metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: start_time
dtype: float32
- name: end_time
dtype: float32
- name: input_features
sequence:
sequence: float32
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 17954827537.56
num_examples: 21680
- name: validation
num_bytes: 4410540089.225
num_examples: 5325
download_size: 20504188270
dataset_size: 22365367626.785004
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*