Experimental audio dataset for squeak and rattle identification

This dataset is a collection of squeak and rattle audio files experimentally created in RMIT NVH laboratory. This is a useful database potentially be used for development of features for Machine Learning models. There are two independent datasets for squeak and rattle. The rattle audio samples were organised into 8 classes based on the beam length, impact material and the decay duration of the excitation signal. The squeak sound samples were categorised to 12 classes based on the sliding beam length, contact material type, and the oscillation frequency.