Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Whisper Turns Stronger: Augmenting Wav2Vec 2.0 for Superior ASR in Low-Resource Languages

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Augmenting fine-tuning audio with three cheap transforms—pitch shift, Gaussian noise, and band-stop filtering—cuts Wav2Vec2's word error rate by a third and character error rate by half in Arabic, Russian, and Portuguese, and beats a…

desk verdict The 33.9%/53.2% claims compare clean vs. augmented Wav2Vec2, not Whisper, and the CER numbers are internally inconsistent; the augmentation recipe is plausible, but this version can't be trusted. read the letter →

arxiv 2501.00425 v1 pith:5HJ5JC5T submitted 2024-12-31 cs.CL cs.SDeess.AS

Or Haim Anidjar , Revital Marbel , Roi Yozevitch This is my paper · ORCID
classification cs.CLcs.SDeess.AS
keywords automaticspeechrecognitionWav2Vec2dataaugmentationlow-resourcelanguagespitchshiftGaussiannoiseband-stopfilterMozillaCommonVoice
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 a simple data-augmentation recipe substantially improves automatic speech recognition (ASR) in low-resource languages. Fine-tuning Wav2Vec2-xlsr-53 on clean Mozilla Common Voice audio plus 20 percent each of pitch-shifted, Gaussian-noised, and band-stop-filtered copies of the same clips lowers the average word error rate by 33.9 percent and the average character error rate by 53.2 percent relative to fine-tuning on clean audio alone. The augmented model also posts lower WER than a Whisper-tiny baseline on the same test sets for Arabic, Russian, and Portuguese. The paper presents this as a cheap, model-agnostic path to better speech recognition where labeled data is scarce.

What carries the argument

The load-bearing mechanism is the three-way augmentation schedule applied to Wav2Vec2-xlsr-53, a self-supervised speech encoder pre-trained on 53 languages and adapted with a CTC output head. The authors construct a 160-percent training set: the original audio plus three independent 20-percent subsets, each transformed by one of three operations—pitch shift, Gaussian-noise addition, and band-stop filtering. The band-stop filter spectrally masks a frequency range to simulate accent variation, Gaussian noise adds robustness, and pitch shift expands the speaker and prosody landscape. The claim is that this particular combination, at 20-percent proportion per transform, produces lower CER than any single transform or any pair, with the full three-transform model giving the best Arabic CER of 19.0 percent in the initial Arabic-only sweep.

What would settle it

Re-run Whisper-tiny on the same Common Voice test sets after applying the same punctuation removal and Whisper's custom text normalizer, and compute its CER; if Whisper's WER then drops below the augmented Wav2Vec2 values (27.6, 35.8, 31.8) or its CER is lower than 9.0, 10.2, 11.5, the central outperformance claim fails.

Watch

Extended reading notes

Core claim

The central discovery is that a modest amount of augmented audio in the fine-tuning set sharpens a pre-trained Wav2Vec2 model's transcriptions far more than the sheer size of the added data would predict. For each language, the authors fine-tune Wav2Vec2-xlsr-53 on the clean train split plus an additional 20 percent of that split, copied under three transformations: pitch shifting by a uniformly random number of semitones in $[-6,6]$, adding amplitude-scaled Gaussian noise, and applying a band-stop filter that removes a frequency range within the 0--4000 Hz band. Using all three transforms together yields the best results of the seven combinations tested. On held-out Common Voice test sets, the augmented model reduces Arabic WER from 46.5 to 27.6 percent, Russian from 54.6 to 35.8, and Portuguese from 43.3 to 31.8, with corresponding CER falling to 9.0, 10.2, and 11.5 percent. The paper further claims that these numbers beat the Whisper-tiny baseline, whose Arabic WER alone is reported at 90.9 percent.

Load-bearing premise

The claimed outperformance over Whisper rests on the premise that the Whisper-tiny baseline is a fair state-of-the-art comparison, measured under the same preprocessing and scoring conditions as the augmented Wav2Vec2 model, which the paper does not fully establish.

Editorial extensions

If this is right

  • If the recipe transfers, other low-resource languages can adopt the same three augmentations without changing model architecture.
  • The reported gains imply that synthetic spectral and prosodic variation can offset limited labeled audio for pre-trained ASR models.
  • The large CER reductions suggest character-level errors—diacritics in Arabic especially—are disproportionately corrected by augmentation.
  • The method sets a simple, reproducible baseline for future low-resource ASR systems to compare against.

Reading between the lines

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

  • One implication the paper leaves implicit is that the 20-percent augmentation proportion may not be optimal; holding all else fixed and varying the per-transform percentage would map the benefit curve.
  • The recipe's generality could be tested by applying the same three transforms to a different pre-trained ASR architecture, such as HuBERT or a fine-tuned Whisper-small, to see whether the gains are architecture-specific.
  • Computing Whisper-tiny's CER with the same text normalization would reveal whether the character-level advantage is specific to Wav2Vec2 or a general effect of augmentation.
  • A practical takeaway: teams with only dozens of hours of labeled speech can use these three off-the-shelf audio transforms to cut error rates without collecting or labeling new data.
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

3 major / 5 minor

Summary. The paper proposes an end-to-end ASR framework for low-resource languages (Arabic, Russian, Portuguese) by fine-tuning Wav2Vec2-xlsr-53 on a combination of clean audio plus three data augmentations: pitch shift, Gaussian noise, and band-stop filtering. The authors report average relative improvements of 33.9% in WER and 53.2% in CER, and claim that this approach outperforms both a standard Wav2Vec2 baseline and OpenAI's Whisper model. The evaluation uses Mozilla Common Voice training splits of 17 hours (Arabic, Portuguese) and 30 hours (Russian). A related finding is that the augmentation combination was selected based on Arabic-language experiments and then transferred to the other languages.

Significance. If the claims were fully supported, the paper would provide a simple, low-cost recipe for improving Wav2Vec2 in low-resource settings, which is a practically useful result. The work does include a direct empirical comparison of several augmentation combinations and reports both WER and CER, and the chosen augmentations are straightforward and easy to reproduce. However, the headline numbers are not actually computed against the stated baselines, and there are internal inconsistencies in the CER results, so the significance cannot be assessed from the manuscript as written.

major comments (3)
  1. [Abstract, §1.2, Table 2, Table 3, Table 4] The abstract and §1.2 claim that the approach 'outperforms two previous baseline models, which are the pre-trained Wav2Vec2 and the well-known Whisper ASR model, resulting in an average relative improvement of 33.9% in WER and a 53.2% relative improvement in CER.' However, the 33.9% and 53.2% figures are the average clean-versus-augmented Wav2Vec2 improvements in Tables 2 and 3 (Arabic 40.65, Russian 34.43, Portuguese 26.56; Arabic 59.64, Russian 54.26, Portuguese 45.75). They are not improvements over Whisper. Recomputing from Table 4, the relative WER improvements over Whisper are roughly 69.6% (Arabic), 11.8% (Russian), and 9.7% (Portuguese), with no CER values for Whisper. The central claim of outperforming Whisper is therefore not supported by the numbers actually presented.
  2. [§2.1, §6.3, Table 4] The Whisper baseline is not an apples-to-apples comparison. Section 2.1 explicitly notes that Whisper's WER is computed using a custom text normalizer before scoring, but the paper never states whether this normalizer was applied in the Table 4 evaluation. If it was not, the reported Whisper WERs are inflated. Additionally, Table 4 uses Whisper-tiny, the smallest Whisper variant, with no fine-tuning, while the proposed model is fine-tuned on domain data. To substantiate the claim of outperforming Whisper, the authors need to evaluate a properly normalized Whisper model (ideally with several sizes, with and without fine-tuning) and report its CER as well.
  3. [§6.1, Table 3, §6.4] The CER values are internally inconsistent. Arabic clean CER is reported as 37.5% in §6.1 and Figure 3, but as 22.3% in Table 3; the augmented CER is 19.0% in §6.1 and Figure 3, but 9.0% in Table 3. Section 6.4 states the reduction as 'from 37.3% to 19.0%'. These factor-of-two discrepancies for the same metric and language mean the 53.2% improvement claim cannot be verified from the manuscript. The authors must reconcile the numbers and explain which settings (e.g., with or without diacritics) produce each value.
minor comments (5)
  1. [§4.2] The phrase 'random samples from a uniform distribution over 0.001 to 0.03 Hz' is odd; the unit Hz is likely a typo, and the range for the noise amplitude ratio should be stated in a dimensionless way.
  2. [§5.3.1] The example of Portuguese sentences yields an 85.7% WER and 17.3% CER; it would be helpful to explain how these values are computed, as the word-level error seems surprisingly high for such a short sentence.
  3. [Table 4] The table title states 'WER and CER results' but the CER column is entirely N/A; the title should be changed to reflect that only WER is reported, and the reason (no CER in Radford et al.) should be stated in the caption in a more precise way.
  4. [General] Several references are incomplete or inconsistently formatted (e.g., 'Avci and Akpolat [2006]' without full details in the text, and some URLs are missing). The paper would benefit from a thorough reference cleanup.
  5. [§1.2, §6.2] The selection of the augmentation combination is based on the Arabic development results (§6.1), and then the same recipe is applied to Russian and Portuguese without any validation on those languages. This should be acknowledged as a limitation or addressed by reporting a validation split for each language.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported gains are direct measured comparisons; the abstract's misleading attribution to Whisper is a comparison-validity issue, not a circular derivation.

full rationale

This paper is purely empirical and contains no derivation chain in which a prediction reduces to its own inputs. The central results (33.9% WER and 53.2% CER relative improvements) are arithmetic summaries of the measured Clean vs. Augmented columns in Tables 2 and 3; they are not fitted parameters renamed as predictions. The augmentation recipe is selected after observing Arabic CER results and then applied to Russian and Portuguese, which raises a model-selection/generalization concern but is not definitional circularity. The only self-citations (Anidjar et al. 2020, 2023) appear in background discussion about speaker change detection and do not carry the paper's argument. No uniqueness theorem, ansatz, or fitted input is imported from the authors' prior work. The abstract's claim that the 33.9%/53.2% improvements are relative to both Wav2Vec2 and Whisper is factually misleading because Whisper is compared separately in Table 4 and only its WER is reported against Wav2Vec2-tiny, and the CER/WER inconsistencies across sections are reporting-quality problems, not circular reasoning. Since no step in the paper is equivalent by construction to its own input, the circularity score is 0.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on manually chosen augmentation hyperparameters and several unstated domain assumptions about dataset quality, augmentation validity, and baseline fairness. No new entities are introduced. The free parameters matter because the headline improvements are measured under one specific augmentation recipe selected using the Arabic evaluation set.

free parameters (5)
  • Augmentation ratio per method = 20% of training split per augmentation (60% total for all three)
    Chosen by hand and tuned on Arabic in Section 6.1; the reported results depend on this ratio.
  • Pitch shift semitone range = [-6, 6]
    Random semitone range chosen without reported ablation; affects augmented audio diversity (Section 4.3).
  • Gaussian noise amplitude range = uniform 0.001 to 0.03 (paper says Hz)
    Hand-set noise level; the description is dimensionally unclear and the exact implementation is not reproducible (Section 4.2).
  • Band-stop filter parameters = bandwidth fraction and steepness unspecified; frequency range 0-4000 Hz
    Parameters are not fully reported, so the augmentation is not reproducible (Section 4.1).
  • Fine-tuning hyperparameters = learning rate 3e-4, batch size 16, 500 warmup steps
    Standard choices from Wav2Vec2; no sweep or sensitivity analysis is reported (Section 6).
assumptions (4)
  • domain assumption The Common Voice train and test splits used for evaluation are valid and the text labels are accurate enough for WER/CER measurement.
    The paper relies on Common Voice quality but itself notes in Section 3 that the dataset contains samples with silent audio and mismatched labels.
  • domain assumption The augmentation operations preserve the transcription of the original audio.
    Stated in Section 4: augmented data remains viable as long as augmentations do not interfere with transcription; no verification is provided that band-stop or pitch-shift preserve intelligibility.
  • ad hoc to paper Whisper-tiny is a representative state-of-the-art Whisper baseline and its WERs are comparable without applying its custom text normalizer.
    Section 2.1 notes Whisper uses a custom text normalizer before WER computation, yet Table 4 compares raw WERs; the baseline choice favors the proposed model.
  • domain assumption Wav2Vec2-xlsr-53 is an appropriate pre-trained backbone for Arabic, Russian, and Portuguese.
    The paper uses xlsr-53 pre-trained on 53 languages (Section 5.1); this is standard practice, but no language-specific justification is offered.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Whisper Turns Stronger: Augmenting Wav2Vec 2.0 for Superior ASR in Low-Resource Languages." pith.science (2026). https://pith.science/paper/5HJ5JC5T

@misc{pith2026250100425,
  author       = {Pith},
  title        = {Pith review of: Whisper Turns Stronger: Augmenting Wav2Vec 2.0 for Superior ASR in Low-Resource Languages},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5HJ5JC5T}},
  note         = {Machine review of arXiv:2501.00425}
}
read the original abstract

Approaching Speech-to-Text and Automatic Speech Recognition problems in low-resource languages is notoriously challenging due to the scarcity of validated datasets and the diversity of dialects. Arabic, Russian, and Portuguese exemplify these difficulties, being low-resource languages due to the many dialects of these languages across different continents worldwide. Moreover, the variety of accents and pronunciations of such languages complicate ASR models' success. With the increasing popularity of Deep Learning and Transformers, acoustic models like the renowned Wav2Vec2 have achieved superior performance in the Speech Recognition field compared to state-of-the-art approaches. However, despite Wav2Vec2's improved efficiency over traditional methods, its performance significantly declines for under-represented languages, even though it requires significantly less labeled data. This paper introduces an end-to-end framework that enhances ASR systems fine-tuned on Wav2Vec2 through data augmentation techniques. To validate our framework's effectiveness, we conducted a detailed experimental evaluation using three datasets from Mozilla's Common Voice project in Arabic, Russian, and Portuguese. Additionally, the framework presented in this paper demonstrates robustness to different diacritics. Ultimately, our approach outperforms two previous baseline models, which are the pre-trained Wav2Vec2 and the well-known Whisper ASR model, resulting in an average relative improvement of 33.9\% in Word Error Rate and a 53.2\% relative improvement in Character Error Rate.

Figures

Figures reproduced from arXiv: 2501.00425 by the authors.

Figure 1
Figure 1. The framework presented in this paper. unseen data. The results showed that fine-tuning the Wav2Vec2-xlsr-53 model on augmented data improved its performance compared to fine-tuning on clean data. Finally, the model architecture and flow are presented in [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. An example of two identical sentences using different diacritics. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. CER bar-chart for different combinations of augmentation methods. It is clear that the more augmentations [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Enhancing Neural Spoken Language Recognition: An Exploration with Multilingual Datasets

    cs.SD 2025-01 reject novelty 3.0 of 10

    A funnel-shaped TDNN plus 1x1 layers is reported to reach 97% language identification accuracy on ten Common Voice languages, but the evaluation is insufficient to support the claim.

Reference graph

Works this paper leans on

28 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [2]

    Dialect-specific models for automatic speech recognition of african american vernacular english

    Rachel Dorn. Dialect-specific models for automatic speech recognition of african american vernacular english. In Proceedings of the Student Research Workshop Associated with RANLP 2019, pages 16–20,

    work page 2019

  2. [5]

    Data augmentation based unsupervised pre-training for low-resource speech recognition

    Hong Luo, Xiao Xie, Penghua Li, and Feiyang Xin. Data augmentation based unsupervised pre-training for low-resource speech recognition. In 2024 36th Chinese Control and Decision Conference (CCDC) , pages 5007–5012,

    work page 2024

  3. [9]

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

    Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. wav2vec 2.0: A framework for self-supervised learning of speech representations. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Bal- can, and H. Lin, editors, Advances in Neural Information Processing Systems , volume 33, pages 12449– 12460. Curran Associates, Inc., 2020b. URL https://proc...

    work page 2020

  4. [11]

    Applying wav2vec2 for Speech Recognition on Bengali Common Voices Dataset

    HAZ Shahgir, Khondker Salman Sayeed, and Tanjeem Azwad Zaman. Applying wav2vec2 for speech recognition on bengali common voices dataset. arXiv preprint arXiv:2209.06581,

    work page Pith review arXiv

  5. [13]

    Hubert: How much can a bad teacher benefit asr pre-training? In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6533–6537

    Wei-Ning Hsu, Yao-Hung Hubert Tsai, Benjamin Bolte, Ruslan Salakhutdinov, and Abdelrahman Mohamed. Hubert: How much can a bad teacher benefit asr pre-training? In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6533–6537. IEEE,

    work page 2021

  6. [16]

    Unsupervised methods for speaker diarization: An integrated and iterative approach

    Stephen H Shum, Najim Dehak, Réda Dehak, and James R Glass. Unsupervised methods for speaker diarization: An integrated and iterative approach. IEEE Transactions on Audio, Speech, and Language Processing , 21(10): 2015–2028,

    work page 2015

  7. [18]

    Speaker-dependent live quranic verses recitation recognition system using sphinx-4 framework

    Aurish Hammad Hafeez, Khawaja Mohiuddin, and Sohaib Ahmed. Speaker-dependent live quranic verses recitation recognition system using sphinx-4 framework. In 17th IEEE International Multi Topic Conference 2014, pages 333–337. IEEE,

    work page 2014

  8. [20]

    A Speech Representation Anonymization Framework via Selective Noise Perturbation

    Minh Tran and Mohammad Soleymani. Towards privacy-preserving speech representation for client-side data sharing. arXiv preprint arXiv:2203.14171,

    work page Pith review arXiv

Show all 28 references
  1. [21]

    Fleurs: Few-shot learning evaluation of universal representations of speech

    Alexis Conneau, Min Ma, Simran Khanuja, Yu Zhang, Vera Axelrod, Siddharth Dalmia, Jason Riesa, Clara Rivera, and Ankur Bapna. Fleurs: Few-shot learning evaluation of universal representations of speech. arXiv preprint arXiv:2205.12446,

    arXiv

  2. [22]

    Common voice: A massively-multilingual speech corpus

    Rosana Ardila, Megan Branson, Kelly Davis, Michael Henretty, Michael Kohler, Josh Meyer, Reuben Morais, Lindsay Saunders, Francis M Tyers, and Gregor Weber. Common voice: A massively-multilingual speech corpus. arXiv preprint arXiv:1912.06670,

    1912 arXiv

  3. [23]

    V oice-based gender recognition using neural network

    Kavita Chachadi and SR Nirmala. V oice-based gender recognition using neural network. InInformation and Communi- cation Technology for Competitive Strategies (ICTCS 2020), pages 741–749. Springer,

    2020

  4. [24]

    Data augmentation for low resource languages

    Anton Ragni, Kate M Knill, Shakti P Rath, and Mark JF Gales. Data augmentation for low resource languages. In INTERSPEECH 2014: 15th Annual Conference of the International Speech Communication Association , pages 810–814. International Speech Communication Association (ISCA),

    2014

  5. [25]

    Synthetic data augmen- tation for improving low-resource asr

    Bao Thai, Robert Jimerson, Dominic Arcoraci, Emily Prud’hommeaux, and Raymond Ptucha. Synthetic data augmen- tation for improving low-resource asr. In 2019 IEEE Western New York Image and Signal Processing Workshop (WNYISPW), pages 1–9. IEEE,

    2019

  6. [26]

    Gaussian error linear units (gelus)

    Dan Hendrycks and Kevin Gimpel. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415,

    arXiv

  7. [27]

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

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805,

    arXiv

  8. [1996]

    Investigating transformer encoders and fusion strategies for speech emotion recognition in emergency call center conversations

    Theo Deschamps-Berger, Lori Lamel, and Laurence Devillers. Investigating transformer encoders and fusion strategies for speech emotion recognition in emergency call center conversations. In Companion Publication of the 2022 International Conference on Multimodal Interaction, p...

    2022

  9. [2004]

    Towards fast and accurate streaming end-to-end asr

    Bo Li, Shuo-yiin Chang, Tara N Sainath, Ruoming Pang, Yanzhang He, Trevor Strohman, and Yonghui Wu. Towards fast and accurate streaming end-to-end asr. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6069–6073. IEEE,

    2020

  10. [2013]

    O’Shea, Johnathan Corgan, and T

    Timothy J. O’Shea, Johnathan Corgan, and T. Charles Clancy. Unsupervised representation learning of structured radio communication signals. In 2016 First International Workshop on Sensing, Processing and Learning for Intelligent Machines (SPLINE), pages 1–5,

    2016

  11. [2014]

    Robust speech recognition via large-scale weak supervision

    14 Whisper Turns Stronger: Augmenting Wav2Vec 2.0 for Superior ASR in Low-Resource LanguagesA PREPRINT Alec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, and Ilya Sutskever. Robust speech recognition via large-scale weak supervision. arXiv preprint arXiv:2...

    arXiv

  12. [2016]

    Thanh Toan Nguyen, Minh Tam Pham, Thanh Tam Nguyen, Thanh Trung Huynh, Quoc Viet Hung Nguyen, Thanh Tho Quan, et al

    doi:10.1109/SPLIM.2016.7528397. Thanh Toan Nguyen, Minh Tam Pham, Thanh Tam Nguyen, Thanh Trung Huynh, Quoc Viet Hung Nguyen, Thanh Tho Quan, et al. Structural representation learning for network alignment with self-supervised anchor links. Expert Systems with Applications, 16...

    2016

  13. [2017]

    Lstm based similarity measurement with spectral clustering for speaker diarization

    Qingjian Lin, Ruiqing Yin, Ming Li, Hervé Bredin, and Claude Barras. Lstm based similarity measurement with spectral clustering for speaker diarization. arXiv preprint arXiv:1907.10393,

    1907 arXiv

  14. [2018]

    Hierarchical conditional end-to-end asr with ctc and multi-granular subword units

    Yosuke Higuchi, Keita Karube, Tetsuji Ogawa, and Tetsunori Kobayashi. Hierarchical conditional end-to-end asr with ctc and multi-granular subword units. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 7797–7801. IEEE,

    2022

  15. [2019]

    Improving noise robust automatic speech recognition with single-channel time-domain enhancement network

    Keisuke Kinoshita, Tsubasa Ochiai, Marc Delcroix, and Tomohiro Nakatani. Improving noise robust automatic speech recognition with single-channel time-domain enhancement network. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASS...

    2020

  16. [2020]

    Personalizing asr for dysarthric and accented speech with limited data

    Joel Shor, Dotan Emanuel, Oran Lang, Omry Tuval, Michael Brenner, Julie Cattiau, Fernando Vieira, Maeve McNally, Taylor Charbonneau, Melissa Nollstadt, et al. Personalizing asr for dysarthric and accented speech with limited data. arXiv preprint arXiv:1907.13511,

    1907 arXiv

  17. [2021]

    Dialectal arabic speech recognition using cnn-lstm based on end-to-end deep learning

    Hamzah A Alsayadi, Salah Al-Hagree, Fahd A Alqasemi, and Abdelaziz A Abdelhamid. Dialectal arabic speech recognition using cnn-lstm based on end-to-end deep learning. In 2022 2nd International Conference on Emerging Smart Technologies and Applications (eSmarTA), pages 1–8. IEEE,

    2022

  18. [2022]

    Clovacall: Korean goal-oriented dialog speech corpus for automatic speech recognition of contact centers

    Jung-Woo Ha, Kihyun Nam, Jingu Kang, Sang-Woo Lee, Sohee Yang, Hyunhoon Jung, Eunmi Kim, Hyeji Kim, Soojin Kim, Hyun Ah Kim, et al. Clovacall: Korean goal-oriented dialog speech corpus for automatic speech recognition of contact centers. arXiv preprint arXiv:2004.09367,

    2004 arXiv

  19. [2023]

    A thousand words are worth more than one recording: Nlp based speaker change point detection

    Or Haim Anidjar, Chen Hajaj, Amit Dvir, and Issachar Gilad. A thousand words are worth more than one recording: Nlp based speaker change point detection. arXiv preprint arXiv:2006.01206,

    2006 arXiv

  20. [2024]

    Anuroop Sriram, Michael Auli, and Alexei Baevski

    doi:10.1109/CCDC62350.2024.10587399. Anuroop Sriram, Michael Auli, and Alexei Baevski. Wav2vec-aug: Improved self-supervised training with limited data. arXiv preprint arXiv:2206.13654,

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.