---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: fuzzy_tokens
    dtype:
      array2_d:
        shape:
        - null
        - 1024
        dtype: float32
  - name: token_length
    dtype: int32
  splits:
  - name: train
    num_bytes: 943664568
    num_examples: 1195
  download_size: 924190129
  dataset_size: 943664568
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

this is very much a clone of

https://huggingface.co/datasets/MrDragonFox/Elise

but converted to audio embedding vectors at 25 Hz using 

https://huggingface.co/bosonai/higgs-audio-v2-tokenizer