taras-sereda/uk-pods

uk-pods - speech datasets of Ukrainian podcasts. Preparation Clone the dataset repository and extract the content of clips.tar.gz archive. git clone https://huggingface.co/datasets/taras-sereda/uk-pods cd uk-pods && tar -zxvf clips.tar.gz To use these manifests for training/inference with NeMo [1] modify audio_filepath to absolute locations of audio files extracted in previous step. # data_root=<clonned_repo_dir> # /home/ubuntu/uk-pods data_root=$(realpath .)… See the full description on the dataset page: https://huggingface.co/datasets/taras-sereda/uk-pods.