762' in some places and 'SpeechOcean762' in others; please standardize throughout, including table captions and figure labels.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the GOP comparisons are empirical and each metric is defined independently of the human ratings; test-set threshold/alpha fitting create overfitting concerns but not circular construction.

full rationale

The paper's claims are empirical comparisons of GOP scoring functions applied to external corpora (MPC and SpeechOcean762); there is no formal derivation that presupposes its conclusions. The GOP equations (Eqs. 1-6) are definitions, not derived from human ratings or from each other: GOPMaxLogit is a frame-wise max of raw logits, GOPMargin a mean margin, GOPVarLogit a variance, and GOPCombined a convex combination of GOPMargin and GOPDNN. Human phoneme accuracy labels in SpeechOcean762 are independent of the model logits. The only fitted quantities are the classification threshold (chosen to maximize MCC on the test set, Sec. 2.5), the alpha in GOPCombined (Eq. 6, unreported), and the second-order polynomial used to compute PCC/MSE; these are evaluation-protocol choices, not prediction-equivalent constructions. They create overfitting/selection concerns (winner PCC of 0.442/0.456 is selected post hoc on the same test set without confidence intervals), but they do not make any reported result identical to its input by construction. The self-citations (refs. 10, 11, 30) are background/dataset references and are not load-bearing. Therefore no significant circularity.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

No invented entities. The central claim rests on the reliability of the acoustic model and the validity of simulated errors in MPC.

free parameters (2)
  • alpha (GOPCombined weight) = not reported
    Eq. 6 weights the logit margin against GOPDNN; the paper never states the value used, and results may depend on it.
  • MCC-optimized threshold percentile = dataset-specific
    Section 2.5: threshold chosen to maximize MCC on the evaluation data, inflating reported metrics.
assumptions (3)
  • domain assumption CTC segmentation gives accurate phoneme boundaries
    Section 2.3: forced alignment via CTC-segmentation; the authors acknowledge alignment errors for child/non-native speech in Section 4.
  • ad hoc to paper Simulated phoneme substitutions approximate real errors
    Section 2.4: MPC has no mispronunciation annotations, so errors are artificially generated; representatives of natural errors is unverified.
  • domain assumption Fine-tuned wav2vec2 logits are a reliable phoneme likelihood source
    Section 2.3: uses an open-source model without calibration or analysis of logit distributions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evaluating Logit-Based GOP Scores for Mispronunciation Detection." pith.science (2026). https://pith.science/paper/J4IZTHPH

@misc{pith2026250612067,
  author       = {Pith},
  title        = {Pith review of: Evaluating Logit-Based GOP Scores for Mispronunciation Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J4IZTHPH}},
  note         = {Machine review of arXiv:2506.12067}
}
read the original abstract

Pronunciation assessment relies on goodness of pronunciation (GOP) scores, traditionally derived from softmax-based posterior probabilities. However, posterior probabilities may suffer from overconfidence and poor phoneme separation, limiting their effectiveness. This study compares logit-based GOP scores with probability-based GOP scores for mispronunciation detection. We conducted our experiment on two L2 English speech datasets spoken by Dutch and Mandarin speakers, assessing classification performance and correlation with human ratings. Logit-based methods outperform probability-based GOP in classification, but their effectiveness depends on dataset characteristics. The maximum logit GOP shows the strongest alignment with human perception, while a combination of different GOP scores balances probability and logit features. The findings suggest that hybrid GOP methods incorporating uncertainty modeling and phoneme-specific weighting improve pronunciation assessment.

Figures

Figures reproduced from arXiv: 2506.12067 by the authors.

Figure 1
Figure 1. Comparison of mispronunciation error rates by phoneme (Max Logit vs. human rater) in the SpeechOcean762 dataset. Blue bars show GOPMaxLogit-predicted error rates, while red bars indicate human-rated phoneme accuracy. logit-based approaches across both datasets, reducing overlap compared to GOPDNN. It presents a trade-off between classifi￾cation accuracy from posterior probabilities and human correla￾tion from logit-… view at source ↗
Figure 2
Figure 2. Comparison of GOP score distributions across MPC and Speechocean762 datasets. Finally, to analyze the alignment between GOP-based mis￾pronunciation detection and human-rated phoneme accuracy, we investigated whether the GOP scoring method with the high￾est correlation to human ratings, GOPMaxLogit, effectively iden￾tifies mispronounced phonemes [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

39 extracted references · 35 canonical work pages

  1. [1]

    Evaluating Logit-Based GOP Scores for Mispronunciation Detection

    Introduction In today’s interconnected world, globalization has led to in- creased movement across borders for work, education, and other opportunities. For individuals who are adapting to a new linguistic environment, learning the local language is essential for social integration, career advancement, and overall well- being [1]. Effective communication ...

    work page Pith review arXiv 2025

  2. [2]

    Methodology 2.1. Definition of GOP The GOP score, first introduced by Witt and Young [9], quanti- fies pronunciation quality by comparing the likelihood of a hy- pothesized phoneme to competing alternatives. For a phoneme p aligned to an audio segment, the original GOP formulation computes: GOPoriginal(p) = log P (X|p) 1 N P q∈Q P (X|q) (1) where P (X|p) ...

    work page

  3. [3]

    Of all these measures, GOP Margin achieves the highest accuracy (0.851), MCC (0.347)

    Results Table 1 presents the evaluation scores for posterior probabil- ity based (first column), logit-based (second to fourth columns) and hybrid GOP scores (last column) on the MPC dataset. Of all these measures, GOP Margin achieves the highest accuracy (0.851), MCC (0.347). It also outperforms other approaches in F1-score (0.415) and precision (0.347),...

    work page

  4. [4]

    Discussion and Conclusion In this work, we have analyzed differences in probability-based and logit-based GOP for pronunciation assessment across two datasets, MPC and SpeechOcean762. To answer our RQ, our findings indicate that logit-based methods achieve a better classification performance than probability-based GOP; how- ever, their effectiveness depen...

    work page

  5. [5]

    Acknowledgements This publication is part of the project Responsible AI for V oice Diagnostics (RAIVD) with file number NGF.1607.22.013 of the research programme NGF AiNed Fellowship Grants which is financed by the Dutch Research Council (NWO)

    work page

  6. [6]

    Combin- ing language training and work experience for refugees with low- literacy levels: a mixed-methods case study,

    A. Kuschel, N. Hansen, L. Heyse, and R. P. Wittek, “Combin- ing language training and work experience for refugees with low- literacy levels: a mixed-methods case study,” Journal of Interna- tional Migration and Integration, vol. 24, no. 4, pp. 1635–1661, 2023

    work page 2023

  7. [7]

    English pronunciation for a global world,

    R. Walker, E.-L. Low, and J. Setter, “English pronunciation for a global world,” Oxford, October 2021, Last visited: 2025-02-10. [Online]. Available: https://centaur.reading.ac.uk/101017/

    work page 2021

  8. [8]

    Jenkins, The phonology of English as an international lan- guage

    J. Jenkins, The phonology of English as an international lan- guage. Oxford University Press, 2000

    work page 2000

Show all 39 references
  1. [9]

    What do adult english learners say about their pronuncia- tion and linguistic self-confidence?

    J. Zoss, “What do adult english learners say about their pronuncia- tion and linguistic self-confidence?” MinneTESOL Journal, 2016

    2016

  2. [10]

    An examination of automatic speech recogni- tion (asr)-based computer-assisted pronunciation training (capt) for less-proficient efl students using the technology acceptance model,

    H.-W. Hsu, “An examination of automatic speech recogni- tion (asr)-based computer-assisted pronunciation training (capt) for less-proficient efl students using the technology acceptance model,” International Journal of Technology in Education, vol. 7, no. 3, pp. 456–473, 2024

    2024

  3. [11]

    Computer-assisted pronunciation training: A systematic review,

    M. Amrate and P. hua Tsai, “Computer-assisted pronunciation training: A systematic review,”ReCALL, no. 1, pp. 22–42, 2024

    2024

  4. [12]

    Improving mis- pronunciation detection and diagnosis for non-native learners of the arabic language,

    N. Alrashoudi, H. Al-Khalifa, and Y . Alotaibi, “Improving mis- pronunciation detection and diagnosis for non-native learners of the arabic language,” Discover Computing, vol. 28, no. 1, p. 1, 2025

    2025

  5. [13]

    A Framework for Phoneme-Level Pronunciation Assessment Using CTC,

    X. Cao, Z. Fan, T. Svendsen, and G. Salvi, “A Framework for Phoneme-Level Pronunciation Assessment Using CTC,” in Inter- speech 2024, 2024, pp. 302–306

    2024

  6. [14]

    Use of speech recognition in computer-assisted lan- guage learning

    S. M. Witt, “Use of speech recognition in computer-assisted lan- guage learning.” Ph.D. dissertation, University of Cambridge, 2000

    2000

  7. [15]

    The goodness of pro- nunciation algorithm: a detailed performance study,

    S. Kanters, C. Cucchiarini, and H. Strik, “The goodness of pro- nunciation algorithm: a detailed performance study,” in Speech and Language Technology in Education (SLaTE 2009), 2009, pp. 49–52

    2009

  8. [16]

    Using non-native error patterns to improve pronunciation verification,

    J. van Doremalen, C. Cucchiarini, and H. Strik, “Using non-native error patterns to improve pronunciation verification,” in Inter- speech 2010, 2010, pp. 590–593

    2010

  9. [17]

    Lattice-based gop in automatic pronunciation evaluation,

    Y . Song, W. Liang, and R. Liu, “Lattice-based gop in automatic pronunciation evaluation,” in2010 The 2nd International Confer- ence on Computer and Automation Engineering (ICCAE), vol. 3, 2010, pp. 598–602

    2010

  10. [18]

    Context-aware goodness of pronuncia- tion for computer-assisted pronunciation training,

    J. Shi, N. Huo, and Q. Jin, “Context-aware goodness of pronuncia- tion for computer-assisted pronunciation training,” in Interspeech 2020, 2020, pp. 3057–3061

    2020

  11. [19]

    Acoustic feature mixup for balanced multi-aspect pronunciation assessment,

    H. Do, W. Lee, and G. G. Lee, “Acoustic feature mixup for balanced multi-aspect pronunciation assessment,” CoRR, vol. abs/2406.15723, 2024. [Online]. Available: https://doi.org/10. 48550/arXiv.2406.15723

    2024 doi arXiv

  12. [20]

    Transformer-based multi-aspect multi-granularity non-native en- glish speaker pronunciation assessment,

    Y . Gong, Z. Chen, I.-H. Chu, P. Chang, and J. Glass, “Transformer-based multi-aspect multi-granularity non-native en- glish speaker pronunciation assessment,” in ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022, pp. 7262–7266

    2022

  13. [21]

    Mispronunciation detection with speechblender data augmentation pipeline,

    Y . El Kheir, “Mispronunciation detection with speechblender data augmentation pipeline,” PhD Thesis Report, KTH Royal Institute of Technology, Stockholm, Sweden, 2023, Last visited: 2025-02-

    2023

  14. [22]

    Available: https://urn.kb.se/resolve?urn=urn:nbn:se: kth:diva-339940

    [Online]. Available: https://urn.kb.se/resolve?urn=urn:nbn:se: kth:diva-339940

  15. [23]

    Xls-r: Self-supervised cross-lingual speech representation learning at scale,

    A. Babu, C. Wang, A. Tjandra, K. Lakhotia, Q. Xu, N. Goyal, K. Singh, P. von Platen, Y . Saraf, J. Pino, A. Baevski, A. Con- neau, and M. Auli, “Xls-r: Self-supervised cross-lingual speech representation learning at scale,” in Interspeech 2022, 2022, pp. 2278–2282

    2022

  16. [24]

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

    W.-N. Hsu, B. Bolte, Y .-H. H. Tsai, K. Lakhotia, R. Salakhut- dinov, and A. Mohamed, “Hubert: Self-supervised speech representation learning by masked prediction of hidden units,” IEEE/ACM Trans. Audio, Speech and Lang. Proc. , vol. 29, p. 3451–3460, Oct. 2021. [Online]. Avai...

    2021

  17. [25]

    An improved goodness of pronunciation (gop) measure for pronunci- ation evaluation with dnn-hmm system considering hmm transi- tion probabilities

    S. Sudhakara, M. K. Ramanathi, C. Yarra, and P. K. Ghosh, “An improved goodness of pronunciation (gop) measure for pronunci- ation evaluation with dnn-hmm system considering hmm transi- tion probabilities.” in INTERSPEECH, vol. 2, 2019, pp. 954–958

    2019

  18. [26]

    Regularizing neural networks by penalizing confident output distributions,

    G. Pereyra, G. Tucker, J. Chorowski, L. Kaiser, and G. Hinton, “Regularizing neural networks by penalizing confident output distributions,” 2017. [Online]. Available: https://openreview.net/ forum?id=HkCjNI5ex

    2017

  19. [27]

    Mitigating neural network overconfidence with logit normalization,

    H. Wei, R. Xie, H. Cheng, L. Feng, B. An, and Y . Li, “Mitigating neural network overconfidence with logit normalization,” in In- ternational conference on machine learning. PMLR, 2022, pp. 23 631–23 644

    2022

  20. [28]

    Comparing non-native and native speech: Are l2 productions more variable?

    X. Xie and T. F. Jaeger, “Comparing non-native and native speech: Are l2 productions more variable?” The Journal of the Acoustical Society of America, vol. 147, no. 5, pp. 3322–3347, 2020

    2020

  21. [29]

    Perception of speech sounds in school-aged children with speech sound disorders,

    J. L. Preston, J. R. Irwin, and J. Turcios, “Perception of speech sounds in school-aged children with speech sound disorders,” in Seminars in speech and language, vol. 36, no. 04. Thieme Med- ical Publishers, 2015, pp. 224–233

    2015

  22. [30]

    On the learning property of lo- gistic and softmax losses for deep neural networks,

    X. Li, X. Li, D. Pan, and D. Zhu, “On the learning property of lo- gistic and softmax losses for deep neural networks,” in Proceed- ings of the AAAI Conference on Artificial Intelligence , vol. 34, no. 04, 2020, pp. 4739–4746

    2020

  23. [31]

    Logit margin mat- ters: Improving transferable targeted adversarial attack by logit calibration,

    J. Weng, Z. Luo, S. Li, N. Sebe, and Z. Zhong, “Logit margin mat- ters: Improving transferable targeted adversarial attack by logit calibration,” IEEE Transactions on Information Forensics and Se- curity, vol. 18, pp. 3561–3574, 2023

    2023

  24. [32]

    Speech intelligibility assessment of dysarthric speech by using goodness of pronunci- ation with uncertainty quantification,

    E. J. Yeo, K. Choi, S. Kim, and M. Chung, “Speech intelligibility assessment of dysarthric speech by using goodness of pronunci- ation with uncertainty quantification,” in Interspeech 2023, 2023, pp. 166–170

    2023

  25. [33]

    A new DNN-based high qual- ity pronunciation evaluation for computer-aided language learn- ing (CALL),

    W. Hu, Y . Qian, and F. K. Soong, “A new DNN-based high qual- ity pronunciation evaluation for computer-aided language learn- ing (CALL),” in Interspeech, 2013, pp. 1886–1890

    2013

  26. [34]

    Ctc-segmentation of large corpora for german end-to-end speech recognition,

    L. K ¨urzinger, D. Winkelbauer, L. Li, T. Watzel, and G. Rigoll, “Ctc-segmentation of large corpora for german end-to-end speech recognition,” in International Conference on Speech and Com- puter. Springer, 2020, pp. 267–278

    2020

  27. [35]

    Simple and effective zero-shot cross-lingual phoneme recognition,

    Q. Xu, A. Baevski, and M. Auli, “Simple and effective zero-shot cross-lingual phoneme recognition,” in Interspeech 2022, 2022, pp. 2113–2117

    2022

  28. [36]

    My pronunciation coach: Improving english pronunciation with an automatic coach that listens,

    C. Cucchiarini, W. Nejjari, and H. Strik, “My pronunciation coach: Improving english pronunciation with an automatic coach that listens,” Language Learning in Higher Education , vol. 1, no. 2, pp. 365–376, 2012

    2012

  29. [37]

    speechocean762: An open-source non- native english speech corpus for pronunciation assessment,

    J. Zhang, Z. Zhang, Y . Wang, Z. Yan, Q. Song, Y . Huang, K. Li, D. Povey, and Y . Wang, “speechocean762: An open-source non- native english speech corpus for pronunciation assessment,” inIn- terspeech 2021, 2021, pp. 3710–3714

    2021

  30. [38]

    Etlt 2021: Shared task on automatic speech recognition for non-native children’s speech,

    R. Gretter, M. Matassoni, D. Falavigna, A. Misra, C. Leong, K. Knill, and L. Wang, “Etlt 2021: Shared task on automatic speech recognition for non-native children’s speech,” in Inter- speech 2021, 2021, pp. 3845–3849

    2021

  31. [39]

    A hi- erarchical context-aware modeling approach for multi-aspect and multi-granular pronunciation assessment,

    F.-A. Chao, T.-H. Lo, T.-I. Wu, Y .-T. Sung, and B. Chen, “A hi- erarchical context-aware modeling approach for multi-aspect and multi-granular pronunciation assessment,” in Interspeech 2023 , 2023, pp. 974–978

    2023

Pith tools

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