philgzl/mls-hq-urgent-track1

Multilingual LibriSpeech HQ (MLS-HQ)

This is a mirror of the Multilingual LibriSpeech HQ (MLS-HQ) data used in URGENT 2025 Track 1. The original files were converted from FLAC to Opus to reduce the size and accelerate streaming.

• Sampling rate: 48 kHz (resampled from 44.1 kHz to support Opus format)
• Channels: 1
• Format: Opus
• Splits:
  • spanish: 150 hours, 36031 utterances
  • german: 150 hours, 35890 utterances
  • french: 150 hours, 36078 utterances
• License: CC0 1.0
• Source: https://huggingface.co/datasets/kohei0209/mls_hq_urgent_track1
• Paper: Interspeech 2025 URGENT Speech Enhancement Challenge

Usage

import io

import soundfile as sf
from datasets import Features, Value, load_dataset

for item in load_dataset(
    "philgzl/mls-hq-urgent-track1",
    split="spanish",
    streaming=True,
    features=Features({"audio": Value("binary"), "name": Value("string")}),
):
    print(item["name"])
    buffer = io.BytesIO(item["audio"])
    x, fs = sf.read(buffer)
    # do stuff...

Citation

@inproceedings{saijo2025interspeech,
  title = {Interspeech 2025 {URGENT} {Speech} {Enhancement} {Challenge}},
  author = {Kohei Saijo and Wangyou Zhang and Samuele Cornell and Robin Scheibler and Chenda Li and Zhaoheng Ni and Anurag Kumar and Marvin Sach and Yihui Fu and Wei Wang and Tim Fingscheidt and Shinji Watanabe},
  booktitle = {Proc. Interspeech},
  pages = {858--862},
  year = {2025}
}