metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transcription
dtype: string
- name: audio_len
dtype: float64
- name: transcript_len
dtype: int64
- name: len_ratio
dtype: float64
- name: num_feature_vecs
dtype: int64
splits:
- name: train
num_bytes: 11412772812.720085
num_examples: 31954
- name: val
num_bytes: 863584013.7317481
num_examples: 2553
download_size: 11764004926
dataset_size: 12276356826.451834
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*