Pith. sign in

REVIEW 4 major objections 6 minor 26 references

IIITH-BUT system for IWSLT 2025 low-resource Bhojpuri to Hindi speech translation

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

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

Pith's one-line read Careful fine-tuning lifts low-resource speech translation to 36.4 BLEU

desk verdict A competent, honest IWSLT system description whose headline claim about hyperparameter and augmentation gains rests on dev-set scores that collapse on test. read the letter →

arxiv 2506.04714 v1 pith:5J4ZNCOK submitted 2025-06-05 cs.CL eess.AS

Bhavana Akkiraju , Aishwarya Pothula , Santosh Kesiraju , Anil Kumar Vuppala This is my paper
classification cs.CLeess.AS
keywords Bhojpuri-Hindispeechtranslationlow-resourceSeamlessM4Tfine-tuningSpecAugmentspeedperturbationhyperparameteroptimizationcross-lingualtransferIWSLT2025
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 argues that for low-resource Bhojpuri-to-Hindi speech translation, careful choice of fine-tuning hyperparameters plus simple audio augmentation delivers large gains over default settings, and that a mid-sized model can outperform a larger one. Fine-tuning the SeamlessM4T medium model with a moderate learning rate, batch size 32, label smoothing 0.1, longer warmup, and SpecAugment reaches 36.4 BLEU on the development set, while the same configuration reaches only 9.9 BLEU on the official test set. A contrastive model jointly fine-tuned on Marathi-Hindi and Bhojpuri-Hindi data, then adapted to Bhojpuri for one epoch, scores 36.0 on dev and 10.2 on test. The paper interprets the large dev-test gap as evidence of domain mismatch or test-data quality issues rather than a failure of the method.

What carries the argument

The load-bearing machinery is the SeamlessM4T medium model (a 1.2B-parameter end-to-end speech-to-text translation model with a 24-layer conformer speech encoder and 12-layer transformer decoder), fine-tuned on the official 20-hour Bhojpuri-Hindi training set. Two audio augmentations carry the data-scarcity argument: SpecAugment, which masks random time and frequency spans of the spectrogram, and speed perturbation, which resamples audio at 0.9x, 1.0x, and 1.1x rates to triple the training data. A sequential cross-lingual fine-tuning procedure, joint training on Marathi-Hindi plus Bhojpuri-Hindi followed by Bhojpuri-only adaptation, supplies the transfer-learning contrast. Hyperparameter choices are made by grid search on development BLEU, with batch size, learning rate, label smoothing, warmup steps, patience, and beam size varied independently.

What would settle it

Take the exact primary configuration and evaluate it on an independently collected Bhojpuri-Hindi test set matched to the training domain; if BLEU stays near 10 rather than near 36, the dev-based hyperparameter selection does not transfer. Alternatively, compare the reported configuration against a small random hyperparameter search on the same test set: if random configurations match or beat the reported 9.9 BLEU within noise, the claimed optimization is not load-bearing.

Watch

Extended reading notes

Core claim

The paper's central claim is that systematic hyperparameter optimization and established augmentation techniques, applied to a pre-trained end-to-end speech translation model, substantially improve translation quality in a low-resource setting. With SeamlessM4T medium as the backbone, the best configuration uses a learning rate of 1e-5, batch size 32, label smoothing 0.1, 250 warmup steps, early stopping patience 10, beam size 10, and SpecAugment, yielding 36.41 BLEU on the development set; the larger 2.3B model underperforms the 1.2B medium model (25.5 versus 30.5 BLEU). Speed perturbation alone gives a smaller gain and combining it with SpecAugment slightly hurts, so the primary submission uses SpecAugment only. A contrastive system fine-tuned jointly on Marathi-Hindi and Bhojpuri-Hindi, followed by one epoch of Bhojpuri-only fine-tuning, achieves 36.0 dev BLEU and outperforms the primary on the test set (10.2 versus 9.9 BLEU). The paper also reports that the test set scores for all systems are far below dev scores and ties this to domain mismatch and data quality issues rather than to the training procedure.

Load-bearing premise

The load-bearing premise is that the 2.07-hour official development set is representative enough that hyperparameter choices made on it will transfer to the test set, even though the selected system scores 36.4 on dev but only 9.9 on test.

Editorial extensions

If this is right

  • If the dev-set results transfer, low-resource speech translation can be improved by roughly 4-5 BLEU points purely through hyperparameter selection and SpecAugment, without any new data collection.
  • The medium SeamlessM4T variant is preferable to the large variant for a 20-hour low-resource fine-tuning scenario, suggesting that over-parameterization hurts when data is scarce.
  • Cross-lingual joint fine-tuning with a related Indo-Aryan language (Marathi) plus a short Bhojpuri adaptation phase can match or slightly exceed monolingual fine-tuning on the test set.
  • The dev-test gap reported here implies that evaluation sets for low-resource speech translation can be the dominant factor in measured system quality, and that dev-based optimization alone does not guarantee test-set performance.
  • The paper's error analysis shows that audio-transcript misalignment and inconsistent numerical representation in reference texts can depress BLEU independently of model quality.

Reading between the lines

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

  • The 36.4 versus 9.9 dev-test gap is large enough that the absolute numbers should not be read as comparable; if the official test set is noisy or out-of-domain, the reported test scores may mostly reflect dataset artifacts rather than model differences.
  • SpecAugment's benefit is not uniform in the paper's own tables: it helps at batch size 10 but combining it with speed perturbation lowers BLEU relative to SpecAugment alone, so the optimal augmentation choice is likely sensitive to data size and hyperparameters in ways the paper does not fully disentangle.
  • The one-epoch Bhojpuri adaptation after joint Marathi training suggests that catastrophic forgetting is the main risk in cross-lingual transfer, and that the optimal adaptation length is short; a testable extension is to vary the adaptation data size and measure where forgetting begins.
  • Because all hyperparameter choices are made on the 2.07-hour dev set, a fairer evaluation would require held-out dev splits or nested cross-validation to confirm that the selected configuration generalizes beyond that single set.
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 / 6 minor

Summary. The paper describes the IIITH-BUT submission to the IWSLT 2025 low-resource Bhojpuri-to-Hindi speech translation shared task. The authors fine-tune SeamlessM4T (medium and large variants) on the official Bhojpuri-Hindi corpus, systematically explore hyperparameters (learning rate, batch size, label smoothing, warmup steps, early-stopping patience, beam size), test SpecAugment and speed perturbation, and evaluate a contrastive model jointly fine-tuned on Marathi-Hindi followed by Bhojpuri-only adaptation. The primary model achieves 36.4 BLEU on the dev set but 9.9 BLEU on the official test set; the contrastive model achieves 36.0 dev and 10.2 test. The paper also presents a qualitative error analysis of the dev set.

Significance. If the reported improvements were robust across evaluation conditions, the paper would be a useful system description for a very low-resource language pair, showing that careful configuration of a large pre-trained speech translation model and simple data augmentation can yield large gains on the dev distribution. The authors are honest in reporting the large dev-test gap, and the error analysis of numerical inconsistencies and audio-transcript misalignment is informative. However, the headline claims in the abstract are calibrated only to dev-set numbers; the test set shows a rank inversion between the two submissions, and the absence of a test-set baseline means the central generalization claim is not established. With corrections and appropriate caveats, the paper can be made sound for a system-description venue.

major comments (4)
  1. [§5, Table 8] The central claim in the Abstract that "careful selection of hyperparameters and the application of simple yet effective augmentation techniques significantly improve performance in low-resource settings" is supported only by dev-set BLEU. The selected configuration reaches 36.4 on dev but 9.9 on the official test set, while the contrastive model, which ranks lower on dev (36.0), ranks higher on test (10.2). This rank inversion directly contradicts the dev-based optimization criterion and indicates that the dev set is not representative of the test condition. The paper attributes the gap to "potential domain mismatch between the datasets or possible data quality issues" without testing either explanation. To substantiate the headline claim, the authors should either report a default-hyperparameter/no-augmentation baseline on the test set showing that the chosen settings still improve test BLEU, or explicitly limit the claim to the dev distribution and discuss the generalization concern. As written, the abstract overstates what two test-set evaluations can establish.
  2. [Abstract, §4.5, Table 5] The abstract and the contributions bullet state that the augmentation techniques improve BLEU by "an average of 2.1 points" while "expanding our training data by 3x." The 2.1-point average is not recoverable from Table 5: the increments over the no-augmentation baseline (31.8) are +1.9 (SpecAugment only), +0.9 (speed perturbation only), and +0.6 (SA+SP), whose average is 1.13, not 2.1. Furthermore, only speed perturbation triples the data through the three speed factors; SpecAugment does not multiply the corpus size threefold. This is a factual inconsistency in the headline claim and needs correction.
  3. [§4.4, §5] All hyperparameter choices (learning rate, batch size, label smoothing, warmup, patience, beam size, augmentation on/off) are made using a single run on the 2.07-hour dev set, and no test-set ablation is reported. Because the dev-to-test gap (36.4 vs. 9.9) suggests substantial distribution shift, the absence of any test-set comparison means there is no evidence that the selected hyperparameters or augmentation techniques are beneficial on the actual evaluation condition. Adding at least one test-set comparison, such as the no-augmentation or default-configuration model, would make the generalization claim testable and would strengthen the paper substantially.
  4. [§4.4, §5, Tables 5–6] The paper reports no variance, error bars, or repeated runs; on a 2.07-hour dev set, the differences among configurations in Tables 5 and 6 are small (e.g., 33.7 vs. 32.4 for SA alone vs. SA+SP) and may reflect noise. The word "significantly" in the abstract therefore needs statistical support, or it should be replaced with a descriptive term such as "consistently on the dev set." Without this, the strength of the claim exceeds what a single-run dev-set comparison can support.
minor comments (6)
  1. [§5.1] The error analysis lists three qualitative patterns (reference-target word-count mismatches, noisy recordings, numerical inconsistencies) but does not quantify their frequencies; providing counts or proportions would make the analysis more useful.
  2. [§4.1] The text says "We used only the official IWSLT 2025 shared task dataset" but the next sentence introduces an additional Marathi-Hindi corpus; rephrase to avoid ambiguity.
  3. [§4.3, Table 2] The statement that the medium model "consistently yielded better translation results" is based on a single dev-set comparison; either add test-set numbers for both sizes or soften the wording.
  4. [§4.5, Table 5] Table 5 does not report the beam size or early-stopping patience used for the augmentation comparison; please state these settings so the comparison with Table 6 is unambiguous.
  5. [§1] The phrase "The remaining of this paper" should be "The remainder of this paper."
  6. [References] Several reference entries contain formatting errors (e.g., "Popovi ´c" with a stray space, "GowSmith" without a space, "Ashwin" in the IWSLT 2025 author list); please check the references against the venue's style guidelines.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an empirical system description reporting measured BLEU scores on held-out dev and test sets, and no claim reduces to its own inputs by construction.

full rationale

The paper contains no derivation chain whose conclusion is equivalent to its premises. Hyperparameters and augmentation settings are tuned on the development set, and the selected configuration is then evaluated on the official IWSLT test set, with separate dev and test numbers reported in Table 8. The central claim that hyperparameter selection and augmentation improve performance is supported by dev-set comparisons in Tables 3-6; the test result is reported independently and is not used to set any constant. The large dev-to-test drop (36.4 to 9.9) and rank inversion with the contrastive model (36.0 dev, 10.2 test) are concerns about generalization and dev-set overfitting, not circularity. The self-citations to Kesiraju et al. (2023a,b) appear only in the related-work survey and are not load-bearing for any experimental result. The abstract's statement of a 2.1-point average augmentation gain is not fully recoverable from Table 5, but this is an internal reporting inconsistency, not a case of a fitted parameter being renamed as a prediction. No equation, fitting step, or self-citation chain equates an output with an input, so the circularity score is 0.

Assumptions & free parameters 10 free parameters · 5 assumptions · 0 invented entities

The central claims rest on a small officially provided corpus, a pretrained backbone, and several hyperparameters tuned on the dev set. The most fragile inputs are the dev-to-test transfer assumption and the lack of a pretrained-backbone baseline; both are obscured by reporting only fine-tuned dev scores. No new physical or architectural entities are introduced.

free parameters (10)
  • Learning rate = 1e-5
    Selected as best among 1e-6, 1e-5, and 2e-5 on dev BLEU (Section 4.4, Table 3).
  • Label smoothing = 0.1
    Selected as best among 0, 0.1, and 0.2 on dev BLEU (Section 4.4, Table 4).
  • Batch size = 32
    Selected as best among 5, 10, 32, and 64 on dev BLEU (Section 4.4, Tables 3 and 6).
  • Warmup steps = 250
    Selected as best among 100, 250, 350, and 400 on dev BLEU (Section 4.4, Table 6).
  • Early stopping patience = 10
    Selected as best among 5, 10, and 20 epochs on dev BLEU (Section 4.4, Table 6).
  • Beam size = 10
    Selected as best among 1, 5, and 10 on dev BLEU (Section 5, Table 6).
  • Augmentation on/off = SpecAugment on, speed perturbation off
    Best dev BLEU among none, speed perturbation only, SpecAugment only, and both (Section 4.5, Tables 5 and 6).
  • SpecAugment mask limits = Time mask max 30 frames, frequency mask max 30 mel bins
    Set without an ablation study (Section 4.5.1).
  • Speed perturbation factors = 0.9x, 1.0x, 1.1x
    Set without an ablation study (Section 4.5.2).
  • Additional Bhojpuri-only fine-tuning epochs for contrastive model = 1
    Selected as best among 1, 2, and convergence in Table 7.
assumptions (5)
  • domain assumption The official IWSLT Bhojpuri-Hindi and Marathi-Hindi datasets are correctly aligned and suitable for training and evaluation.
    All results use the official splits (Section 4.1); no data cleaning or manual verification is reported.
  • domain assumption Fine-tuning all parameters of SeamlessM4T improves over the pretrained model for this language pair.
    No zero-shot or frozen-backbone baseline is reported (Sections 3.1 and 4.3).
  • domain assumption Marathi-Hindi data provides useful cross-lingual signal for Bhojpuri-Hindi.
    Assumed from linguistic similarity between Indo-Aryan languages (Section 4.6).
  • domain assumption BLEU and chrF++ are valid proxies for translation quality.
    Evaluation uses only these automatic metrics (Section 3.3).
  • domain assumption SpecAugment and speed perturbation do not corrupt the training signal.
    Augmentations are applied without inspecting or validating the generated variants (Section 4.5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of IIITH-BUT system for IWSLT 2025 low-resource Bhojpuri to Hindi speech translation." pith.science (2026). https://pith.science/paper/5J4ZNCOK

@misc{pith2026250604714,
  author       = {Pith},
  title        = {Pith review of: IIITH-BUT system for IWSLT 2025 low-resource Bhojpuri to Hindi speech translation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5J4ZNCOK}},
  note         = {Machine review of arXiv:2506.04714}
}
read the original abstract

This paper presents the submission of IIITH-BUT to the IWSLT 2025 shared task on speech translation for the low-resource Bhojpuri-Hindi language pair. We explored the impact of hyperparameter optimisation and data augmentation techniques on the performance of the SeamlessM4T model fine-tuned for this specific task. We systematically investigated a range of hyperparameters including learning rate schedules, number of update steps, warm-up steps, label smoothing, and batch sizes; and report their effect on translation quality. To address data scarcity, we applied speed perturbation and SpecAugment and studied their effect on translation quality. We also examined the use of cross-lingual signal through joint training with Marathi and Bhojpuri speech data. Our experiments reveal that careful selection of hyperparameters and the application of simple yet effective augmentation techniques significantly improve performance in low-resource settings. We also analysed the translation hypotheses to understand various kinds of errors that impacted the translation quality in terms of BLEU.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

26 extracted references · 14 canonical work pages

  1. [1]

    Idris Abdulmumin, Victor Agostinelli, Tanel Alumäe, Antonios Anastasopoulos, Ashwin, Luisa Bentivogli, Ondřej Bojar, Claudia Borg, Fethi Bougares, Roldano Cattoni, Mauro Cettolo, Lizhong Chen, William Chen, Raj Dabre, Yannick Estève, Marcello Federico, Marco Gaido, Dávid Javorský, Marek Kasztelnik, and 30 others. 2025. Findings of the iwslt 2025 evaluatio...

    2025

  2. [2]

    Milind Agarwal, Sweta Agrawal, Antonios Anastasopoulos, Luisa Bentivogli, Ond r ej Bojar, Claudia Borg, Marine Carpuat, Roldano Cattoni, Mauro Cettolo, Mingda Chen, William Chen, Khalid Choukri, Alexandra Chronopoulou, Anna Currey, Thierry Declerck, Qianqian Dong, Kevin Duh, Yannick Est \`e ve, Marcello Federico, and 43 others. 2023. https://doi.org/10.18...

    work page doi:10.18653/v1/2023.iwslt-1.1 2023

  3. [3]

    Ibrahim Said Ahmad, Antonios Anastasopoulos, Ond r ej Bojar, Claudia Borg, Marine Carpuat, Roldano Cattoni, Mauro Cettolo, William Chen, Qianqian Dong, Marcello Federico, Barry Haddow, D \'a vid Javorsk \'y , Mateusz Krubi \'n ski, Tsz Kin Lam, Xutai Ma, Prashant Mathur, Evgeny Matusov, Chandresh Maurya, John McCrae, and 25 others. 2024. https://doi.org/1...

    doi:10.18653/v1/2024.iwslt-1.1 2024

  4. [4]

    Luisa Bentivogli, Mauro Cettolo, Marco Gaido, Alina Karakanta, Alberto Martinelli, Matteo Negri, and Marco Turchi. 2021. Cascade versus direct speech translation: Do the differences still make a difference? arXiv preprint arXiv:2106.01045

    work page Pith review arXiv 2021

  5. [5]

    Alexandre B \'e rard, Olivier Pietquin, Christophe Servan, and Laurent Besacier. 2016. Listen and translate: A proof of concept for end-to-end speech-to-text translation. arXiv preprint arXiv:1612.01744

    arXiv 2016

  6. [6]

    Seamless Communication, Loïc Barrault, Yu-An Chung, Mariano Cora Meglioli, David Dale, Ning Dong, Paul-Ambroise Duquenne, Hady Elsahar, Hongyu Gong, Kevin Heffernan, John Hoffman, Christopher Klaiber, Pengwei Li, Daniel Licht, Jean Maillard, Alice Rakotoarison, Kaushik Ram Sadagopan, Guillaume Wenzek, Ethan Ye, and 49 others. 2023. Seamlessm4t: Massively ...

    arXiv 2023

  7. [7]

    Ortega, Rodolfo Zevallos, and William Chen

    John E. Ortega, Rodolfo Zevallos, and William Chen. 2023. https://doi.org/10.18653/v1/2023.iwslt-1.23 QUESPA submission for the IWSLT 2023 dialect and low-resource speech translation tasks . In Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023), pages 261--268, Toronto, Canada (in-person and online). Association f...

    work page doi:10.18653/v1/2023.iwslt-1.23 2023

  8. [8]

    Edward Gow-Smith, Alexandre Berard, Marcely Zanon Boito, and Ioan Calapodescu. 2023. https://doi.org/10.18653/v1/2023.iwslt-1.10 NAVER LABS E urope ' s multilingual speech translation systems for the IWSLT 2023 low-resource track . In Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023), pages 144--158, Toronto, Can...

    work page doi:10.18653/v1/2023.iwslt-1.10 2023

Show all 26 references
  1. [9]

    Ye Jia, Ron J Weiss, Fadi Biadsy, Wolfgang Macherey, Melvin Johnson, Zhifeng Chen, and Yonghui Wu. 2019. Direct speech-to-speech translation with a sequence-to-sequence model. arXiv preprint arXiv:1904.06037

    2019 arXiv

  2. [10]

    Santosh Kesiraju, Karel Bene s , Maksim Tikhonov, and Jan C ernock \'y . 2023 a . https://doi.org/10.18653/v1/2023.iwslt-1.19 BUT systems for IWSLT 2023 M arathi - H indi low resource speech translation task . In Proceedings of the 20th International Conference on Spoken Langu...

    2023 doi

  3. [11]

    Santosh Kesiraju, Marek Sarvaš, Tomáš Pavlíček, Cécile Macaire, and Alejandro Ciuba. 2023 b . https://doi.org/10.21437/Interspeech.2023-2506 Strategies for improving low resource speech to text translation relying on pre-trained asr models . In Interspeech 2023, pages 2148--2152

    2023 doi

  4. [12]

    Tom Ko, Vijayaditya Peddinti, Daniel Povey, and Sanjeev Khudanpur. 2015. Audio augmentation for speech recognition. In Interspeech, volume 2015, page 3586

    2015

  5. [13]

    Jonathan Mbuya and Antonios Anastasopoulos. 2023. https://doi.org/10.18653/v1/2023.iwslt-1.24 GMU systems for the IWSLT 2023 dialect and low-resource speech translation tasks . In Proceedings of the 20th International Conference on Spoken Language Translation (IWSLT 2023), pag...

    2023 doi

  6. [14]

    John E Ortega, Rodolfo Joel Zevallos, Ibrahim Sa’id Ahmad, and William Chen. 2024. QUESPA Submission for the IWSLT 2024 Dialectal and Low-resource Speech Translation Task . In Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 2024), pages 125--133

    2024

  7. [15]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. https://doi.org/10.3115/1073083.1073135 B leu: a method for automatic evaluation of machine translation . In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311--3...

    2002

  8. [16]

    Daniel S Park, William Chan, Yu Zhang, Chung-Cheng Chiu, Barret Zoph, Ekin D Cubuk, and Quoc V Le. 2019. Specaugment: A simple data augmentation method for automatic speech recognition. arXiv preprint arXiv:1904.08779

    2019 arXiv

  9. [17]

    Maja Popovi \'c . 2017. https://doi.org/10.18653/v1/W17-4770 chr F ++: words helping character n-grams . In Proceedings of the Second Conference on Machine Translation, pages 612--618, Copenhagen, Denmark. Association for Computational Linguistics

    2017 doi

  10. [18]

    Matt Post. 2018. https://doi.org/10.18653/v1/W18-6319 A call for clarity in reporting BLEU scores . In Proceedings of the Third Conference on Machine Translation: Research Papers, pages 186--191, Brussels, Belgium. Association for Computational Linguistics

    2018 doi

  11. [19]

    Matt Post, Gaurav Kumar, Adam Lopez, Damianos Karakos, Chris Callison-Burch, and Sanjeev Khudanpur. 2013. Improved speech-to-text translation with the fisher and callhome spanish-english speech translation corpus. In Proceedings of the 10th international workshop on spoken lan...

    2013

  12. [20]

    Balaji Radhakrishnan, Saurabh Agrawal, Raj Prakash Gohil, Kiran Praveen, Advait Vinay Dhopeshwarkar, and Abhishek Pandey. 2023. https://doi.org/10.18653/v1/2023.iwslt-1.43 SRI - B ' s systems for IWSLT 2023 dialectal and low-resource track: M arathi- H indi speech translation ...

    2023 doi

  13. [21]

    Nathaniel Romney Robinson, Kaiser Sun, Cihan Xiao, Niyati Bafna, Weiting Tan, Haoran Xu, Henry Li Xinyuan, Ankur Kejriwal, Sanjeev Khudanpur, Kenton Murray, and 1 others. 2024. JHU IWSLT 2024 Dialectal and Low-resource System Description . In Proceedings of the 21st Internatio...

    2024

  14. [22]

    Akshaya Vishnu Kudlu Shanbhogue, Ran Xue, Soumya Saha, Daniel Zhang, and Ashwinkumar Ganesan. 2023. https://doi.org/10.18653/v1/2023.iwslt-1.21 Improving low resource speech translation with data augmentation and ensemble strategies . In Proceedings of the 20th International C...

    2023 doi

  15. [23]

    Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Dmytro Okhonko, and Juan Pino. 2020 a . https://doi.org/10.18653/v1/2020.aacl-demo.6 Fairseq S 2 T : Fast speech-to-text modeling with fairseq . In Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for ...

    2020 doi

  16. [24]

    Changhan Wang, Yun Tang, Xutai Ma, Anne Wu, Sravya Popuri, Dmytro Okhonko, and Juan Pino. 2020 b . Fairseq s2t: Fast speech-to-text modeling with fairseq. arXiv preprint arXiv:2010.05171

    2020 arXiv

  17. [25]

    Wikipedia contributors . 2025. 2011 census of india --- Wikipedia , the free encyclopedia. https://en.wikipedia.org/w/index.php?title=2011_census_of_India&oldid=1289695299. [Online; accessed 1-June-2025]

    2025

  18. [26]

    Maria Zafar, Antonio Castaldo, Prashanth Nayak, Rejwanul Haque, Neha Gajakos, and Andy Way. 2024. The SETU-DCU submissions to IWSLT 2024 low-resource speech-to-text translation tasks . In Proceedings of the 21st International Conference on Spoken Language Translation (IWSLT 20...

    2024

Pith tools

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