An open acoustic features benchmark for industrial machine fault diagnosis

This dataset is derived from the MIMII (Malfunctioning Industrial Machine Investigation and Inspection) dataset and provides a standardized benchmark of acoustic features for industrial machine fault diagnosis. The original audio recordings were preprocessed and converted into structured CSV files containing extracted acoustic features to facilitate machine learning, anomaly detection, and predictive maintenance research. The benchmark includes data from four machine types: Fan Pump Slide Rail Valve Each machine type contains recordings under three signal-to-noise ratio (SNR) conditions: −6 dB 0 dB 6 dB For each operating condition, feature files are provided for multiple machine IDs and preprocessing variants (Raw and LPF).