metadata
dataset_info:
features:
- name: audio
dtype: audio
- name: id
dtype: string
- name: text
dtype: string
- name: singer
dtype: string
- name: token_wavlm_large_1024_6
dtype: string
- name: token_espnet_hubert_1024_9
dtype: string
splits:
- name: train
num_bytes: 591502027
num_examples: 3500
- name: validation
num_bytes: 8608441
num_examples: 50
- name: test
num_bytes: 31369300
num_examples: 206
download_size: 576512726
dataset_size: 631479768
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
Discrete units extracted from HuBERT and WavLM for the Opencpop dataset