Elamur Research⌕
☰
home / papers / 2026 / ParlaSpeech 3.0: Richly Annotated Spoken Parliamentary Corpora of Croatian, Czech, Polish, and Serbian

ParlaSpeech 3.0: Richly Annotated Spoken Parliamentary Corpora of Croatian, Czech, Polish, and Serbian

Summary

Spoken corpora are much rarer than written ones, and Slavic spoken resources are especially small and hard to work with. Existing parliamentary speech corpora mostly pair transcripts with audio for training ASR systems. ParlaSpeech 3.0 addresses this gap for Croatian, Czech, Polish and Serbian. It provides over 6,000 hours of parliamentary speech, built automatically by aligning ParlaMint transcripts and speaker metadata with the parliaments' recordings at the sentence level.

The new release adds automatic annotation layers:

  • All four languages: linguistic analysis following Universal Dependencies, sentiment labels on the text, and detection of filled pauses in the audio.
  • Croatian and Serbian only: precise word- and grapheme-level forced alignment and primary stress marking in multisyllabic words.

The authors demonstrate the value of these layers by studying how sentiment relates to acoustic properties of speech. The four corpora are comparable, so researchers can run cross-lingual studies, which no other parliamentary speech collection supports. The data can be downloaded as JSONL or TextGrid and searched through a concordancer. Potential users include linguists, phoneticians, speech technologists and political scientists.

Key points

  • Five enrichment layers: ParlaSpeech-Ling, ParlaSpeech-Senti, ParlaSpeech-Pause, ParlaSpeech-Align and ParlaSpeech-Stress. Czech and Polish currently have only the Ling, Senti and Pause layers.
  • Linguistic tools: CLASSLA-Stanza was used for Croatian and Serbian, and Stanza for Czech and Polish, both within the UD framework.
  • Speech models:
    • Filled pauses are detected with a speech-transformer model (wav2vecbert2-filledPause) whose performance is close to human annotators.
    • Stress is predicted by a transformer speech-encoder model with about 99% word-level accuracy.
    • Word and grapheme alignment uses an MFA-based model, which clearly beats ASR-based timestamps.
  • Sentiment: labels come from a multilingual text transformer tuned for parliamentary language (XLM-R-ParlaSent, trained on ParlaSent).
  • Availability: the corpus is published on CLARIN.SI (hdl.handle.net/11356/1833). Speakers are mostly middle-aged (median age 50), with a noticeable gender imbalance.

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

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/3b7dnbjr75es

@inproceedings{ljubesic2026parlaspeech,
  title = {{ParlaSpeech 3.0: Richly Annotated Spoken Parliamentary Corpora of Croatian, Czech, Polish, and Serbian}},
  author = {Ljubešić, Nikola and Rupnik, Peter and Porupski, Ivan and Pungeršek, Taja Kuzman},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/3b7dnbjr75es},
  url = {https://doi.org/10.63317/3b7dnbjr75es}
}

References

56 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Itai Allouche, Itay Asael, Rotem Rousso, Vered Dassa, Ann Bradlow, Seung-Eun Kim, Matthew Goldrick, and Joseph Keshet. 2025. How Does a Deep Neural Network Look at Lexical Stress? arXiv preprint arXiv:2508.07229. find
  2. ES Atwell, PA Howarth, and DC Souter. 2003. The ISLE corpus: Italian and German spoken learner’s English. ICAME Journal: International Computer Archive of Modern and Medieval English Journal, 27:5–18. find
  3. Loïc Barrault, Yu-An Chung, Mariano Coria Meglioli, David Dale, Ning Dong, Mark Duppenthaler, Paul-Ambroise Duquenne, Brian Ellis, Hady Elsahar, Justin Haaheim, John Hoffman, Min-Jae Hwang, Hirofumi Inaguma, Christopher Klaiber, Ilia Kulikov, Pengwei Li, Daniel Licht, Jean Maillard, Ruslan Mavlyutov… find our card
  4. Sebastian P. Bayerl, Alexander Wolff von Guden-berg, Florian Hönig, Elmar Nöth, and Korbinian Riedhammer. 2022. KSoF: The Kassel state of fluency dataset – a therapy centered dataset of stuttering. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1780–1787. find
  5. Eric Chen, Zhiyun Lu, Hao Xu, Liangliang Cao, Yu Zhang, and James Fan. 2020. A large scale speech sentiment corpus. In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 6549–6555. find
  6. George Christodoulides, Mathieu Avanzi, and Jean-Philippe Goldman. 2014. DisMo: A morphosyntactic, disfluency and multi-word unit annotator. an evaluation on a corpus of French spontaneous and read speech. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LRE… find
  7. Ringki Das and Thoudam Doren Singh. 2023. Multimodal sentiment analysis: a survey of methods, trends, and challenges. ACM Computing Surveys, 55(13s):1–38. find
  8. Nina Dobrushina and Elena Sokur. 2022. Spoken corpora of Slavic languages. Russian Linguistics, 46(2):77–93. find
  9. Tomaž Erjavec, Matyáš Kopp, Nikola Ljubešić, Taja Kuzman, Paul Rayson, Petya Osenova, Maciej Ogrodniczuk, Çağrı Çöltekin, Danijel Koržinek, Katja Meden, et al. 2025. ParlaMint II: advancing comparable parliamentary corpora across Europe. Language Resources and Evaluation, 59(3):2071–2102. find
  10. Florian Eyben, Klaus R Scherer, Björn W Schuller, Johan Sundberg, Elisabeth André, Carlos Busso, Laurence Y Devillers, Julien Epps, Petri Laukka, Shrikanth S Narayanan, et al. 2015. The Geneva minimalistic acoustic parameter set (GeMAPS) for voice research and affective computing. IEEE transactions… find
  11. Darja Fišer, Jakob Lenardič, and Tomaž Erjavec. 2018. CLARIN’s key resource families. In Proceedings of the eleventh international conference on language resources and evaluation (LREC 2018). find
  12. John J Godfrey, Edward C Holliman, and Jane McDaniel. 1992. SWITCHBOARD: Telephone speech corpus for research and development. In Acoustics, speech, and signal processing, ieee international conference on, volume 1, pages 517–520. IEEE Computer Society. find
  13. Lakshmish Kaushik, Abhijeet Sangwan, and John HL Hansen. 2015. Automatic audio sentiment extraction using keyword spotting. In IN-TERSPEECH, pages 2709–2713. find
  14. Andreas Kirkedal, Marija Stepanović, and Barbara Plank. 2020. find
  15. FT Speech: Danish Parliament Speech Corpus. In Interspeech 2020, pages 442–446. find
  16. Colin Lea, Vikramjit Mitra, Aparna Joshi, Sachin Kajarekar, and Jeffrey P Bigham. 2021. SEP-28k: A dataset for stuttering event detection from podcasts with people who stutter. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 6798–6802. IEE… find
  17. Dolores Lemmenmeier. 2023. Spoken language corpora: Approaches for facilitating linguistic research. Ph.D. thesis, University of Zurich. find
  18. Huan Liao, Qinke Ni, Yuancheng Wang, Yiheng Lu, Haoyue Zhan, Pengyuan Xie, Qiang Zhang, and Zhizheng Wu. 2025. NVSpeech: An Integrated and Scalable Pipeline for Human-Like Speech Modeling with Paralinguistic Vocalizations. find
  19. Nikola Ljubešić, Danijel Koržinek, Peter Rupnik, and Ivo-Pavao Jazbec. 2022. ParlaSpeech-HR – A freely available ASR dataset for Croatian bootstrapped from the ParlaMint corpus. In Proceedings of the workshop ParlaCLARIN III within the 13th language resources and evaluation Conference, pages 111–116… find
  20. Nikola Ljubešić, Ivan Porupski, and Peter Rupnik. 2025a. Identifying Filled Pauses in Speech Across South and West Slavic Languages. In Proceedings of the 10th Workshop on Slavic Natural Language Processing (Slavic NLP 2025), pages 1–8, Vienna, Austria. Association for Computational Linguistics. find
  21. Nikola Ljubešić, Ivan Porupski, and Peter Rupnik. 2025b. Identifying Primary Stress Across Related Languages and Dialects with Transformer-based Speech Encoder Models. In Proceedings of Interspeech 2025, pages 5768–5772. find
  22. Nikola Ljubešić, Peter Rupnik, and Danijel Koržinek. 2024. The ParlaSpeech Collection of Automatically Generated Speech and Text Datasets from Parliamentary Proceedings. In International Conference on Speech and Computer, pages 137– 150. Springer. find
  23. Nikola Ljubešić, Peter Rupnik, Ivan Porupski, Nejc Robida, and Mirna Potočnjak. 2025. Dataset for primary stress identification in Croatian and related languages and dialects. Slovenian language resource repository CLARIN.SI. find
  24. Robbie Love, Claire Dembry, Andrew Hardie, Vaclav Brezina, and Tony McEnery. 2017. The Spoken BNC2014: Designing and building a spoken corpus of everyday conversations. International Journal of Corpus Linguistics, 22(3):319–344. find
  25. Yanying Mao, Qun Liu, and Yu Zhang. 2024. Sentiment analysis methods, applications, and challenges: A systematic literature review. Journal of King Saud University – Computer and Information Sciences, 36(4):102048. find
  26. Michael McAuliffe, Michael Socolof, Sarah Mihuc, Michael Wagner, and Morgan Sonderegger. 2017. Montreal forced aligner: Trainable text-speech alignment using kaldi. In Proceedings of Interspeech 2017, pages 498–502. find
  27. Michal Mochtak, Peter Rupnik, Taja Kuzman Pungeršek, and Nikola Ljubešić. 2025. ParlaSent: Mapping Sentiment in Political Discourse with Large Language Models. Political Research Exchange, 7(1). find
  28. Michal Mochtak, Peter Rupnik, and Nikola Ljubešić. 2024. The ParlaSent multilingual training dataset for sentiment identification in parliamentary proceedings. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)… find
  29. Alexis Nasr, Frederic Bechet, Benoit Favre, Thierry Bazillon, Jose Deulofeu, and Andre Valli. 2014. Automatically enriching spoken corpora with syntactic information for linguistic studies. In LREC, pages 854–858. find
  30. Elenmari Pletikos Olof, Jelena Vlašić Duić, and Jordan Bićanić. 2023. Perception of standard croatian pitch accents by speakers from pitch accent and from stress accent regiolects. In Proceedings of the 20th International Congress of Phonetic Sciences, pages 361–365. Prag: Guarant International. find
  31. Peng Qi, Yuhao Zhang, Yuhui Zhang, Jason Bolton, and Christopher D. Manning. 2020. Stanza: A Python Natural Language Processing Toolkit for Many Human Languages. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 101–108. find our card
  32. Amrit Romana, Minxue Niu, Matthew Perez, and Emily Mower Provost. 2024. FluencyBank Timestamped: An updated data set for disfluency detection and automatic intended speech recognition. Journal of Speech, Language, and Hearing Research, 67(11):4203–4215. find
  33. Rotem Rousso, Eyal Cohen, Joseph Keshet, and Eleanor Chodroff. 2024. Tradition or Innovation: A Comparison of Modern ASR Methods for Forced Alignment. In Proceedings of Interspeech 2024, pages 1525–1529. find
  34. Svetlana Savchuk. 2009. Spoken texts representation in the Russian National Corpus: Spoken and Accentologic sub-corpora. In NLP, Corpus Linguistics, Corpus Based Grammar Research. Fifth International Conference, Smolenice, Slovakia, 25-27 November 2009. Proceedings, pages 310– 320. find
  35. Björn Schuller, Stefan Steidl, Anton Batliner, Julia Hirschberg, Judee K Burgoon, Alice Baird, Aaron Elkins, Yue Zhang, Eduardo Coutinho, and Keelan Evanini. 2016. find
  36. The INTERSPEECH 2016 computational paralinguistics challenge: deception, sincerity and native language. In 17TH Annual Conference of the International Speech Communication Association (Interspeech 2016), Vols 1-5, volume 8, pages 2001–2005. ISCA. find
  37. Suwon Shon, Ankita Pasad, Felix Wu, Pablo Brusco, Yoav Artzi, Karen Livescu, and Kyu J Han. 2022. Slue: New benchmark tasks for spoken language understanding evaluation on natural speech. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 792… find
  38. Per Erik Solberg and Pablo Ortiz. 2022. The Norwegian parliamentary speech corpus. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 1003–1008, Marseille, France. European Language Resources Association. find
  39. Luka Terčon and Nikola Ljubešić. 2023. CLASSLA-Stanza: The next step for linguistic processing of south slavic languages. arXiv preprint. find
  40. Anja Virkkunen, Aku Rouhe, Nhan Phan, and Mikko Kurimo. 2023. Finnish parliament ASR corpus: Analysis, benchmarks and statistics. Language Resources and Evaluation, 57(4):1645–1670. find
  41. Alicja Wieczorkowska. 2025. Methodology for Obtaining High-Quality Speech Corpora. Applied Sciences, 15(4):1848. find
  42. Yaru Wu, Fabian Suchanek, Ioana Vasilescu, Lori Lamel, and Martine Adda-Decker. 2022. Using a knowledge base to automatically annotate speech corpora and to identify sociolinguistic variation. In 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1054–1360. European Language Res… find
  43. Lori Young and Stuart Soroka. 2012. Affective News: The Automated Coding of Sentiment in Political Texts. Political Communication, 29(2):205–231. find
  44. Vicky Zayats, Trang Tran, Richard Wright, Courtney Mansfield, and Mari Ostendorf. 2019. Disfluencies and human speech transcription errors. arXiv preprint arXiv:1904.04398. 11. — [arXiv:1904.04398](https://arxiv.org/abs/1904.04398) find
  45. Language Resource References Paul Boersma and David Weenink. 2025. Praat: Doing Phonetics by Computer. [Computer program]. Retrieved 28 August 2025. find
  46. Erjavec, Tomaž and Kopp, Matyáš and Ogrodniczuk, Maciej and Osenova, Petya et al. 2023. Multilingual comparable corpora of parliamentary debates ParlaMint 4.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1859. ↗
  47. Kopp, Matyáš and Ljubešić, Nikola. 2024. Parliamentary spoken corpus of Czech ParlaSpeech-CZ 1.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1785. ↗
  48. Koržinek, Danijel and Ljubešić, Nikola. 2024. Parliamentary spoken corpus of Polish ParlaSpeech-PL 1.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1686. ↗
  49. Ljubešić, Nikola and Koržinek, Danijel and Rupnik, Peter. 2024a. Parliamentary spoken corpus of Croatian ParlaSpeech-HR 2.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1914. ↗
  50. Ljubešić, Nikola and Rupnik, Peter and Koržinek, Danijel. 2024b. Parliamentary spoken corpus of Serbian ParlaSpeech-RS 1.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1834. ↗
  51. Ljubešić, Nikola and Rupnik, Peter and Porupski, Ivan and Kuzman Pungeršek, Taja and Koržinek, Danijel and Kopp, Matyáš. 2025. Spoken corpora of parliamentary debates ParlaSpeech 3.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1833. ↗
  52. Ljubešić, Nikola and Rupnik, Peter and van Noord, Rik. 2023. XLM-R-Parla (Revision 8c7f9b2). Hugging Face. PID 10.57967/hf/6717. — [doi:10.57967/hf/6717](https://doi.org/10.57967/hf/6717) find
  53. Mochtak, Michal and Rupnik, Peter and Meden, Katja and Ljubešić, Nikola. 2023. The multilingual sentiment dataset of parliamentary debates ParlaSent 1.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1868. ↗
  54. Rupnik, Peter and Ljubešić, Nikola and Mochtak, Michal. 2023. XLM-R-ParlaSent (Revision a04de02). Hugging Face. PID 10.57967/hf/6718. — [doi:10.57967/hf/6718](https://doi.org/10.57967/hf/6718) find
  55. Rupnik, Peter and Ljubešić, Nikola and Porupski, Ivan and Verdonik, Darinka. 2024. wav2vecbert2-filledPause (Revision 5e75061). Hugging Face. PID 10.57967/hf/6732. — [doi:10.57967/hf/6732](https://doi.org/10.57967/hf/6732) find
  56. Verdonik, Darinka and Dobrovoljc, Kaja and Rupnik, Peter and Ljubešić, Nikola and Majhenič, Simona and Čibej, Jaka and Schmidt, Thomas. 2024. Training corpus of spoken Slovenian ROG 1.0. Slovenian language resource repository CLARIN.SI. PID http://hdl.handle.net/11356/1992. ↗

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/ljubesic-2026-parlaspeech-3-0-richly-annotated/