This is the download page for the PATCOR corpus, as referenced in the paper Van de Loo et al. (2019) Effective weakly supervised semantic frame induction using expression sharing in hierarchical hidden Markov models.
Due to the size of the corpus (8.7G unpacked), we offer it as a direct download from: https://www.clips.uantwerpen.be/patcor.zip
The unzipped file contains 10 directories, one for each participant in the experiment, labeled pp1 to pp10. Each pp-folder has the same structure, as illustrated for pp1:
pp1_transcriptions.txt
all orthographic transcriptions for pp1110802144557_transcription.txt
orthographic transcriptions for session 110802144557110802144557_phon_transcr.txt
all phonetic transcriptions for session 110802144557110802144557.pjt
a subdirectory for session110802144557
1.wav
complete audio recording channel 12.wav
complete audio recording channel 2move_1_110802143844_audio.wav
audio recording for move_1move_1_110802143844_framedescription.xml
frame description for move_1move_1_110802143844_transcription.txt
orthographic transcription for move_1move_39_110802145753_phon_transcr.txt
phonetic transcription for move_1