Spoken corpora of parliamentary debates ParlaSpeech 4.0 (transcriptions)

Name: Spoken corpora of parliamentary debates ParlaSpeech 4.0 (transcriptions)
License: https://creativecommons.org/licenses/by-sa/4.0/

Ljubešić, Nikola; Koržinek, Danijel; Porupski, Ivan; Dobranić, Filip; Rupnik, Peter

Spoken corpora of parliamentary debates ParlaSpeech 4.0 (transcriptions)

CLARIN.SI data & tools

Authors: Ljubešić, Nikola ; Koržinek, Danijel ; Porupski, Ivan ; Dobranić, Filip and Rupnik, Peter

Item identifier: http://hdl.handle.net/11356/1885

Project URL: https://clarinsi.github.io/parlaspeech/

Demo URL: https://clarinsi.github.io/parlaspeech/concordancer/concordancer-guide.html

Referenced by: https://aclanthology.org/2026.lrec-1.447.pdf

Date issued: 2026-10-22

Type: corpus, text

Size: 77327430 words, 3901578 sentences

Language(s): Croatian , Czech , Polish , Serbian , Slovenian

Description: The ParlaSpeech corpora are built from the transcripts of parliamentary proceedings of Croatian, Serbian, Slovenian, Polish, and Czech parliaments available in the ParlaMint 4.0 corpus (http://hdl.handle.net/11356/1859), and the parliamentary recordings available from the parliaments' YouTube channels. An instance in this dataset is a transcript sentence with the corresponding metadata and the aligned audio. This entry contains the transcriptions and various levels of enrichment, while the audio files in FLAC format are available at http://hdl.handle.net/11356/1889. This version's main extensions are the addition of Slovenian language, and the addition of silent pauses as an enrichment layer, and CAP topic labels as speech-level metadata. Current enrichment layers are the following: * ParlaSpeech-FilledPause: automatic annotations of filled pauses ("eerm") * ParlaSpeech-SilentPause: automatic annotations of silent pauses * ParlaSpeech-Align: precise word- and grapheme-level alignment (HR, RS only) * ParlaSpeech-Stress: Labelled primary stress in multisyllabic words (HR, RS only) * ParlaSpeech-Ling: Universal Dependencies (UD) formatted linguistic annotations (lemma, part-of-speech, syntax, etc.) * ParlaSpeech-Senti: sentiment estimation based on the transcript * ParlaSpeech-Topic: CAP topic label based on the transcript Data size per parliament is the following: * Croatia (HR): 923k sentences, 3,110 hours, 324k filled pauses, 972k silent pauses, 11M word stresses * Czechia (CZ): 718k sentences, 1,218 hours, 200k filled pauses, 683k silent pauses, no word stresses * Poland (PL): 535k sentences, 1,009 hours, 200k filled pauses, 180k silent pauses, no word stresses * Serbia (RS): 291k sentences, 896 hours, 74k filled pauses, 308k silent pauses, 2M word stresses * Slovenia (SI): 1,435k sentences, 4,080 hours, 626k filled pauses, 2,171k silent pauses, no word stresses The data are available in the following formats: * JSONL: master format, containing all the data. Distributed as a newline delimited JSON, where each line is a valid JSON serialization. Mostly intended for computerized processing. * VERT: vertical format intended for concordancers with text, links to audio, linguistic annotations, sentiment, filled pauses, and primary word stress (where available). * TextGrid (HR and RS only), available through the old repository entry http://hdl.handle.net/11356/1833: word- and grapheme alignment, primary word stress, and filled pauses. This format's intended use is with the Praat software (https://www.fon.hum.uva.nl/praat/) for research and applications in phonetics and other speech-focused disciplines. For a detailed dataset schema description and examples, please see our dedicated website: https://clarinsi.github.io/parlaspeech/.