metadata
dataset_info:
features:
- name: key
dtype: int64
- name: audio
dtype: string
- name: transcript
dtype: string
- name: audio_paths
dtype:
audio:
sampling_rate: 16000
- name: input_features
sequence:
sequence: float32
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 4178080929.5
num_examples: 3500
download_size: 1514342399
dataset_size: 4178080929.5
configs:
- config_name: default
data_files:
- split: train
path: data/train-*