besacier/ALFFA

African Language Ressources ALFFA ===== African Language Ressources : alffa.imag.fr SWAHILI directory ### # Swahili Data from Hadrien Gelas PhD data collection (2009-XXXX) # # Description of folders and data # # (PhD supervised by L. Besacier (LIG) & F. Pellegrino (DDL) ### ### /AUDIO_AND_TRN/ Contains 6 set of approximately 2hours of audio (eg. /1set/SWH-15-20101104.zip) + Corresponding transcriptions (eg. /1set/SWH-15-20101104.txt) + Categorized Reference transcription per audio quality of set 6, used as a test set in experiment (filename: PROPER-01-REF-RFI-6SET.trn) ### /DICT/ # /ORIGINAL-65k-SWH.dic Original Swahili dictionary based on 65k most frequent words and basic g2p rules # /LAST-SWH-65K.dic Swahili dictionary after a few xperiment adding prons for i) english, ii) accronyms ### /PHONE/ # /SWH.phone Phone list used in .dic ### /TEXT/ # /00-LM_SWH-CORPUS.txt 28 M Words Full text data grabbed from online newspaper and cleaned as much as it could All files below are extracted from this file # /01-CLN4-TRN.txt Training data for LM # 02-CLN4-DEV.txt Dev data for LM # 03-CLN4-TST.txt Testing data for LM