TahaNauman/SonicEDA ? reverse-engineered prompt

Build me a Python project that takes an MP3 file and turns it into a simple audio analysis report. I want it to load the song, plot the waveform, compare sample rates, measure RMS energy and zero crossing rate, generate a spectrogram, and extract common audio features like tempo, spectral centroid, rolloff, bandwidth, MFCCs, and chroma. Then combine those values into a clean DataFrame, run some basic exploratory analysis, and save the charts into a plots folder.

Please make each stage easy to run on its own, and keep the shared settings in one place so the sample rate, frame size, and file path are consistent. I’d also like a plain language glossary that explains the terms without jargon. If anything is unclear, look up current librosa and pandas docs online so the code is up to date. The end result should feel educational, with clear outputs and simple visualizations rather than anything machine learning focused.