~94 hours / 72,245 utterances of Turkish TV and talk-program speech (16 kHz
mono WAV) with systematically repaired transcripts. This is a derivative of
ulaspolat/tsc-tr-filtered-94h,
itself a filtered subset of the ISSAI Turkish Speech Corpus
(MIT license). Audio is unchanged; only the text column was modified.
