Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Voice Conversion Improves Cross-Domain Robustness for Spoken Arabic Dialect Identification

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 Voice conversion, applied as a training-time re-synthesis that shares target speakers across dialect classes, substantially improves Arabic dialect identification on unseen domains by removing speaker identity shortcuts.

desk verdict A genuinely useful bias-mitigation idea backed by a clean controlled experiment, but the headline cross-domain gain rests on an author-built dataset and unmatched training steps, so read it with caution. read the letter →

arxiv 2505.24713 v1 pith:H7LVYKOV submitted 2025-05-30 cs.CL cs.SDeess.AS

Badr M. Abdullah , Matthew Baas , Bernd Möbius , Dietrich Klakow This is my paper
classification cs.CLcs.SDeess.AS
keywords ArabicdialectidentificationvoiceconversionkNN-VCcross-domainrobustnessspeakerbiasMADIS-5MMSfine-tuningdataaugmentation
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 argues that voice conversion—re-synthesizing training speech in a different speaker's voice—can make Arabic dialect identification (ADI) robust to recording conditions and genres that never appear in training. Fine-tuning a massively multilingual speech model on a mix of natural and converted speech raises average zero-shot cross-domain accuracy from 60.2% to 80.7% on a newly built four-domain test set and pushes in-domain ADI-5 accuracy to 85.3%, above the previous 84.7% ensemble result. The authors' controlled experiments indicate the operative mechanism is not generic data augmentation but bias removal: converting all dialects' training utterances into a shared set of target voices destroys the accidental correlation between speaker identity and dialect label that models otherwise exploit. The paper thus positions voice conversion as a data-centric cure for the shortcut-learning that limits dialect and accent classification.

What carries the argument

The central object is the re-synthesized training set $\tilde{\mathcal{D}} = \{(C_\theta(x_i, v_i), y_i)\}$ created by nearest-neighbor voice conversion (kNN-VC), a text-free method that transfers an utterance into a target voice from a few reference samples. Each natural segment $x_i$ is converted into a target voice $v_i$ drawn from a small pool of Arabic speakers, and the same pool is used for every dialect so that speaker identity is no longer predictive of the dialect label. The analysis experiment toggles this property directly: a unified speaker pool (unbiased) produces strong in-domain and cross-domain results, while dialect-disjoint pools (biased) drive accuracy to chance. This contrast is what separates voice conversion from ordinary acoustic perturbation in the paper's argument.

What would settle it

Evaluate the released model and the natural-speech MMS baseline on an external multi-domain Arabic speech set labeled by a different team of annotators; if the relative accuracy gain over the baseline falls toward the roughly 11% achieved by pitch-shift augmentation instead of the reported 34.1%, the claim that voice conversion specifically removes cross-domain fragility would fail. A second check: if the same VC recipe applied to a corpus whose speakers already overlap across dialects does not shrink the gap to the natural baseline, the speaker-bias explanation would need revision.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that training a wav2vec2-family MMS model on re-synthesized speech—produced by nearest-neighbor voice conversion with target voices shared across all dialect classes—makes the model classify dialect rather than speaker. The best configuration reaches 85.32% in-domain accuracy, a 12.35% relative improvement over the natural-speech MMS baseline, and 80.73% average accuracy across radio, TEDx, TV drama, and theater, a 34.07% relative cross-domain improvement. Traditional audio augmentations, including SpecAugment, pitch shift, simulated room impulse response, and additive noise, lag behind even when combined (81.64% in-domain, 70.75% cross-domain). A controlled experiment isolates the mechanism: training on converted speech only, with a speaker pool shared across dialects, gives 83.38% in-domain and 76.61% cross-domain, whereas a deliberately biased pool with a disjoint set of voices per dialect collapses to 27.33% in-domain and 24.32% cross-domain, close to chance. From this the paper concludes that speaker-dialect correlation in training data is a major source of cross-domain fragility and that voice conversion removes that shortcut.

Load-bearing premise

The load-bearing premise is that the authors' manually curated MADIS-5 test set—about 12 hours of radio, TEDx, TV-drama, and theater speech labeled by two native Arabic speakers—is a fair and representative measure of how ADI systems behave on unseen real-world domains.

Editorial extensions

If this is right

  • In-domain ADI-5 accuracy becomes 85.3%, ahead of the 84.7% fusion of ResNet and ECAPA systems, so a single VC-trained model replaces an ensemble.
  • Cross-domain average accuracy jumps from 60.2% to 80.7% on MADIS-5, with radio and TEDx accuracy roughly matching in-domain levels.
  • Training on converted speech alone, without any natural speech, is enough for large gains (83.4% in-domain, 76.6% cross-domain), so the method works even when the available labeled audio cannot be released.
  • Traditional augmentation (SpecAugment, pitch shift, RIR, noise), even combined, gives at most 81.6% in-domain and 70.8% cross-domain, so the paper's gain is not duplicated by generic acoustic perturbation.

Reading between the lines

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

  • Not claimed by the paper but directly testable: a natural-speech corpus whose speaker set overlaps across dialects should reproduce much of the VC gain; if it does not, the speaker-bias explanation is incomplete.
  • A neighbouring extension: the same shared-target-voice recipe could be applied to accent identification or to clinical speech classification, with the expected gain proportional to how strongly speaker identity predicts the label in the training data.
  • A further step not in the paper: instead of a single global speaker pool, one could use a mixture of target voices whose assignment to dialect labels changes across training epochs, which would preserve dialect phonetics while still breaking the speaker shortcut.
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 studies cross-domain robustness of spoken Arabic dialect identification (ADI). It proposes a data-centric training strategy that uses nearest-neighbor voice conversion to re-synthesize each training utterance in a small set of target voices, then fine-tunes MMS on the union of natural and re-synthesized speech. The authors introduce a new manually curated four-domain evaluation set, MADIS-5, and report that their best model reaches 85.32% in-domain accuracy on ADI-5 (a new state of the art) and 80.73% average cross-domain accuracy on MADIS-5, a relative improvement of 34.07% over an MMS baseline. A controlled experiment comparing an 'unbiased' voice pool shared across dialects with a 'biased' dialect-specific voice pool is used to argue that voice conversion improves ADI by removing a speaker--dialect shortcut.

Significance. If the empirical claims hold, the paper makes a practical and conceptually useful contribution: it shows that a simple, transcription-free voice conversion method can improve both in-domain and zero-shot cross-domain ADI, and it provides a mechanistic explanation (speaker-bias removal) that is uncommon in this literature. The release of the model and the MADIS-5 dataset is a concrete asset for the community. The in-domain state-of-the-art result is on a standard benchmark, which lends credibility. However, the central cross-domain conclusion rests on a self-curated, non-externally validated dataset, and several comparisons are not fully controlled, so the generalization claim is not yet established at the level the paper states.

major comments (4)
  1. [§4.1/§4.3, Table 1] The baseline and augmented/VC models are not trained under matched conditions. The speech baselines are fine-tuned for 6 epochs on N natural samples (§4.1), while all augmentation and VC models are trained for 3 epochs on the combined natural plus re-synthesized data (§4.3). For 2×N data this matches the baseline's number of sample presentations, but for 5×N data (the 'All Augmentations' row and the four-voice VC row) the model sees 15N samples versus 6N for the baseline. The headline +34.07% cross-domain improvement over the MMS baseline therefore conflates the VC method with a 2.5× increase in data and optimization steps. Please add a matched-step comparison, e.g., train the baseline for 15 effective epochs on N, or subsample the 5×N data so that all models see the same total number of samples.
  2. [§3.2/§5.3, Table 1] The entire cross-domain evaluation, including the paper's strongest claim of +34.07% relative improvement and 'state-of-the-art results across all domains,' is measured on MADIS-5, a dataset manually curated, segmented, and labeled by the authors. No external or independently annotated corpus is used to validate that this benchmark reflects real-world cross-domain conditions. The annotation section reports only that the two annotators agreed categorically except for 2.3% of radio segments labeled as MSA versus dialect. To make the cross-domain claim load-bearing, the authors should provide a detailed annotation protocol, per-domain utterance counts and durations, and ideally evaluate on an existing independent Arabic dialect corpus (e.g., a subset of MGB-5 or ADI-17) to show that the gains are not artifacts of the curation choices.
  3. [Table 1/Table 3] No variability or significance measures are reported. The VC rows in Table 1 are stated to be averaged over four runs with different target-speaker sets, but no standard deviation, confidence interval, or significance test is given, and the MMS baseline appears to be a single run. The large differences in Table 3 are unlikely to be noise, but the finer comparisons in Table 1 (e.g., VC with two voices versus four voices, or VC versus the all-augmentations row) require at least error bars and a paired test such as McNemar's test on utterance-level predictions before the relative ranking can be interpreted reliably.
  4. [§6, Table 3] The controlled experiment does not fully isolate the speaker-bias factor. The unbiased condition uses a unified set of 12 target speakers shared across all dialects, while the biased condition uses 60 dialect-specific speakers (12 per dialect). The two conditions therefore differ not only in the speaker–dialect association but also in the total number of target voices, the amount of speaker variability, and possibly the acoustic diversity of the re-synthesized training set. To support the claim that voice conversion 'eliminates the speaker bias,' the biased condition should use the same 12 pooled voices but with a deterministic disjoint mapping from dialects to voices, so that only the association is manipulated while voice-pool size is held constant.
minor comments (5)
  1. [Abstract] The abstract says 'consistent improvements of up to +34.1% in accuracy across domains,' but Table 1 defines this as a relative improvement over the MMS baseline; the abstract should say 'relative improvement' to avoid readers interpreting it as an absolute accuracy gain.
  2. [§5.1] The text states that the phone-based SVM scores 57.90% and Arabic BERT 59.40% in the cross-domain setting, but Table 1 lists their averages as 58.58% and 59.55%, respectively; the in-text numbers appear to be typos or refer to a different subset and should be aligned with the table.
  3. [§6] The parenthetical explanation in the unbiased condition says 'target speakers for VC are uniformly distributed across dialects,' which is difficult to reconcile with the preceding sentence that a unified set of 12 target speakers is used across all dialects; please clarify whether the same voice pool is used for every dialect or whether target speakers were chosen to represent different dialects.
  4. [§4.3] The target voices are described as 'native Arabic voices from LibriVox audio books,' but their dialect backgrounds (e.g., MSA-only vs. regional dialects) are not reported; since the method is meant to preserve dialect cues while changing speaker identity, the dialectal content of the target voices is potentially relevant and should be stated or analyzed.
  5. [Table 2] The state-of-the-art comparison in Table 2 appears to use previously reported numbers that may come from different experimental setups; please state explicitly whether the same ADI-5 train/validation/test split is used for all systems, since MMS-VC is trained on additional re-synthesized data.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported VC gains are empirical results on held-out benchmarks, not consequences of the method's definition or of self-citation.

full rationale

The paper makes no theoretical claim whose derivation reduces to its own inputs. Voice conversion is applied as a fixed, externally published method (k-NN VC) and its effect is measured on held-out natural speech: the ADI-5 evaluation set and the newly collected MADIS-5 set. The number of target voices is a hyperparameter, not a parameter fitted to test labels, and the test segments are explicitly left unmodified. The Section 6 unbiased/biased experiment is a controlled manipulation: 'unbiased' is defined by sharing the same target speakers across dialects, and 'biased' by using disjoint dialect-specific target speakers; observing that the unbiased model achieves far higher accuracy is an empirical confirmation of the speaker-bias mechanism, not a conclusion entailed by the construction alone, because accuracy is always computed on natural, unmodified test utterances. The author-overlap citations ([13], [16], which include co-author Matthew Baas) are tool and prior-method references; the paper does not derive its central claim from these citations, and the cited k-NN VC method is independently published and used as a component. The MADIS-5 benchmark is author-created and manually labeled, which is a legitimate concern for external validity but is not circularity: no label or domain is derived from the model under test, and no fitted quantity is renamed as a prediction.

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

No invented entities. The only axioms are empirical domain assumptions about data quality, VC fidelity, and domain representativeness. These are typical for an applied ML paper and are partially mitigated by releasing the dataset and model.

assumptions (3)
  • domain assumption The coarse-grained dialect labels in ADI-5 and MADIS-5 are correct and representative.
    The paper relies on the ground-truth quality of the training labels (Aljazeera broadcast) and on the two native speakers' annotation of MADIS-5 (Section 3.2). Errors here would directly affect all reported accuracies.
  • domain assumption k-NN voice conversion preserves dialect-discriminative content of the source utterance while changing only speaker identity.
    The entire method and the bias-mitigation explanation in Section 6 assume that after conversion the re-synthesized speech still contains the phonetic and lexical cues needed for dialect classification. The in-domain results on natural test speech provide indirect support, but the assumption is not formally proven.
  • domain assumption The four MADIS-5 domains (radio, TEDx, TV dramas, theater) are a representative sample of unseen domains for Arabic speech.
    The cross-domain generalization claim is measured only on this self-curated set; if the set is not representative of real deployment conditions, the reported 34% gain may not transfer to other domains.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Voice Conversion Improves Cross-Domain Robustness for Spoken Arabic Dialect Identification." pith.science (2026). https://pith.science/paper/H7LVYKOV

@misc{pith2026250524713,
  author       = {Pith},
  title        = {Pith review of: Voice Conversion Improves Cross-Domain Robustness for Spoken Arabic Dialect Identification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H7LVYKOV}},
  note         = {Machine review of arXiv:2505.24713}
}
read the original abstract

Arabic dialect identification (ADI) systems are essential for large-scale data collection pipelines that enable the development of inclusive speech technologies for Arabic language varieties. However, the reliability of current ADI systems is limited by poor generalization to out-of-domain speech. In this paper, we present an effective approach based on voice conversion for training ADI models that achieves state-of-the-art performance and significantly improves robustness in cross-domain scenarios. Evaluated on a newly collected real-world test set spanning four different domains, our approach yields consistent improvements of up to +34.1% in accuracy across domains. Furthermore, we present an analysis of our approach and demonstrate that voice conversion helps mitigate the speaker bias in the ADI dataset. We release our robust ADI model and cross-domain evaluation dataset to support the development of inclusive speech technologies for Arabic.

Figures

Figures reproduced from arXiv: 2505.24713 by the authors.

Figure 1
Figure 1. Cross-domain performance evaluation: Starting by fine-tuning the massively multilingual speech (MMS) model as a baseline, voice conversion (MMS + VC) consistently outper￾forms traditional data augmentation (MMS + Aug.) across four unseen domains, achieving up to +34.1% average relative im￾provement in Arabic dialect identification accuracy. evaluations of ADI systems have been limited to in-domain set￾tings where te… view at source ↗
Figure 2
Figure 2. Dialect distribution across the domains of MADIS-5. 3.2. Evaluation Dataset: MADIS-5 Benchmark We manually curate a dataset for multi-domain ADI in speech (MADIS-5) to facilitate evaluation of cross-domain robustness of ADI systems. Our dataset comprises ∼12 hours of speech (4854 utterances) collected from four different public sources with varying similarity to the TV broadcast domain of ADI￾5. The recordings were … 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]

    Voice Conversion Improves Cross-Domain Robustness for Spoken Arabic Dialect Identification

    Introduction Arabic is the native language of more than 320 million peo- ple geographically distributed across the Middle East and North Africa [1]. Throughout the Arabic-speaking world, Modern Standard Arabic (MSA) serves as the official language and the medium of formal communication and news broadcasts. How- ever, MSA is not naturally acquired and func...

    work page Pith review arXiv 2025

  2. [2]

    Given an Arabic speech sample x, the goal is to predict the speaker’s dialecty ∈ Y, where Y is a closed set of dialects

    Voice Conversion for ADI We formalize ADI as a classification problem. Given an Arabic speech sample x, the goal is to predict the speaker’s dialecty ∈ Y, where Y is a closed set of dialects. To do so, we require a dataset of N natural speech samples, each paired with a dialect annotation: D = n (xi, yi) oN i=1 (1) The dataset D is used to train a model v...

    work page

  3. [3]

    Datasets 3.1. Training Dataset: MGB-3 ADI-5 As our training dataset, we use the MGB-3 ADI-5 dataset, which is a widely-used ADI resource with coarse-grained di- alect labels derived from Aljazeera TV broadcast [7]. It con- sists of approximately 14.6k samples ( ∼53.6 hours) contain- ing speech segments of MSA as well as four Arabic dialect groups based on...

    work page

  4. [4]

    Experimental Setup To evaluate the effectiveness of voice conversion for improv- ing ADI, we compare it against several strong text and speech baselines as well as various audio augmentation techniques. 4.1. Text and Speech Baselines For text baselines, we transcribe the datasets using two pub- licly available models: (1) a universal phoneme recognizer fo...

    work page

  5. [5]

    The dialect labels were then verified by another native Arabic speaker with competence and keen interest in different Arabic dialects

    The recordings were manually segmented and labeled by a native Arabic speaker with linguistic expertise (PhD in Compu- tational Linguistics) and extensive exposure to Arabic language variation. The dialect labels were then verified by another native Arabic speaker with competence and keen interest in different Arabic dialects. The data sources are: • Radi...

    work page 2024

  6. [6]

    in-the-wild

    Experiments and Results 5.1. Baselines Text baselines. On the in-domain test set, the phone-based SVM (66.82%) outperforms both the character-based SVM (50.00%) and Arabic BERT (62.73%). This shows that text- based classifiers trained on ASR transcripts are not reliable for the ADI task since ASR models are trained on MSA speech and normalize dialect-spec...

    work page

  7. [7]

    Here, we investigate why voice conversion yields such substantial im- provements and better cross-domain generalizations

    Model Analysis In the previous section, we established that voice conversion is an effective method for training robust ADI systems. Here, we investigate why voice conversion yields such substantial im- provements and better cross-domain generalizations. Our hy- pothesis is that re-synthesizing the training data using voice conversion helps normalize spea...

    work page

  8. [8]

    The remarkable performance gain achieved through voice conversion (+34.07% cross-domain) cannot be matched by traditional data augmentation techniques

    Discussion and Conclusion Our experiments demonstrated that voice conversion signifi- cantly improves the generalization of ADI systems, particularly in cross-domain scenarios. The remarkable performance gain achieved through voice conversion (+34.07% cross-domain) cannot be matched by traditional data augmentation techniques. Our analysis with controlled...

    work page

Show all 39 references
  1. [9]

    We sincerely thank Aravind Krishnan for his valuable com- ments on the work presented in this paper

    Acknowledgements We thank the anonymous reviewers for their positive feedback. We sincerely thank Aravind Krishnan for his valuable com- ments on the work presented in this paper. This research is funded by the Deutsche Forschungsgemeinschaft (DFG, Ger- man Research Foundation...

  2. [10]

    Brown and S

    K. Brown and S. Ogilvie, Concise encyclopedia of languages of the world. Elsevier, 2010

    2010

  3. [11]

    G. Khan, M. P. Streck, and J. C. Watson, The Semitic languages: An international handbook. Walter de Gruyter, 2011, vol. 36

    2011

  4. [12]

    N. Y . Habash, Introduction to Arabic natural language process- ing. Morgan & Claypool Publishers, 2010

    2010

  5. [13]

    Casablanca: Data and models for multidialectal Arabic speech recognition,

    B. Talafha, K. Kadaoui, S. M. Magdy, M. Habiboullah, C. M. Chafei, A. O. El-Shangiti, H. Zayed, M. C. Tourad, R. Alhamouri, R. Assi, A. Alraeesi, H. Mohamed, F. Alwajih, A. Mohamed, A. El Mekki, E. M. B. Nagoudi, B. D. M. Saadia, H. A. Alsayadi, W. Al-Dhabyani, S. Shatnawi, Y ...

    2024

  6. [14]

    V oxArabica: A robust dialect-aware Arabic speech recognition system,

    A. Waheed, B. Talafha, P. Sullivan, A. Elmadany, and M. Abdul- Mageed, “V oxArabica: A robust dialect-aware Arabic speech recognition system,” in Proceedings of ArabicNLP 2023. Singa- pore (Hybrid): Association for Computational Linguistics, Dec. 2023

    2023

  7. [15]

    On the robust- ness of arabic speech dialect identification,

    P. Sullivan, A. Elmadany, and M. Abdul-Mageed, “On the robust- ness of arabic speech dialect identification,” in Interspeech 2023, 2023, pp. 5326–5330

    2023

  8. [16]

    Speech recognition challenge in the wild: Arabic mgb-3,

    A. Ali, S. V ogel, and S. Renals, “Speech recognition challenge in the wild: Arabic mgb-3,” in 2017 IEEE Automatic Speech Recog- nition and Understanding Workshop (ASRU). IEEE, 2017, pp. 316–322

    2017

  9. [17]

    Convolutional neural networks and language embeddings for end-to-end dialect recognition,

    S. Shon, A. Ali, and J. Glass, “Convolutional neural networks and language embeddings for end-to-end dialect recognition,” arXiv preprint arXiv:1803.04567, 2018

    2018 arXiv

  10. [18]

    The mgb-5 challenge: Recogni- tion and dialect identification of dialectal arabic speech,

    A. Ali, S. Shon, Y . Samih, H. Mubarak, A. Abdelali, J. Glass, S. Renals, and K. Choukri, “The mgb-5 challenge: Recogni- tion and dialect identification of dialectal arabic speech,” in 2019 IEEE Automatic Speech Recognition and Understanding Work- shop (ASRU). IEEE, 2019, pp. ...

    2019

  11. [19]

    Transformer-based arabic dialect identification,

    W. Lin, M. Madhavi, R. K. Das, and H. Li, “Transformer-based arabic dialect identification,” in2020 International Conference on Asian Language Processing (IALP). IEEE, 2020, pp. 192–196

    2020

  12. [20]

    Yet another model for arabic di- alect identification,

    A. Kulkarni and H. Aldarmaki, “Yet another model for arabic di- alect identification,” in Proceedings of ArabicNLP 2023 , 2023, pp. 435–440

    2023

  13. [21]

    An overview of voice conversion systems,

    S. H. Mohammadi and A. Kain, “An overview of voice conversion systems,” Speech Communication, vol. 88, pp. 65–82, 2017

    2017

  14. [22]

    V oice conversion can improve asr in very low-resource settings,

    M. Baas and H. Kamper, “V oice conversion can improve asr in very low-resource settings,” in Proc. Interspeech 2022, 2022, pp. 3513–3517

    2022

  15. [23]

    Asr data augmentation in low- resource settings using cross-lingual multi-speaker tts and cross- lingual voice conversion,

    E. Casanova, C. Shulby, A. Korolev, A. C. Junior, A. d. S. Soares, S. Alu ´ısio, and M. A. Ponti, “Asr data augmentation in low- resource settings using cross-lingual multi-speaker tts and cross- lingual voice conversion,” in Proc. Interspeech 2023 , 2023, pp. 1244–1248

    2023

  16. [24]

    V oice conversion based aug- mentation and a hybrid cnn-lstm model for improving speaker- independent keyword recognition on limited datasets,

    Y . A. Wubet and K.-Y . Lian, “V oice conversion based aug- mentation and a hybrid cnn-lstm model for improving speaker- independent keyword recognition on limited datasets,” IEEE Ac- cess, vol. 10, pp. 89 170–89 180, 2022

    2022

  17. [25]

    V oice conversion with just nearest neighbors,

    M. Baas, B. van Niekerk, and H. Kamper, “V oice conversion with just nearest neighbors,” in Interspeech 2023 , 2023, pp. 2053– 2057

    2023

  18. [26]

    Autovc: Zero-shot voice style transfer with only au- toencoder loss,

    K. Qian, Y . Zhang, S. Chang, X. Yang, and M. Hasegawa- Johnson, “Autovc: Zero-shot voice style transfer with only au- toencoder loss,” in International Conference on Machine Learn- ing. PMLR, 2019, pp. 5210–5219

    2019

  19. [27]

    Speech resynthesis from dis- crete disentangled self-supervised representations,

    A. Polyak, Y . Adi, J. Copet, E. Kharitonov, K. Lakhotia, W.-N. Hsu, A. Mohamed, and E. Dupoux, “Speech resynthesis from dis- crete disentangled self-supervised representations,” inInterspeech 2021, 2021, pp. 3615–3619

    2021

  20. [28]

    Hifi-gan: Generative adversarial net- works for efficient and high fidelity speech synthesis,

    J. Kong, J. Kim, and J. Bae, “Hifi-gan: Generative adversarial net- works for efficient and high fidelity speech synthesis,” Advances in neural information processing systems , vol. 33, pp. 17 022– 17 033, 2020

    2020

  21. [29]

    Specaugment: A simple data augmentation method for automatic speech recognition,

    D. S. Park, W. Chan, Y . Zhang, C.-C. Chiu, B. Zoph, E. D. Cubuk, and Q. V . Le, “Specaugment: A simple data augmentation method for automatic speech recognition,” Interspeech 2019 , p. 2613, 2019

    2019

  22. [30]

    Where are you from? geolocating speech and ap- plications to language identification,

    P. Foley, M. Wiesner, B. Odoom, L. P. Garcia Perera, K. Murray, and P. Koehn, “Where are you from? geolocating speech and ap- plications to language identification,” in Proceedings of NAACL. Association for Computational Linguistics, Jun. 2024

    2024

  23. [31]

    Multilingual tedx corpus for speech recognition and translation,

    E. Salesky, M. Wiesner, J. Bremerman, R. Cattoni, M. Negri, M. Turchi, D. W. Oard, and M. Post, “Multilingual tedx corpus for speech recognition and translation,” in Proceedings of Inter- speech, 2021

    2021

  24. [32]

    Estimating the level of dialectness predicts inter-annotator agreement in multi-dialect Arabic datasets,

    A. Keleg, W. Magdy, and S. Goldwater, “Estimating the level of dialectness predicts inter-annotator agreement in multi-dialect Arabic datasets,” in Proceedings of ACL , Bangkok, Thailand, Aug. 2024

    2024

  25. [33]

    ALDi: Quantifying the Arabic level of dialectness of text,

    A. Keleg, S. Goldwater, and W. Magdy, “ALDi: Quantifying the Arabic level of dialectness of text,” in Proceedings of EMNLP , Singapore, Dec. 2023

    2023

  26. [34]

    When sparse traditional models outperform dense neural networks: the curious case of discriminating between similar languages,

    M. Medvedeva, M. Kroon, and B. Plank, “When sparse traditional models outperform dense neural networks: the curious case of discriminating between similar languages,” in Proceedings of the Fourth Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial), 2017, pp...

    2017

  27. [35]

    The interplay of variant, size, and task type in Arabic pre-trained language models,

    G. Inoue, B. Alhafni, N. Baimukan, H. Bouamor, and N. Habash, “The interplay of variant, size, and task type in Arabic pre-trained language models,” in Proceedings of the Sixth Arabic Natural Language Processing Workshop. Kyiv, Ukraine (Online): As- sociation for Computational...

    2021

  28. [36]

    XLSR-53: Uni- versal cross-lingual speech representations,

    A. Babu, C. Wang, A. Tjandra, K. Lakhotia, Q. Xu, N. Goyal, K. Singh, P. von Platen, Y . Saraf, J. Pinoet al., “XLSR-53: Uni- versal cross-lingual speech representations,” inProc. Interspeech, 2021, pp. 3429–3433

    2021

  29. [37]

    XLS-R: Self-supervised cross-lingual speech representa- tion learning at scale,

    A. Baevski, W.-N. Hsu, Q. Xu, A. Babu, J. Gu, and A. Con- neau, “XLS-R: Self-supervised cross-lingual speech representa- tion learning at scale,”IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 30, pp. 2634–2650, 2022

    2022

  30. [38]

    Scaling speech technology to 1,000+ languages,

    V . Pratap, A. Tjandra, B. Shi, P. Tomasello, A. Babu, S. Kundu, A. Elkahky, Z. Ni, A. Vyas, M. Fazel-Zarandi et al. , “Scaling speech technology to 1,000+ languages,” Journal of Machine Learning Research, vol. 25, no. 97, pp. 1–52, 2024

    2024

  31. [39]

    MUSAN: A Music, Speech, and Noise Corpus,

    D. Snyder, G. Chen, and D. Povey, “MUSAN: A Music, Speech, and Noise Corpus,” 2015, arXiv:1510.08484v1

    2015 arXiv

Pith tools

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