36,589 audio clips of human non-speech vocalizations categorized into 6 distinct classes including laughter, sigh, cough, sneeze, snore, and throat clearing. The data includes recordings from 3,265 unique subjects with associated metadata for demographic analysis.
Use Cases
- Train an audio classification model to distinguish between 'cough' and 'sneeze' labels
- Perform demographic analysis of vocal patterns using the 'age' and 'gender' metadata columns
- Develop health monitoring systems by detecting 'snore' or 'throat clearing' events in audio streams
Strengths
- 36,589 audio recordings across 6 vocal sound categories
- Metadata for 3,265 subjects including age and gender labels
- Baseline code provided for classification tasks using the ICASSP2022 benchmark