Danish Legislative Speech Corpus

This dataset contains text and speaker data for around 1.7 million snippets of legislative speech from Denmark's Parliament, Folketinget. The data set draws in part on the ParlSpeech V2 data set, in part on Folketinget's publicly available XML transcripts. In order to homogenize the lengths of text units, longer speeches are broken down into snippets of 3-5 sentences each. Version 2 of the data extends the data coverage to late 2024. While the first version is retained here for archival purposes, most users will probably want the newest version (i.e. "paradf-v2").