metadata
dataset_info:
features:
- name: audio
dtype:
audio:
sampling_rate: 16000
- name: text
dtype: string
- name: speaker_id
dtype: int64
- name: language
dtype: string
- name: speaker_embedding
list: float32
length: 256
- name: cond_prompt_speech_tokens
list: int64
length: 375
splits:
- name: train
num_bytes: 65036678332.24
num_examples: 237352
download_size: 63941792647
dataset_size: 65036678332.24
configs:
- config_name: default
data_files:
- split: train
path: data/train-*