metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transkription
dtype: string
- name: source
dtype: string
- name: predicted_transkription
dtype: string
splits:
- name: train
num_bytes: 20603875516.34
num_examples: 546561
download_size: 21827534712
dataset_size: 20603875516.34
configs:
- config_name: default
data_files:
- split: train
path: data/train-*