DAS-10

This dataset contains demodulated distributed acoustic sensing (DAS) signals acquired using a laboratory-built time-gated digital optical frequency domain reflectometry (TGD-OFDR) system. The measurements were collected by replaying environmental audio samples through a loudspeaker to generate vibration-induced strain along the sensing fiber. The dataset consists of 2,000 labeled samples distributed across 10 classes: rain thunderstorm train background (ambient noise) chainsaw door_wood_knock drilling glass_breaking hand_saw jackhammer Each sample has a duration of 2 s and contains 10 spatial sensing channels. The released data are demodulated DAS vibration signals sampled at an effective rate of 50 kHz. This dataset is intended for research on DAS signal processing, vibration perception, and learning-based event classification under distributed fiber-optic sensing scenarios.