---
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: 591502027.0
    num_examples: 3500
  - name: validation
    num_bytes: 8608441.0
    num_examples: 50
  - name: test
    num_bytes: 31369300.0
    num_examples: 206
  download_size: 576512726
  dataset_size: 631479768.0
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](https://wenet.org.cn/opencpop/) dataset