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: 16624719566.846472
num_examples: 41349
download_size: 15670547947
dataset_size: 16624719566.846472
configs:
- config_name: default
data_files:
- split: train
path: data/train-*