Open Source

Code, models, and datasets from our research

Projects

Datasets

urgent2024-sqa

79

Dataset Description This dataset comprises noisy and enhanced speech from the URGENT Speech Enhancement Challenge (NeurIPS 2024), curated for SQA/SE research. Each entry includes audio/IDs with a comprehensive suite of objective and model-predicted quality metrics, and human MOS where available (see Data fields). The MOS label for each speech sample were collected from 8 distinct human subjects through Amazon Mechanical Turk (MTurk) platform, following the P.808 recommendation.… See the full description on the dataset page: https://huggingface.co/datasets/urgent-challenge/urgent2024-sqa.

speech

Edit