Samuelsantos777/psg-audio-v3-unofficial-mirror

pretty_name: Psg-audio-v3 ---

PSG-Audio v3 — Unofficial Complete Mirror

Unofficial complete mirror of the publicly released PSG-Audio Version 3 dataset. This repository preserves the original files without modification and provides a reliable, high-speed mirror through the Hugging Face Hub for the research community.

Overview

PSG-Audio v3 is one of the largest publicly available multimodal sleep datasets, combining overnight clinical polysomnography (PSG) with synchronized environmental audio recordings.

The dataset was created for sleep medicine and obstructive sleep apnea (OSA) research, providing synchronized physiological signals, environmental audio, and expert clinical annotations.

This repository redistributes the original public release without modification, preserving the original directory structure and filenames.

Key Features

Repository Structure

PSG-Audio-v3/
│
├── APNEA_EDF/
│   ├── Original EDF polysomnography recordings
│   └── *.edf
│
├── APNEA_RML/
│   ├── Original expert annotations
│   └── *.rml
│
├── APNEA_RML_clean/
│   ├── Cleaned clinical annotations
│   └── *.rml
│
└── SHA256SUMS.txt

Included Contents

✅ Original EDF recordings

✅ Original RML annotations

✅ Cleaned annotation files

✅ Original filenames

✅ Original directory structure

✅ SHA-256 integrity manifest

Dataset Statistics

Typical Applications

• —Sleep staging
• —Sleep apnea detection
• —Respiratory event detection
• —Audio-based sleep analysis
• —EEG signal processing
• —Biomedical signal processing
• —Multimodal machine learning
• —Foundation models for sleep medicine
• —Computational neuroscience

Usage

Reading EDF files (Python)

import pyedflib

edf = pyedflib.EdfReader("00001001_1.edf")

channel_labels = edf.getSignalLabels()

mic_idx = next(
    i for i, label in enumerate(channel_labels)
    if "mic" in label.lower()
)

audio = edf.readSignal(mic_idx)

sample_rate = edf.getSampleFrequency(mic_idx)

print(sample_rate)  # 48000 Hz

edf.close()

Reading annotations (Python)

from lxml import etree

tree = etree.parse("00001001.rml")

events = tree.xpath("//ScoredEvent[Type='Stages|Stages']")

for event in events:
    start = float(event.find("Start").text)
    duration = float(event.find("Duration").text)
    stage = event.find("EventConcept").text

    print(start, duration, stage)

Mirror Integrity

This repository redistributes the original public dataset without modification.

• —No files modified
• —No preprocessing applied
• —No renaming
• —Original directory hierarchy preserved
• —SHA-256 checksums included for integrity verification

Researchers are encouraged to verify downloaded files using the provided checksum manifest.

Why this Mirror?

Large biomedical datasets are frequently distributed through institutional FTP servers that may be slow, geographically restricted, or unavailable.

Publishing PSG-Audio on the Hugging Face Hub provides:

• —Faster downloads
• —Resumable transfers
• —Global CDN distribution
• —Integration with Python and Hugging Face libraries
• —Easier reproducibility
• —Long-term public accessibility

This repository does not replace the original distribution source.

Original Dataset

Korompili, G. et al.

PSG-Audio (Version 3)

Science Data Bank

Dataset DOI

https://doi.org/10.11922/sciencedb.00345

Associated publication

Korompili, G. et al.

PSG-Audio, a scored polysomnography dataset with simultaneous audio recordings for sleep apnea studies.

Scientific Data, Volume 8, Article 197 (2021)

https://doi.org/10.1038/s41597-021-00977-w

Citation

If you use this dataset in scientific work, please cite the original dataset and publication rather than this mirror.

@article{korompili2021psgaudio,
  title={PSG-Audio, a scored polysomnography dataset with simultaneous audio recordings for sleep apnea studies},
  author={Korompili, Georgia and others},
  journal={Scientific Data},
  volume={8},
  pages={197},
  year={2021},
  doi={10.1038/s41597-021-00977-w}
}

Original publication

https://doi.org/10.1038/s41597-021-00977-w

Original dataset

https://doi.org/10.11922/sciencedb.00345

License

According to the original dataset providers, PSG-Audio Version 3 is distributed under the:

Creative Commons Attribution 4.0 International (CC BY 4.0)

This repository redistributes the original files under the same license declaration.

No ownership is claimed over any part of the dataset.

All scientific credit remains exclusively with the original authors.

Disclaimer

This repository is an unofficial mirror created solely to improve accessibility, reproducibility, and long-term preservation of the publicly released dataset.

It is not affiliated with, endorsed by, sponsored by, or maintained by the original authors, Science Data Bank, or any associated institution.

Acknowledgements

Special thanks to the original PSG-Audio authors for releasing this valuable dataset and enabling reproducible research in:

• —Sleep medicine
• —Biomedical engineering
• —Artificial intelligence
• —Computational neuroscience
• —Biomedical signal processing
• —Audio analysis
• —Multimodal machine learning

Their contribution continues to benefit the global research community.