metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: audio_id
dtype: string
- name: condition_on_prev
sequence: int64
- name: whisper_transcript
dtype: string
splits:
- name: train
num_bytes: 574075446
num_examples: 745
download_size: 438438594
dataset_size: 574075446
configs:
- config_name: default
data_files:
- split: train
path: data/train-*