metadata
dataset_info:
features:
- name: text
dtype: string
- name: description
dtype: string
- name: url
dtype: string
- name: short
dtype: int64
- name: audio_tokens
sequence: int64
- name: md5
dtype: string
- name: start_us
sequence: float64
- name: end_us
sequence: float64
- name: words
sequence: string
splits:
- name: train
num_bytes: 43164739
num_examples: 867
download_size: 8645119
dataset_size: 43164739
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
This dataset contains no audio file. The audio_tokens column is generated using WavTokenizer, to be used by OuteTTS 500M v0.2 model.
If you want to re-use this data to train for TTS model, download the full dataset containing WAV file from https://huggingface.co/datasets/ngxson/ThisTTS_dataset