Speech + text dataset collection based on the ParlaMint data. Paper describing the construction process: https://www.arxiv.org/abs/2409.15397.