300 speech samples provided in noisy and 22 different enhanced versions, totaling 6,900 audio clips with human-labeled mean opinion scores (MOS). This collection was developed for the 2024 URGENT Speech Enhancement Challenge at NeurIPS to evaluate speech quality assessment and enhancement algorithms.
Use Cases
- Train speech quality assessment (SQA) models to predict human MOS values from audio signals
- Benchmark the performance of speech enhancement (SE) systems against 22 distinct algorithm outputs
- Analyze the variance in human-labeled MOS across different noise types and enhancement methods
Strengths
- Includes 300 original noisy speech samples and 6,600 enhanced variations
- Features human-labeled Mean Opinion Scores (MOS) for subjective quality evaluation
- Covers 22 different speech enhancement (SE) system outputs for every input sample
- Developed as part of the official NeurIPS 2024 URGENT Speech Enhancement Challenge