---
title: "A Comparative Study of Parkinsonian Speech Corpora for Deep"
description: "The study also proposes a reproducible evaluation protocol and gives guidance for designing future clinical speech corpora. 25 references."
authors: ["Clara Ponchard", "Pierre Serrano"]
year: 2026
venue: "Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC)"
source: "https://aclanthology.org/2026.bucc-1.2/"
doi: "10.63317/27zb48j5vv5f"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/ponchard-2026-comparative-study-parkinsonian-speech-corpora/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/ponchard-2026-comparative-study-parkinsonian-speech-corpora/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/20/) / A Comparative Study of Parkinsonian Speech Corpora for Deep Learning-Based Detection of Dysarthria

# A Comparative Study of Parkinsonian Speech Corpora for Deep Learning-Based Detection of Dysarthria

**Authors:** Clara Ponchard and Pierre Serrano · **Year:** 2026 · **Venue:** [Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC)](https://md.research.elamur.ai/free/en/venues/bucc/) ([BUCC 2026](https://md.research.elamur.ai/free/en/venues/bucc/2026/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.bucc-1.2/) · [PDF](https://aclanthology.org/2026.bucc-1.2.pdf) · [DOI 10.63317/27zb48j5vv5f](https://doi.org/10.63317/27zb48j5vv5f) · [OpenAlex](https://openalex.org/W7166535335)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Hypokinetic dysarthria is a speech disorder that affects most people with Parkinson's disease. Clinicians still assess it mainly by ear, and automatic detectors are usually built and tested on a single dataset. The authors ask whether existing Parkinsonian speech corpora are similar enough to be pooled. They test this indirectly: a classifier is trained on one corpus or on combinations of corpora, then evaluated within the same corpus, on other corpora, and in out-of-domain settings. The four corpora are Neurovoz, IPVSD, MDVR-KCL and AHN, totalling 380 speakers of Spanish, Italian, English and French. They differ in speech tasks, recording equipment and medication state. The model borrows from speaker-recognition and deepfake-detection practice. A pre-trained Wav2Vec 2.0 XLS-R stays frozen, and only an MHFA head that combines several SSL layers is trained. Multi-corpus training improved robustness and generalization, but some corpus combinations were much less compatible than others. The study also proposes a reproducible evaluation protocol and gives guidance for designing future clinical speech corpora.

## Key points
- **Corpora:**
  - Neurovoz: Castilian Spanish, 53 PD and 54 HC.
  - IPVSD: Italian, 28 PD and 37 HC.
  - MDVR-KCL: English, 16 PD and 21 HC, recorded on a smartphone to mimic a phone call.
  - AHN: French, 120 PD and 51 HC.
  - The paper gives the PD total as 211, but the per-corpus figures add up to 217.
- **AHN's distinctive features:** it had not been used for classification before. It is the only corpus with OFF-DOPA recordings, and it includes neurologist ratings on UPDRS Item 18 (scores 0–3). The task analysed is a sentence repetition targeting plosive consonants.
- **Architecture:** a frozen Wav2Vec 2.0 XLS-R feeds a multi-head feature aggregation (MHFA) head with hidden dimension 128. This lets the model weight the most useful SSL layers rather than relying only on the last-layer embeddings used in most earlier work.
- **Experimental setup:**
  - Data split: 80/20 train/test, with 20% of the training portion held out for development.
  - Robustness: five random speaker-disjoint splits, reported as mean ± SD.
  - Input: audio resampled to 16 kHz and cut into random 1-second chunks.
  - Training: Adam with cross-entropy loss for at least 100 epochs.
- **Evaluation:** the authors recommend fixed-point AUC over plain accuracy for comparing studies. They note that the field has no standard cross-corpus benchmark.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Institut national de recherche en sciences et technologies du numérique](https://md.research.elamur.ai/free/en/labs/institut-national-de-recherche-en-sciences-et-technologies-du-numerique/) — Research institute, Le Chesnay-Rocquencourt, France · 85 works · 25,498 citations

📚 Shelves (7): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 493 works · [🩺 Medicine](https://md.research.elamur.ai/free/en/shelves/medicine/) · 275 works · [🩺 Physiology](https://md.research.elamur.ai/free/en/shelves/medicine/physiology/) · 35 works

## Cite

Cite as: Clara Ponchard and Pierre Serrano. 2026. A Comparative Study of Parkinsonian Speech Corpora for Deep Learning-Based Detection of Dysarthria. Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{ponchard2026comparative,
  title = {{A Comparative Study of Parkinsonian Speech Corpora for Deep Learning-Based Detection of Dysarthria}},
  author = {Clara Ponchard and Pierre Serrano},
  booktitle = {Proceedings of the 19th Workshop on Building and Using Comparable Corpora (BUCC)},
  year = {2026},
  doi = {10.63317/27zb48j5vv5f},
  url = {https://doi.org/10.63317/27zb48j5vv5f}
}
```

- [← previous A Comparative Study of Multilingual Fine-tuning and Prompting for Automatic Text Readability Classification in Galician](https://md.research.elamur.ai/free/en/papers/rey-2026-comparative-study-multilingual-fine-tuning/)
- [next → A Comprehensive Full-Form Lexicon for Arabic NLP and Speech Technology](https://md.research.elamur.ai/free/en/papers/haralambous-2026-comprehensive-full-form-lexicon-arabic/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/20/)

## References

25 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Emiro J Ibarra, Julián D Arias-Londoño, Matías Zañartu, and Juan I Godino-Llorente. 2023. Towards a corpus (and language)-independent screening of parkinson’s disease from voice and speech through domain adaptation. Bioengineering, 10(11):1316. [find](https://scholar.google.com/scholar?q=Emiro+J+Ibarra%2C+Juli%C3%A1n+D+Arias%2DLondo%C3%B1o%2C+Mat%C3%ADas+Za%C3%B1artu%2C+and+Juan+I+Godino%2DLlorente%2E+2023%2E+Towards+a+corpus+%28and+language%29%2Dindependent+screening+of+parkinson%E2%80%99s+disease+from+voice+and+speech+through+dom)
2. Arun Babu, Changhan Wang, Andros Tjandra, Kushal Lakhotia, Qiantong Xu, Naman Goyal, Kritika Singh, Patrick von Platen, Yatharth Saraf, Juan Pino, et al. 2022. [Xls-r: Self-supervised cross-lingual speech representation learning at scale](https://md.research.elamur.ai/free/en/papers/babu-2021-xls-r-self-supervised-cross/). In Proc.
3. Interspeech 2022, pages 2278– 2282. [find](https://scholar.google.com/scholar?q=Interspeech+2022%2C+pages+2278%E2%80%93+2282%2E)
4. Ondrej Klempíř and Roman Krupička. 2024. Analyzing wav2vec 1.0 embeddings for cross-database parkinson’s disease detection and speech features extraction. Sensors, 24:5520. [find](https://scholar.google.com/scholar?q=Ondrej+Klemp%C3%AD%C5%99+and+Roman+Krupi%C4%8Dka%2E+2024%2E+Analyzing+wav2vec+1%2E0+embeddings+for+cross%2Ddatabase+parkinson%E2%80%99s+disease+detection+and+speech+features+extraction%2E+Sensors%2C+24%3A5520%2E)
5. Frederic L. Darley, Arnold E. Aronson, and J. R. Brown. 1969. Clusters of Deviant Speech Dimensions in the Dysarthrias. Journal of Speech and Hearing Research, 12(3):462–496. [find](https://scholar.google.com/scholar?q=Frederic+L%2E+Darley%2C+Arnold+E%2E+Aronson%2C+and+J%2E+R%2E+Brown%2E+1969%2E+Clusters+of+Deviant+Speech+Dimensions+in+the+Dysarthrias%2E+Journal+of+Speech+and+Hearing+Research%2C+12%283%29%3A462%E2%80%93496%2E)
6. I. Laaridh. 2017. Évaluation de la parole dysarthrique. Ph.D. thesis, Université d’Avignon. [find](https://scholar.google.com/scholar?q=I%2E+Laaridh%2E+2017%2E+%C3%89valuation+de+la+parole+dysarthrique%2E+Ph%2ED%2E+thesis%2C+Universit%C3%A9+d%E2%80%99Avignon%2E)
7. Stanley Fahn and Richard L. Elton. 1987. The unified parkinson’s disease rating scale. In Stanley Fahn, C. David Marsden, Donald Calne, and Melvin Goldstein, editors, Recent Developments in Parkinson’s Disease, volume 2, pages 153–163. Macmillan Health Care Information, Florham Park. [find](https://scholar.google.com/scholar?q=Stanley+Fahn+and+Richard+L%2E+Elton%2E+1987%2E+The+unified+parkinson%E2%80%99s+disease+rating+scale%2E+In+Stanley+Fahn%2C+C%2E+David+Marsden%2C+Donald+Calne%2C+and+Melvin+Goldstein%2C+editors%2C+Recent+Developments+in+Parkinson%E2%80%99)
8. J. A. Logemann, H. B. Fisher, B. Boshes, and E. R. Blonsky. 1978. Frequency and cooccurrence of vocal tract dysfunctions in the speech of a large sample of parkinson patients. Journal of Speech and Hearing Disorders, 43(1):47–57. [find](https://scholar.google.com/scholar?q=J%2E+A%2E+Logemann%2C+H%2E+B%2E+Fisher%2C+B%2E+Boshes%2C+and+E%2E+R%2E+Blonsky%2E+1978%2E+Frequency+and+cooccurrence+of+vocal+tract+dysfunctions+in+the+speech+of+a+large+sample+of+parkinson+patients%2E+Journal+of+Speech+and+He)
9. Ana S. Ozbolt, Laura Moro-Velazquez, Irene Lina, Adam A. Butala, and Najim Dehak. 2022. Things to consider when automatically detecting parkinson’s disease using the phonation of sustained vowels: Analysis of methodological issues. Applied Sciences, 12(3):991. [find](https://scholar.google.com/scholar?q=Ana+S%2E+Ozbolt%2C+Laura+Moro%2DVelazquez%2C+Irene+Lina%2C+Adam+A%2E+Butala%2C+and+Najim+Dehak%2E+2022%2E+Things+to+consider+when+automatically+detecting+parkinson%E2%80%99s+disease+using+the+phonation+of+sustained+vowels%3A+Ana)
10. Alain Ghio, Gilles Pouchoulin, Antoine Giovanni, Corinne Fredouille, Bernard Teston, Joana Révis, Jean-François Bonastre, Danièle Robert-Rochet, Ping Yu, Maurice Ouaknine, Marie-Dominique Guarella, Christine Spezza, Thierry Legou, and Alain Marchal. 2007. Approches complémentaires pour l’évaluation… [find](https://scholar.google.com/scholar?q=Alain+Ghio%2C+Gilles+Pouchoulin%2C+Antoine+Giovanni%2C+Corinne+Fredouille%2C+Bernard+Teston%2C+Joana+R%C3%A9vis%2C+Jean%2DFran%C3%A7ois+Bonastre%2C+Dani%C3%A8le+Robert%2DRochet%2C+Ping+Yu%2C+Maurice+Ouaknine%2C+Marie%2DDominique+Guarella%2C+Ch)
11. J. Peng, O. Plchot, T. Stafylakis, L. Mošner, L. Burget, and J. Černocký. 2022. An attention-based backend allowing efficient fine-tuning of transformer models for speaker verification. In Proceedings of the IEEE Spoken Language Technology Workshop (SLT), pages 555–562. [find](https://scholar.google.com/scholar?q=J%2E+Peng%2C+O%2E+Plchot%2C+T%2E+Stafylakis%2C+L%2E+Mo%C5%A1ner%2C+L%2E+Burget%2C+and+J%2E+%C4%8Cernock%C3%BD%2E+2022%2E+An+attention%2Dbased+backend+allowing+efficient+fine%2Dtuning+of+transformer+models+for+speaker+verification%2E+In+Proceedings)
12. David Gimeno-Gómez, Catarina Botelho, Anna Pompili, Alberto Abad, and Carlos-D Martínez-Hinarejos. 2025. Unveiling interpretability in self-supervised speech representations for parkinson’s diagnosis. IEEE Journal of Selected Topics in Signal Processing. [find](https://scholar.google.com/scholar?q=David+Gimeno%2DG%C3%B3mez%2C+Catarina+Botelho%2C+Anna+Pompili%2C+Alberto+Abad%2C+and+Carlos%2DD+Mart%C3%ADnez%2DHinarejos%2E+2025%2E+Unveiling+interpretability+in+self%2Dsupervised+speech+representations+for+parkinson%E2%80%99s+diagnosis%2E)
13. S. Pinto, A. Ghio, B. Teston, and F. Viallet. 2010. La dysarthrie au cours de la maladie de parkinson. Revue Neurologique, 166(10):800–810. [find](https://scholar.google.com/scholar?q=S%2E+Pinto%2C+A%2E+Ghio%2C+B%2E+Teston%2C+and+F%2E+Viallet%2E+2010%2E+La+dysarthrie+au+cours+de+la+maladie+de+parkinson%2E+Revue+Neurologique%2C+166%2810%29%3A800%E2%80%93810%2E)
14. Emmy Postma and Cristian Tejedor-Garcia. 2025. Evaluating the Effectiveness of Pre-Trained Audio Embeddings for Classification of Parkinson’s Disease Speech Data. In Interspeech 2025, pages 4603–4607. [find](https://scholar.google.com/scholar?q=Emmy+Postma+and+Cristian+Tejedor%2DGarcia%2E+2025%2E+Evaluating+the+Effectiveness+of+Pre%2DTrained+Audio+Embeddings+for+Classification+of+Parkinson%E2%80%99s+Disease+Speech+Data%2E+In+Interspeech+2025%2C+pages+4603%E2%80%934607%2E)
15. David Grabli. 2017. Maladie de parkinson et syndromes parkinsoniens : les signes moteurs. La Presse Médicale, 46(2, Part 1):187–194. [find](https://scholar.google.com/scholar?q=David+Grabli%2E+2017%2E+Maladie+de+parkinson+et+syndromes+parkinsoniens+%3A+les+signes+moteurs%2E+La+Presse+M%C3%A9dicale%2C+46%282%2C+Part+1%29%3A187%E2%80%93194%2E)
16. S. Sapir. 2014. Multiple factors are involved in the dysarthria associated with parkinson’s disease. Journal of Speech, Language, and Hearing Research, 57(4):1330–1343. [find](https://scholar.google.com/scholar?q=S%2E+Sapir%2E+2014%2E+Multiple+factors+are+involved+in+the+dysarthria+associated+with+parkinson%E2%80%99s+disease%2E+Journal+of+Speech%2C+Language%2C+and+Hearing+Research%2C+57%284%29%3A1330%E2%80%931343%2E)
17. L. Hartelius and P. Svensson. 2009. Speech and swallowing symptoms associated with parkinson’s disease and multiple sclerosis: A survey. Folia Phoniatrica et Logopaedica, 46(1):9–17. [find](https://scholar.google.com/scholar?q=L%2E+Hartelius+and+P%2E+Svensson%2E+2009%2E+Speech+and+swallowing+symptoms+associated+with+parkinson%E2%80%99s+disease+and+multiple+sclerosis%3A+A+survey%2E+Folia+Phoniatrica+et+Logopaedica%2C+46%281%29%3A9%E2%80%9317%2E)
18. Bernard Teston, Alain Ghio, and Bernard Galindo. 1999. A multisensor data acquisition and processing system for speech production investigation. pages 2251–2254. [find](https://scholar.google.com/scholar?q=Bernard+Teston%2C+Alain+Ghio%2C+and+Bernard+Galindo%2E+1999%2E+A+multisensor+data+acquisition+and+processing+system+for+speech+production+investigation%2E+pages+2251%E2%80%932254%2E)
19. M. Hireš, P. Drotár, N. D. Pah, Q. C. Ngo, and D. K. Kumar. 2023. On the inter-dataset generalization of machine learning approaches to parkinson’s disease detection from voice. International Journal of Medical Informatics, 179:105237. [find](https://scholar.google.com/scholar?q=M%2E+Hire%C5%A1%2C+P%2E+Drot%C3%A1r%2C+N%2E+D%2E+Pah%2C+Q%2E+C%2E+Ngo%2C+and+D%2E+K%2E+Kumar%2E+2023%2E+On+the+inter%2Ddataset+generalization+of+machine+learning+approaches+to+parkinson%E2%80%99s+disease+detection+from+voice%2E+International+Journal+)
20. X. Wang, H. Delgado, H. Tak, et al. 2024. Asvspoof 5: crowdsourced speech data, deepfakes, and adversarial attacks at scale. In Proceedings of the ASVspoof 2024 Workshop, pages 1–8. [find](https://scholar.google.com/scholar?q=X%2E+Wang%2C+H%2E+Delgado%2C+H%2E+Tak%2C+et+al%2E+2024%2E+Asvspoof+5%3A+crowdsourced+speech+data%2C+deepfakes%2C+and+adversarial+attacks+at+scale%2E+In+Proceedings+of+the+ASVspoof+2024+Workshop%2C+pages+1%E2%80%938%2E)
21. A. K. Ho, R. Iansek, C. Marigliani, J. L. Bradshaw, and S. Gates. 1998. Speech impairment in a Salah Zaiem, Youcef Kemiche, Titouan Parcollet, Slim Essid, and Mirco Ravanelli. 2025. Speech self-supervised representations benchmarking: a case for larger probing heads. Computer Speech & Language, 89:1… [find](https://scholar.google.com/scholar?q=A%2E+K%2E+Ho%2C+R%2E+Iansek%2C+C%2E+Marigliani%2C+J%2E+L%2E+Bradshaw%2C+and+S%2E+Gates%2E+1998%2E+Speech+impairment+in+a+Salah+Zaiem%2C+Youcef+Kemiche%2C+Titouan+Parcollet%2C+Slim+Essid%2C+and+Mirco+Ravanelli%2E+2025%2E+Speech+self%2Dsuperv)
22. Language Resource References Dimauro, Giovanni and Di Nicola, Vincenzo and Bevilacqua, Vitoantonio and Caivano, Danilo and Girardi, Francesco. 2017. Assessment of Speech Intelligibility in Parkinson’s Disease Using a Speech-To-Text System. [find](https://scholar.google.com/scholar?q=Language+Resource+References+Dimauro%2C+Giovanni+and+Di+Nicola%2C+Vincenzo+and+Bevilacqua%2C+Vitoantonio+and+Caivano%2C+Danilo+and+Girardi%2C+Francesco%2E+2017%2E+Assessment+of+Speech+Intelligibility+in+Parkinson%E2%80%99s)
23. Ghio, Alain and Pouchoulin, Gilles and Viallet, François and Giovanni, Antoine and Woisard, Virginie and Crevier-Buchman, Lise and Hirsch, Fabrice and Fauth, Camille and Fredouille, Corinne. 2021. Du recueil à l’exploitation des corpus de parole « pathologique » : comment accéder à la variation phys… — [doi:10.4000/corpus.5677](https://doi.org/10.4000/corpus.5677) [find](https://scholar.google.com/scholar?q=Ghio%2C+Alain+and+Pouchoulin%2C+Gilles+and+Viallet%2C+Fran%C3%A7ois+and+Giovanni%2C+Antoine+and+Woisard%2C+Virginie+and+Crevier%2DBuchman%2C+Lise+and+Hirsch%2C+Fabrice+and+Fauth%2C+Camille+and+Fredouille%2C+Corinne%2E+2021%2E+Du+)
24. Jaeger, Hagen and Trivedi, Dhaval and Stadtschnitzer, Michael. 2019. Mobile Device Voice Recordings at King’s College London (MDVR-KCL) from both early and advanced Parkinson’s disease patients and healthy controls. Zenodo. PID https://doi.org/10.5281/zenodo.2867216. — [doi:10.5281/zenodo.2867216](https://doi.org/10.5281/zenodo.2867216) [↗](https://doi.org/10.5281/zenodo.2867216)
25. Mendes-Laureano, J. and Gómez-García, J. A. and Guerrero-López, A. and others. 2024. NeuroVoz: a Castillian Spanish corpus of parkinsonian speech. PID https://doi.org/10.1038/s41597-024-04186-z. — [doi:10.1038/s41597-024-04186-z](https://doi.org/10.1038/s41597-024-04186-z) [↗](https://doi.org/10.1038/s41597-024-04186-z)

*Updated 1 November 2026*