Story perspectives

New Datasets Enhance Sound Classification for Environmental Research

10/31/2025

27 6

Summary

Story summary

Researchers developed DataSEC and DataSED datasets containing real environmental recordings to improve machine learning classification of sounds such as birds, vehicles, and machinery.
Classification relies on manual labeling and subdivision into subclasses to ensure precise identification of sound sources.
The datasets reflect outdoor noise variability and aim to support comprehensive environmental noise evaluation.
Recordings are standardized to a 44.1 kHz sampling rate to ensure realism and consistency.

1 of 1

Story summary

Researchers developed DataSEC and DataSED datasets containing real environmental recordings to improve machine learning classification of sounds such as birds, vehicles, and machinery.
Classification relies on manual labeling and subdivision into subclasses to ensure precise identification of sound sources.
The datasets reflect outdoor noise variability and aim to support comprehensive environmental noise evaluation.
Recordings are standardized to a 44.1 kHz sampling rate to ensure realism and consistency.