metadata
dataset_info:
features:
- name: audio
dtype: audio
- name: caption
dtype: string
- name: transcript
dtype: string
- name: pred_caption
dtype: string
splits:
- name: train
num_bytes: 741962594
num_examples: 5000
download_size: 735685799
dataset_size: 741962594
configs:
- config_name: default
data_files:
- split: train
path: data/train-*