Sign in to view source links and access this dataset
Description
A curated collection of 28,564 non-speech vocal burst audio samples. The dataset spans 18 categories, including laughter, crying, cough, and sigh, with sample counts provided per category. It was created by author 0x3 and last updated on Hugging Face in May 2026.
Use Cases
Train audio classification models based on the 18 distinct vocal burst categories.
Develop emotion recognition systems based on non-speech vocalizations like laughter, crying, and sighs.
Create data augmentation pipelines for audio datasets based on the collection of breath, cough, and sneeze sounds.
Benchmark machine learning models for sound event detection based on the provided category labels and sample counts.
Strengths
Contains 28,564 audio samples, providing a substantial collection for model training.
Explicitly lists sample counts for 18 distinct categories, with over 4,000 samples for categories like cough and laughter.
Audio format is specified as FLAC, which is a lossless compression format.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Description metadata is limited; actual data quality, recording conditions, and demographic information require manual inspection after download.
Provenance
Source
huggingface
Collection Method
Curated collection; specific gathering method is not detailed in the provided description.
Freshness
Last updated 2026-05-31 08:57:42; freshness should be verified.
License information is unknown and should be verified before use.