instinct-org/

espeech_podcasts_chunked_speech_restorised_nfa_aligned

Source dataset: instinct-org/espeech_podcasts_chunked_speech_restorised Aligned dataset: instinct-org/espeech_podcasts_chunked_speech_restorised_nfa_aligned Rows: 2467471 successfully aligned rows This dataset adds NeMo Forced Aligner metadata for STT training and timestamp quality control. Source rows that did not produce usable CTM alignments are excluded from the published data. Added columns:

nfa_token_alignments: NeMo token/subword CTM spans… See the full description on the dataset page: https://huggingface.co/datasets/instinct-org/espeech_podcasts_chunked_speech_restorised_nfa_aligned.