metadata
dataset_info:
features:
- name: filename
dtype:
audio:
sampling_rate: 24000
- name: filesize
dtype: int64
- name: transcript
dtype: string
- name: transcript_wav2vec
dtype: string
- name: levenshtein
dtype: float64
- name: duration
dtype: float64
- name: num_words
dtype: int64
- name: client_id
dtype: int64
splits:
- name: train
num_bytes: 48041763157.94412
num_examples: 168392
- name: test
num_bytes: 952142549.9266493
num_examples: 3079
- name: dev
num_bytes: 928725728.9787879
num_examples: 3146
download_size: 37092439132
dataset_size: 49922631436.84956
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: dev
path: data/dev-*