metadata
dataset_info:
features:
- name: duration
dtype: float64
- name: speakers
dtype: string
- name: source
dtype: string
- name: gender
dtype: string
- name: keep
dtype: bool
- name: duplicate
dtype: bool
- name: reasons
list: string
- name: clipped
dtype: bool
- name: silence_ratio
dtype: float64
- name: cps
dtype: float64
- name: word_count
dtype: int64
- name: detected_lang
dtype: string
- name: id
dtype: string
- name: text
dtype: string
- name: audio
dtype:
audio:
decode: false
- name: source_shard
dtype: string
- name: source_key
dtype: string
splits:
- name: train
num_bytes: 128875508
num_examples: 2500
download_size: 128577232
dataset_size: 128875508
configs:
- config_name: default
data_files:
- split: train
path: data/train-*