This is a whisper-formatted version of the ivrit.ai Knesset Plenums dataset.
This dataset was created by splitting long audio recordings, along with their respective transcriptions, into audio slices of 30 seconds or less.
Each such slice represents one or more consecutive segments, along with timestamp token data and the previous slice's transcription.
The code for this dataset preparation process is available on the… See the full description on the dataset page:
https://huggingface.co/datasets/ivrit-ai/knesset-plenums-whisper-training.
