sarulab-speech/yodas2_sidon

YODAS2-Sidon

Overview

This dataset is a cleansed version of YODAS-2 with Sidon speech restoration mode for Speech Synthesis and Spoken Language Modeling.

YODAS-2 is a massive, multilingual YouTube-derived dataset. We have applied the Sidon restoration model to remove background noise and enhance audio quality, making it suitable for high-quality generation tasks. We resampled original sidon output to 24kHz due to a storage constraints.

The dataset is provided in WebDataset format for efficient large-scale training.

• Source: YODAS-2 (YouTube-Oriented Dataset for Audio-Visual Speech)
• Format: WebDataset (.tar.gz shards)
• License: CC-BY-3.0

---

Dataset Structure

Each sample in the dataset contains:

• flac — audio file (24 kHz, single channel, restored)
• metadata.json (optional) — metadata including language, YouTube video ID, and transcription

Example (inside a .tar shard):

000001.flac
000001.metadata.json
000002.flac
000002.metadata.json
...

---

How to Use

With 🤗 Datasets

You can load the WebDataset directly with Hugging Face’s datasets library:

from datasets import load_dataset
from huggingface_hub import list_repo_files

repo_id = "sarulab-speech/yodas2_sidon"
subset="en000"
all_files = list_repo_files(repo_id, repo_type="dataset")

urls = [
    f"https://huggingface.co/datasets/{repo_id}/resolve/main/{f}"
    for f in sorted(all_files)
    if f.endswith(".tar.gz") and f.startswith(subset)
]

print(f"Found {len(urls)} shards.")

dataset = load_dataset(
    "webdataset", 
    data_files={"train": urls}, 
    streaming=True
)['train']
from IPython.display import Audio
sample = next(iter(dataset))
audio = sample['flac']
print(sample['metadata.json'])
Audio(audio['array'], rate=audio['sampling_rate'])

Replace subset with the desired subset.

---

Citation

If you use this dataset, please cite Sidon and the original YODAS paper:

@misc{nakata2025sidonfastrobustopensource,
      title={Sidon: Fast and Robust Open-Source Multilingual Speech Restoration for Large-scale Dataset Cleansing}, 
      author={Wataru Nakata and Yuki Saito and Yota Ueda and Hiroshi Saruwatari},
      year={2025},
      eprint={2509.17052},
      archivePrefix={arXiv},
      primaryClass={cs.SD},
      url={[https://arxiv.org/abs/2509.17052](https://arxiv.org/abs/2509.17052)}, 
}

@inproceedings{li2023yodas,
  title={Yodas: Youtube-Oriented Dataset for Audio and Speech},
  author={Li, Xinjian and Takamichi, Shinnosuke and Saeki, Takaaki and Chen, William and Shiota, Sayaka and Watanabe, Shinji},
  booktitle={2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)},
  pages={1--8},
  year={2023},
  organization={IEEE}
}

---

License

This dataset is released under CC-BY-3.0.

---

Acknowledgements

• Original data: YODAS2