metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: input_values
sequence: float32
- name: input_length
dtype: int64
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 15853252317.123964
num_examples: 39529
- name: val
num_bytes: 2016239876.125
num_examples: 5631
download_size: 15873864990
dataset_size: 17869492193.248962
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*