Selected recordings from the MIMII Dataset

This dataset contains a curated subset of audio recordings derived from the original MIMII (Malfunctioning Industrial Machine Investigation and Inspection) dataset. It was compiled specifically for the purpose of training and evaluating machine condition classification models as part of a Master's thesis. Dataset Overview: Target Machines: Pumps, fans, sliders, and valves. Acoustic Conditions: All selected audio files feature a Signal-to-Noise Ratio (SNR) of 0 dB. Data Distribution: The subset is perfectly balanced, containing exactly 450 recordings of the "normal" state and 450 recordings of the "abnormal" state for each of the four machine types. Original Source & Acknowledgment: This is a derived dataset created strictly for educational and research purposes. All original audio data belongs to the creators of the MIMII dataset. Anyone utilizing this subset should refer to and cite the original MIMII dataset publication (Purohit et al., 2019).https://zenodo.org/records/3384388