Dataset · HumynLabs
Noisy audio with medical symptoms
HumynLabs/medical-symptoms-english-audio
English audio recordings with a symptom phrase, intentionally noisy to test model robustness.
What's inside
An English audio dataset where participants speak text about chronic fatigue, headaches, dizziness, and sleep problems. Per the card, the data is deliberately low quality: noisy, non-ideal, and suboptimal examples. It is designed for testing model robustness on degraded input.
What you can do with it
- Testing audio-classification robustness to noise
- Evaluating ASR on degraded recordings
- Test augmentation for medical audio models
- Tasks
- Audio classification
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 242
- Likes
- 470
How to load it
from datasets import load_dataset
ds = load_dataset("HumynLabs/medical-symptoms-english-audio")