Kensho dataset #23
Description
Please add Kensho dataset to your list. It consists of 5,000 hours of training data for the ASR task. The conversations are entirely real, deliberately transcribed, and anonymized. The conversations are spontaneous and differ from crowdsourced datasets (like Common Voice) in this regard.