ESC-50: 2,000 Environmental Audio Recordings for Sound Classification

ESC-50 is a labeled collection of 2000 environmental audio recordings designed for benchmarking sound classification methods. The dataset consists of 5-second-long recordings organized into 50 semantical classes, with 40 examples per class. It was uploaded to Hugging Face by ashutoshm28 and was last updated on 2026-03 07.

Use Cases

• Train audio classification models based on the 50 labeled environmental sound classes.
• Benchmark machine learning methods for environmental sound recognition based on the standardized 5-second recordings.
• Develop sound event detection systems based on the dataset's 5 major categorical groupings.
• Conduct research in acoustic scene analysis based on the collection of 2000 diverse recordings.

Strengths

• Contains 2000 individual audio recordings, providing a substantial corpus for model training.
• Features 50 distinct semantic classes, offering a diverse range of environmental sounds.
• Provides a standardized structure with 40 examples per class and 5-second-long recordings.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-03-07 02:00:49; freshness should be verified.

Provenance

Source

Hugging Face user ashutoshm28.

Collection Method

Likely a curated collection of environmental audio recordings.

Time Range

null

Freshness

Last updated 2026-03-07 02:00:49.

Geography

null