metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: transcription
dtype: string
- name: input_features
sequence:
sequence: float32
- name: input_length
dtype: float64
- name: labels
sequence: int64
- name: wer
dtype: float64
- name: cer
dtype: float64
- name: pred_text
dtype: string
splits:
- name: train
num_bytes: 1529326259.6699386
num_examples: 1466
- name: test
num_bytes: 170041050.7
num_examples: 163
download_size: 330387293
dataset_size: 1699367310.3699386
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*