Pith. sign in

REVIEW 5 major objections 5 minor 53 references

BanglaDialecto: An End-to-End AI-Powered Regional Speech Standardization

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Fine-tuned Whisper and BanglaT5 can standardize Noakhali dialect speech into formal Bangla, with near-perfect transcription on the authors' new dataset.

desk verdict A new Noakhali dialect speech corpus and a plausible end-to-end ASR/MT/TTS pipeline, but the headline numbers likely rest on a non-speaker-disjoint split and the state-of-the-art claim contradicts the paper's own table. read the letter →

arxiv 2411.10879 v1 pith:DKV6FPV5 submitted 2024-11-16 cs.CL cs.AIcs.LGcs.SDeess.AS

Md. Nazmus Sadat Samin , Jawad Ibn Ahad , Tanjila Ahmed Medha , Fuad Rahman , Mohammad Ruhul Amin , Nabeel Mohammed , Shafin Rahman This is my paper · ORCID
classification cs.CLcs.AIcs.LGcs.SDeess.AS
keywords NoakhalidialectBanglaspeechrecognitionstandardizationWhisperfine-tuningT5machinetranslationtext-to-speechlow-resourceASR
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 claims that an end-to-end, fully automatic pipeline can take spoken Noakhali Bangla—a regional dialect with no public speech corpus before this work—and convert it into standard Bangla speech. The pipeline chains three fine-tuned components: Whisper-large V2 transcribes dialect audio into dialect text, BanglaT5 rewrites that text into standard Bangla, and AlignTTS speaks the result. On the authors' new NDD dataset the system reports character error rate 0.8%, word error rate 1.5%, and BLEU 41.6. If these numbers hold on genuinely held-out speakers, the approach would provide a reusable template for standardizing the other roughly 54 Bangla dialects.

What carries the argument

The load-bearing object is the NDD (Noakhali Dialect Dataset): 10 hours of dialect speech from 24 native speakers, segmented into 7,200 five-second chunks, each aligned with a manual dialect transcription and a standard Bangla translation. The argument runs on two fine-tuning stages over this triply-aligned data: Whisper (speech-to-text) is trained on (audio chunk, dialect text) pairs, and BanglaT5 (text-to-text) is trained on (dialect text, standard text) pairs; AlignTTS then synthesizes standard speech from the translated text. The 5-second segmentation is what makes the alignment tractable for transformer models.

What would settle it

Recompute CER, WER, and BLEU using a split that holds out entire speakers or entire source recordings, then compare with the paper's numbers; large degradation would show the reported metrics came from overlap, not generalization.

Watch

Extended reading notes

Core claim

The central claim is that fine-tuning a large multilingual ASR and a multilingual text-to-text model on a modest but carefully aligned dialect corpus is enough to nearly close the gap between a regional Bangla accent and the standard language. Specifically, the paper reports that fine-tuned Whisper-large V2 drops from a CER of 135.2% (pretrained, essentially useless on this dialect) to 0.8%, and from WER 167.5% to 1.5%, on Noakhali dialect speech; fine-tuned BanglaT5 reaches a BLEU score of 41.6 for dialect-to-standard translation, up from 22.7 without fine-tuning. The authors also demonstrate the complete chain by feeding ASR output through BanglaT5 and then AlignTTS to produce standard Bangla voice. The discovery is framed as evidence that large multilingual models, adapted on dialect-specific data, can perform dialect standardization without hand-built pronunciation or grammar rules.

Load-bearing premise

The results assume the data split is clean, meaning no audio from the same speaker or same recording appears in both training and test sets; the paper describes chunking recordings into 5-second pieces and then splitting the chunks, which risks inflating the scores.

Editorial extensions

If this is right

  • A speaker of the Noakhali dialect can, in principle, dictate in their native accent and receive formal Bangla text or speech without human transcription.
  • Fine-tuning is the decisive step: pretrained Whisper and BanglaT5 perform badly on this dialect, while fine-tuned versions reach the reported accuracy.
  • Larger Whisper variants scale monotonically on this task, from base (CER 20.6%) to large V2 (CER 0.8%).
  • The same three-stage recipe can be applied to other Bangla dialects once an equivalent aligned speech-text dataset exists.

Reading between the lines

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

  • The reported test metrics are likely optimistic because the 7,200 segment split into train/validation/test was done without a speaker-disjoint or recording-disjoint partition, so adjacent 5-second chunks of the same utterance can appear on both sides of the split.
  • The BLEU 41.6 for Noakhali is below the 47.43 previously reported by Vashantor on a Noakhali text benchmark; a fair comparison to that benchmark would require the same evaluation protocol.
  • A quick test of the pipeline's true generalization would be to record a new Noakhali speaker, run the system, and measure whether CER and WER remain in single digits; the paper does not report such an out-of-corpus evaluation.
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

5 major / 5 minor

Summary. This paper describes an end-to-end pipeline that converts Noakhali-dialect Bangla speech into standard Bangla speech. The authors collect a 10-hour corpus (NDD) from 24 speakers, segment it into 7200 five-second chunks, and fine-tune Whisper variants for dialect speech-to-text and several sequence-to-sequence models (including BanglaT5) for dialect-to-standard text translation, then use AlignTTS for text-to-speech. The headline results are 0.8% CER and 1.5% WER for Whisper-large V2 and a BLEU of 41.6 for BanglaT5. The paper claims this is the first integrated ASR+MT+TTS system for a Bangla dialect and that it outperforms existing techniques.

Significance. If the reported performance survives a properly disjoint evaluation, the paper would be a useful contribution to low-resource Bangla dialect processing: it provides a new annotated dialect corpus, compares multiple Whisper and MT model sizes under fine-tuning, and makes code and data available through a linked repository. The main value is the integrated pipeline and the new corpus, not any single methodological innovation. However, the current manuscript does not establish that the metrics are independent of the training data, and it does not evaluate the TTS output at all, so the headline claims are not yet supported.

major comments (5)
  1. [§III-B and §IV-A] The train/validation/test partition is described only as a split of the 7200 five-second segments into 6270/810/120 samples, with no statement that the split is disjoint by speaker or by original recording. Since Section III-A reports that the audio comes from 24 speakers and Section III-B chunks each recording into contiguous five-second segments, a segment-level random split will in all likelihood place chunks from the same recording, often from the same read paragraph, in both training and test. Such leakage can artificially lower the reported CER, WER, and BLEU. Please re-run the evaluation with a speaker- or recording-disjoint split, or otherwise provide evidence that segment-level independence is satisfied; please also report the test-set size and confidence intervals for the headline numbers.
  2. [§IV-B, Table II, Table III] The text in Section IV-B states that 'our models performed better than any of the current techniques,' but Table II itself reports a BLEU of 47.43 for Faria et al. [10] on the Noakhali dialect, which is higher than the 41.6 BLEU attributed to BanglaT5 in the same table. In addition, the Ours row in Table II reports CER 20.2 and WER 38.2 for the MT task, while Table III reports 21.3 and 38.2 for the fine-tuned BanglaT5 model. These internal inconsistencies must be corrected before the state-of-the-art claim can be assessed.
  3. [§III-A and §II] The claim in Section III-A that 'no speech signal dataset for the Noakhali dialect has been available to date' is contradicted by the paper's own related-work discussion: Faria et al. [10] (Vashantor) is cited as containing a Noakhali dialect benchmark and an associated BLEU score. The authors should clarify the relation between NDD and Vashantor, state what NDD adds (e.g., speech recordings vs. text-only data, different annotation protocol, larger size), and soften the dataset-novelty claim accordingly.
  4. [§I and §IV-B] The paper's title and abstract promise an end-to-end conversion of dialect speech to standard Bangla speech, but the TTS stage is not evaluated at all. Section IV-B states only that AlignTTS was used to generate standard speech; no MOS, intelligibility, similarity, or even qualitative listening-test results are reported, and Table IV (which should contain example outputs) appears empty in the manuscript. Without any evaluation of the synthesized speech, the end-to-end claim is not substantiated. Please add an evaluation of the TTS output or explicitly restrict the paper's claims to the ASR+MT stages.
  5. [§IV-A and §IV-C] The test set consists of only 120 five-second segments, and the parameter-scaling conclusion in Figure 4(a) is based on small performance differences (e.g., CER 2.0 vs. 1.5 vs. 0.8 across Whisper-small, medium, and large V2) reported without error bars or significance tests. On such a small test set these differences may reflect sampling noise; the authors should report confidence intervals or a significance test before drawing conclusions about model scale.
minor comments (5)
  1. [Table III] The column header 'BLUE Score' should be 'BLEU Score', and BLEU should not be reported with a percent sign in the abstract and text (41.6% should be 41.6).
  2. [§III-B] In the segmentation formula, the index 'j' in 's_i^k = s_i[(k-1)*5 : j*5]' should be 'k'; the surrounding text also mixes t_{i}^{d,j} and t_{i}^{d,k} notation.
  3. [Table I] The 'Max Text' row in Table I has no value; please complete the table entry or remove the row.
  4. [§IV-C] The statement that 'mBART showing the lowest WER' in the ablation study contradicts Table III, where fine-tuned mBART50 has WER 416.8, the highest among the four MT models; please correct this discussion.
  5. [§IV-A] The phrase 'which is a standard and thereby enables rigorous evaluation of model performance' is an incomplete sentence and should be rewritten.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: results come from a conventional supervised fine-tuning evaluation, with only a non-circular data-split provenance concern.

full rationale

The paper's claimed derivation chain is a standard supervised fine-tuning pipeline: NDD provides paired speech, dialect-text, and standard-text examples; Whisper variants are fine-tuned on (s_i^k, t_i^d,k) for speech-to-text, BanglaT5 and other models are fine-tuned on (t_i^d,k, t_i^s,k) for text standardization, and the reported CER, WER, and BLEU scores are computed on a held-out partition of 120 samples and compared against pretrained checkpoints. None of the model outputs is defined in terms of the training targets, no fitted parameter is renamed as a prediction, and no uniqueness theorem or load-bearing self-citation is invoked. The citations to Whisper, BanglaT5, and AlignTTS are external model references, not justifications for the paper's own conclusions. The only substantive concern is that the paper does not state whether the 6270/810/120 split is speaker- or recording-disjoint; if segments from the same original recording appear in both training and test, the absolute metrics could be inflated. That is a data-provenance and independence risk, not a definitional circularity, and it cannot be established from the manuscript text alone. Accordingly, no circular step is exhibited and the circularity score is 0.

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

The central results are empirical machine-learning outcomes, not derivations. The claims rest on dataset quality, annotation fidelity, the validity of the evaluation split, and standard transfer-learning assumptions. No new physical or theoretical entity is introduced; NDD is a dataset resource. The listed hyperparameters are hand-chosen and their values affect the reported numbers.

free parameters (5)
  • 5-second segment length = 5 s
    Chosen to fit transformer input; segmentation can split words and affect ASR and MT alignment.
  • ASR training epochs = 10
    Hand-chosen; no tuning curve or early-stopping analysis is reported.
  • MT training epochs = 25
    Hand-chosen; no tuning curve or early-stopping analysis is reported.
  • Batch sizes = 16 for ASR, 6 for MT
    Hand-chosen hyperparameters that affect the final model weights and metrics.
  • Train/validation/test split = 6270/810/120
    Split is described without a speaker-disjoint guarantee; the assignment directly determines the reported test scores.
assumptions (5)
  • domain assumption NDD human annotations are accurate and consistent
    Section III-A describes manual annotation by 24 evaluators; no inter-annotator agreement or verification protocol is reported.
  • domain assumption 5-second chunks align with the corresponding text chunks
    Section III-B: text split follows the same process, but no forced alignment or manual boundary verification is described.
  • domain assumption Train/test segments are independent
    Section IV-A: the split appears to be segment-level, with no speaker partitioning; if violated, test scores are inflated.
  • domain assumption Fine-tuning generalizes beyond the 24 recorded speakers
    The paper's own Limitations paragraph says the dataset from specific regions may not fully represent dialectal diversity.
  • standard math CER/WER/BLEU are adequate measures of pipeline quality
    Used in Section IV-A, but no human or acoustic evaluation of synthesized standard speech is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of BanglaDialecto: An End-to-End AI-Powered Regional Speech Standardization." pith.science (2026). https://pith.science/paper/DKV6FPV5

@misc{pith2026241110879,
  author       = {Pith},
  title        = {Pith review of: BanglaDialecto: An End-to-End AI-Powered Regional Speech Standardization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DKV6FPV5}},
  note         = {Machine review of arXiv:2411.10879}
}
read the original abstract

This study focuses on recognizing Bangladeshi dialects and converting diverse Bengali accents into standardized formal Bengali speech. Dialects, often referred to as regional languages, are distinctive variations of a language spoken in a particular location and are identified by their phonetics, pronunciations, and lexicon. Subtle changes in pronunciation and intonation are also influenced by geographic location, educational attainment, and socioeconomic status. Dialect standardization is needed to ensure effective communication, educational consistency, access to technology, economic opportunities, and the preservation of linguistic resources while respecting cultural diversity. Being the fifth most spoken language with around 55 distinct dialects spoken by 160 million people, addressing Bangla dialects is crucial for developing inclusive communication tools. However, limited research exists due to a lack of comprehensive datasets and the challenges of handling diverse dialects. With the advancement in multilingual Large Language Models (mLLMs), emerging possibilities have been created to address the challenges of dialectal Automated Speech Recognition (ASR) and Machine Translation (MT). This study presents an end-to-end pipeline for converting dialectal Noakhali speech to standard Bangla speech. This investigation includes constructing a large-scale diverse dataset with dialectal speech signals that tailored the fine-tuning process in ASR and LLM for transcribing the dialect speech to dialect text and translating the dialect text to standard Bangla text. Our experiments demonstrated that fine-tuning the Whisper ASR model achieved a CER of 0.8% and WER of 1.5%, while the BanglaT5 model attained a BLEU score of 41.6% for dialect-to-standard text translation.

Figures

Figures reproduced from arXiv: 2411.10879 by the authors.

Figure 1
Figure 1. (a) Typical Deep Neural Network (DNN) based imple [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. BanglaDialecto system: (a) input dialect speech signals s i are converted into wav form, then it undergoes the process of noise reduction and splitting into manageable 5-second speech segments s i k , Dialect text t i d and standard text t i s then segmented in corresponding chunks t i d,k and t i s,k. (b) The segment s i k and t i d are used to fine-tune the ASR to predict and transcript dialect speech s i k into d… view at source ↗
Figure 3
Figure 3. Interview participant distribution across the Noakhali [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: (a) The impact of increasing Whisper model parameters [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 38 canonical work pages

  1. [10]

    Vashantor: a large-scale multilingual benchmark dataset for automated translation of bangla regional dialects to bangla language,

    F. T. J. Faria, M. B. Moin, A. A. Wase, M. Ahmmed, M. R. Sani, and T. Muhammad, “Vashantor: a large-scale multilingual benchmark dataset for automated translation of bangla regional dialects to bangla language,” arXiv preprint arXiv:2311.11142 , 2023

    arXiv 2023

  2. [1]

    Continuous bengali speech recognition based on deep neural network,

    M. A. Al Amin, M. T. Islam, S. Kibria, and M. S. Rahman, “Continuous bengali speech recognition based on deep neural network,” in 2019 international conference on electrical, computer and communication engineering (ECCE). IEEE, 2019, pp. 1–6

    work page 2019

  3. [2]

    Assessing performance of bengali speech recognizers under real world conditions using gmm-hmm and dnn based methods

    S. Khan, M. Pal, J. Basu, M. S. Bepari, and R. Roy, “Assessing performance of bengali speech recognizers under real world conditions using gmm-hmm and dnn based methods.” in SLTU, 2018, pp. 192–196

    work page 2018

  4. [3]

    Deep learning based large vocabulary continuous speech recognition of an under-resourced language bangladeshi bangla,

    A. M. Samin, M. H. Kobir, S. Kibria, and M. S. Rahman, “Deep learning based large vocabulary continuous speech recognition of an under-resourced language bangladeshi bangla,” Acoustical Science and Technology, vol. 42, no. 5, pp. 252–260, 2021

    work page 2021

  5. [4]

    Whisper augmented end-to- end/hybrid speech recognition system-cyclegan approach

    P. R. Gudepu, G. P. Vadisetti, A. Niranjan, K. Saranu, R. Sarma, M. A. B. Shaik, and P. Paramasivam, “Whisper augmented end-to- end/hybrid speech recognition system-cyclegan approach.” in INTER- SPEECH, 2020, pp. 2302–2306

    work page 2020

  6. [5]

    Analysis of whisper automatic speech recognition performance on low resource language,

    R. S. A. Pratama and A. Amrullah, “Analysis of whisper automatic speech recognition performance on low resource language,” Jurnal Pilar Nusa Mandiri, vol. 20, no. 1, pp. 1–8, 2024

    work page 2024

  7. [6]

    Enhancing whisper model for vietnamese specific domain with data blending and lora fine-tuning,

    N. H. Phung, D. T. Dang, K. D. Ta, K. T. A. Nguyen, T. K. Tran, and C. T. Nguyen, “Enhancing whisper model for vietnamese specific domain with data blending and lora fine-tuning,” in The International Conference on Intelligent Systems & Networks . Springer, 2024, pp. 145–155

    work page 2024

  8. [7]

    Phonological variation and linguistic diversity in bangladeshi dialects: An exploration of sound patterns and sociolinguis- tic significance,

    M. M. Rahman, B. Barman, L. Sharmin, M. R. Uddin, S. B. Yusuf, and U. Rasool, “Phonological variation and linguistic diversity in bangladeshi dialects: An exploration of sound patterns and sociolinguis- tic significance,” in Forum for Linguistic Studies , vol. 6, no. 2, 2024

    work page 2024

Show all 53 references
  1. [8]

    Social factors and dialect variation: An analysis of age, gender, and social class in linguistic practice

    S. H. Hamed and S. Mohamed, “Social factors and dialect variation: An analysis of age, gender, and social class in linguistic practice.” Refereed Journal of Northern Europe Academy for Studies & Research (Denmark), no. 21, 2023

    2023

  2. [9]

    English to bangla machine translation using recurrent neural network,

    S. Siddique, T. Ahmed, M. R. A. Talukder, and M. M. Uddin, “English to bangla machine translation using recurrent neural network,” arXiv preprint arXiv:2106.07225, 2021

    2021 arXiv

  3. [11]

    Yu and L

    D. Yu and L. Deng, Automatic speech recognition . Springer, 2016, vol. 1

    2016

  4. [12]

    Deep learning based bangla speech-to-text conver- sion,

    M. T. Tausif, S. Chowdhury, M. S. Hawlader, M. Hasanuzzaman, and H. Heickal, “Deep learning based bangla speech-to-text conver- sion,” in 2018 5th International Conference on Computational Sci- ence/Intelligence and Applied Informatics (CSII) . IEEE, 2018, pp. 49–54

    2018

  5. [13]

    Deep neural networks in machine translation: An overview

    J. Zhang, C. Zong et al., “Deep neural networks in machine translation: An overview.” IEEE Intell. Syst. , vol. 30, no. 5, pp. 16–25, 2015

    2015

  6. [14]

    Bert: Pre-training of deep bidirectional transformers for language understanding,

    J. Devlin, “Bert: Pre-training of deep bidirectional transformers for language understanding,” arXiv preprint arXiv:1810.04805 , 2018

    2018 arXiv

  7. [15]

    Robust speech recognition via large-scale weak supervi- sion,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak supervi- sion,” in International conference on machine learning . PMLR, 2023, pp. 28 492–28 518

    2023

  8. [16]

    Exploration of whisper fine-tuning strategies for low-resource asr,

    Y . Liu, X. Yang, and D. Qu, “Exploration of whisper fine-tuning strategies for low-resource asr,” EURASIP Journal on Audio, Speech, and Music Processing , vol. 2024, no. 1, p. 29, 2024

    2024

  9. [17]

    Multilingual machine translation with large language models: Empirical results and analysis,

    W. Zhu, H. Liu, Q. Dong, J. Xu, S. Huang, L. Kong, J. Chen, and L. Li, “Multilingual machine translation with large language models: Empirical results and analysis,” arXiv preprint arXiv:2304.04675, 2023

    2023 arXiv

  10. [18]

    Bangla natural language processing: A comprehensive analysis of classical, machine learning, and deep learning-based methods,

    O. Sen, M. Fuad, M. N. Islam, J. Rabbi, M. Masud, M. K. Hasan, M. A. Awal, A. A. Fime, M. T. H. Fuad, D. Sikder et al. , “Bangla natural language processing: A comprehensive analysis of classical, machine learning, and deep learning-based methods,” IEEE Access, vol. 10, pp. 38...

    2022

  11. [19]

    Chinese dialect speech recog- nition: a comprehensive survey,

    Q. Li, Q. Mai, M. Wang, and M. Ma, “Chinese dialect speech recog- nition: a comprehensive survey,” Artificial Intelligence Review , vol. 57, no. 2, p. 25, 2024

    2024

  12. [20]

    Deep investigation of the recent advances in dialectal arabic speech recognition,

    H. A. Alsayadi, A. A. Abdelhamid, I. Hegazy, B. Alotaibi, and Z. T. Fayed, “Deep investigation of the recent advances in dialectal arabic speech recognition,” IEEE Access, vol. 10, pp. 57 063–57 079, 2022

    2022

  13. [21]

    End-to-end automatic speech recogni- tion model for the sudanese dialect,

    A. Mansour and W. F. Mukhtar, “End-to-end automatic speech recogni- tion model for the sudanese dialect,” arXiv preprint arXiv:2212.10826 , 2022

    2022 arXiv

  14. [22]

    Deep speech: Scaling up end-to-end speech recognition,

    A. Hannun, “Deep speech: Scaling up end-to-end speech recognition,” arXiv preprint arXiv:1412.5567 , 2014

    2014 arXiv

  15. [23]

    Tunisian dialectal end-to-end speech recognition based on deepspeech,

    A. Messaoudi, H. Haddad, C. Fourati, M. B. Hmida, A. B. E. Mabrouk, and M. Graiet, “Tunisian dialectal end-to-end speech recognition based on deepspeech,” Procedia Computer Science , vol. 189, pp. 183–190, 2021

    2021

  16. [24]

    Effective training end-to-end asr systems for low-resource lhasa dialect of tibetan language,

    L. Pan, S. Li, L. Wang, and J. Dang, “Effective training end-to-end asr systems for low-resource lhasa dialect of tibetan language,” in 2019 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) . IEEE, 2019, pp. 1152–1156

    2019

  17. [25]

    A real- time bangla local language recognition from voice,

    M. J. K. Noor, F. T. Johora, M. Mahin, and M. A. Rahaman, “A real- time bangla local language recognition from voice,” in Proceedings of the Fourth International Conference on Trends in Computational and Cognitive Engineering: TCCE 2022 . Springer, 2023, pp. 489–501

    2022

  18. [26]

    Classification of bangla regional languages and recogni- tion of artificial bangla speech using deep learning,

    P. S. Hossain, “Classification of bangla regional languages and recogni- tion of artificial bangla speech using deep learning,” Ph.D. dissertation, Brac University, 2022

    2022

  19. [27]

    A neural attention-based encoder-decoder approach for english to bangla transla- tion,

    A. A. Shiam, S. M. Redwan, M. H. Kabir, and J. Shin, “A neural attention-based encoder-decoder approach for english to bangla transla- tion,” Computer Science Journal of Moldova , vol. 91, no. 1, pp. 70–85, 2023

    2023

  20. [28]

    Banglanlg and banglat5: Benchmarks and resources for evaluating low-resource natural language generation in bangla,

    A. Bhattacharjee, T. Hasan, W. U. Ahmad, and R. Shahriyar, “Banglanlg and banglat5: Benchmarks and resources for evaluating low-resource natural language generation in bangla,” 2023

    2023

  21. [29]

    mt5: A massively multilingual pre-trained text-to-text transformer,

    L. Xue, N. Constant, A. Roberts, M. Kale, R. Al-Rfou, A. Siddhant, A. Barua, and C. Raffel, “mt5: A massively multilingual pre-trained text-to-text transformer,” arXiv preprint arXiv:2010.11934 , 2020

    2010 arXiv

  22. [30]

    A comprehensive dialect conversion approach from chittagonian to standard bangla,

    H. R. Milon, S. N. U. Sabbir, A. Inan, and N. Hossain, “A comprehensive dialect conversion approach from chittagonian to standard bangla,” in 2020 IEEE Region 10 Symposium (TENSYMP) . IEEE, 2020, pp. 214– 217

    2020

  23. [31]

    F. Alam, M. Habib, and M. Khan, Bangla Text to Speech using Festival, May 2011

    2011

  24. [32]

    A bangla text-to-speech system using deep neural networks,

    R. S. Raju, P. Bhattacharjee, A. Ahmad, and M. S. Rahman, “A bangla text-to-speech system using deep neural networks,” in2019 International Conference on Bangla Speech and Language Processing (ICBSLP) , 2019, pp. 1–5

    2019

  25. [33]

    End to end bangla speech synthesis,

    P. Bhattacharjee, R. S. Raju, A. Ahmad, and M. S. Rahman, “End to end bangla speech synthesis,” 2021. [Online]. Available: https://arxiv.org/abs/2108.00500

    2021 arXiv

  26. [34]

    Text-to-speech synthesis system for mymensinghiya dialect of bangla language,

    A. Begum, S. M. S. Askari, and U. Sharma, “Text-to-speech synthesis system for mymensinghiya dialect of bangla language,” in Progress in Advanced Computing and Intelligent Engineering: Proceedings of ICACIE 2017, Volume 2 . Springer, 2019, pp. 291–303

    2017

  27. [35]

    A comprehensive overview of large language models,

    H. Naveed, A. U. Khan, S. Qiu, M. Saqib, S. Anwar, M. Usman, N. Akhtar, N. Barnes, and A. Mian, “A comprehensive overview of large language models,” arXiv preprint arXiv:2307.06435 , 2023

    2023 arXiv

  28. [36]

    Sichuan dialect speech recognition with deep lstm network,

    W. Ying, L. Zhang, and H. Deng, “Sichuan dialect speech recognition with deep lstm network,” Frontiers of Computer Science, vol. 14, no. 2, pp. 378–387, 2020

    2020

  29. [37]

    Bengali common voice speech dataset for automatic speech recognition,

    S. Alam, A. Sushmit, Z. Abdullah, S. Nakkhatra, M. Ansary, S. M. Hossen, S. M. Mehnaz, T. Reasat, and A. I. Humayun, “Bengali common voice speech dataset for automatic speech recognition,” arXiv preprint arXiv:2206.14053, 2022

    2022 arXiv

  30. [38]

    Hybrid pipeline for building arabic tunisian dialect-standard arabic neural machine translation model from scratch,

    S. Kchaou, R. Boujelbane, and L. Hadrich, “Hybrid pipeline for building arabic tunisian dialect-standard arabic neural machine translation model from scratch,” ACM Transactions on Asian and Low-Resource Language Information Processing, vol. 22, no. 3, pp. 1–21, 2023

    2023

  31. [39]

    Automatic speech recognition for biomedical data in bengali language,

    S. Kabir, N. Nahar, S. Saha, and M. Rashid, “Automatic speech recognition for biomedical data in bengali language,” arXiv preprint arXiv:2406.12931, 2024

    2024 arXiv

  32. [40]

    Deep learning approach for translating arabic holy quran into italian language,

    H. Hamed, A. M. Helmy, and A. Mohammed, “Deep learning approach for translating arabic holy quran into italian language,” in 2021 In- ternational Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC). IEEE, 2021, pp. 193–199

    2021

  33. [41]

    End-to-end speech recognition for arabic dialects,

    S. Nasr, R. Duwairi, and M. Quwaider, “End-to-end speech recognition for arabic dialects,” Arabian Journal for Science and Engineering , vol. 48, no. 8, pp. 10 617–10 633, 2023

    2023

  34. [42]

    Low resource neural ma- chine translation: Assamese to/from other indo-aryan (indic) languages,

    R. Baruah, R. K. Mundotiya, and A. K. Singh, “Low resource neural ma- chine translation: Assamese to/from other indo-aryan (indic) languages,” Transactions on Asian and Low-Resource Language Information Pro- cessing, vol. 21, no. 1, pp. 1–32, 2021

    2021

  35. [43]

    Development of indonesian- japanese statistical machine translation using lemma translation and additional post-process,

    M. A. Sulaeman and A. Purwarianti, “Development of indonesian- japanese statistical machine translation using lemma translation and additional post-process,” in 2015 International conference on electrical engineering and informatics (ICEEI) . IEEE, 2015, pp. 54–58

    2015

  36. [44]

    End-to-end jordanian dialect speech-to-text self-supervised learning framework,

    A. A. Safieh, I. A. Alhaol, and R. Ghnemat, “End-to-end jordanian dialect speech-to-text self-supervised learning framework,” Frontiers in Robotics and AI , vol. 9, p. 1090012, 2022

    2022

  37. [45]

    Sylheti to standard bangla neural ma- chine translation: A deep learning-based dialect conversion approach

    T. T. Prama and M. M. Anwar, “Sylheti to standard bangla neural ma- chine translation: A deep learning-based dialect conversion approach.”

  38. [46]

    Robust speech recognition via large-scale weak super- vision,

    A. Radford, J. W. Kim, T. Xu, G. Brockman, C. McLeavey, and I. Sutskever, “Robust speech recognition via large-scale weak super- vision,” 2022

    2022

  39. [47]

    Gaussian error linear units (gelus),

    D. Hendrycks and K. Gimpel, “Gaussian error linear units (gelus),” arXiv preprint arXiv:1606.08415, 2016

    2016 arXiv

  40. [48]

    Glu variants improve transformer,

    N. Shazeer, “Glu variants improve transformer,” arXiv preprint arXiv:2002.05202, 2020

    2002 arXiv

  41. [49]

    Aligntts: Effi- cient feed-forward text-to-speech system without explicit alignment,

    Z. Zeng, J. Wang, N. Cheng, T. Xia, and J. Xiao, “Aligntts: Effi- cient feed-forward text-to-speech system without explicit alignment,” in ICASSP 2020-2020 IEEE international conference on acoustics, speech and signal processing (ICASSP) . IEEE, 2020, pp. 6714–6718

    2020

  42. [50]

    Analyzing the input stream for character-level errors in unconstrained text entry evaluations,

    J. O. Wobbrock and B. A. Myers, “Analyzing the input stream for character-level errors in unconstrained text entry evaluations,” ACM Transactions on Computer-Human Interaction (TOCHI) , vol. 13, no. 4, pp. 458–489, 2006

    2006

  43. [51]

    Minimizing word error rate in textual summaries of spoken language,

    K. Zechner and A. Waibel, “Minimizing word error rate in textual summaries of spoken language,” in 1st Meeting of the North American Chapter of the Association for Computational Linguistics , 2000

    2000

  44. [52]

    Bleu: a method for automatic evaluation of machine translation,

    K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu, “Bleu: a method for automatic evaluation of machine translation,” in Proceedings of the 40th annual meeting of the Association for Computational Linguistics , 2002, pp. 311–318

    2002

  45. [53]

    Sparks of large audio models: A survey and outlook,

    S. Latif, M. Shoukat, F. Shamshad, M. Usama, Y . Ren, H. Cuay ´ahuitl, W. Wang, X. Zhang, R. Togneri, E. Cambria et al. , “Sparks of large audio models: A survey and outlook,” arXiv preprint arXiv:2308.12792, 2023

    2023 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.