Rapid Python Tristimulus Centroid Audio Analysis

I have a folder of 12 short-duration sound samples and I need two metrics extracted from every file: the three tristimulus values and the spectral centroid. All data must be processed in Python, ideally with Librosa at the core; you are free to lean on NumPy, SciPy or Pandas for supporting math and I/O.

Please send back:

• A clean, well-commented .py script (or a Jupyter notebook) that loads an arbitrary batch of WAV/AIFF files and outputs a table containing filename, tristimulus 1-2-3, and centroid.
• A CSV (or similar) generated by running that script on my dataset.
• A short README describing assumptions (e.g., window size, hop length, any pre-emphasis) so I can reproduce results or tweak parameters.

I will share the audio samples as soon as we start; the total runtime is under ten minutes, so processing time is negligible. Accuracy matters more than fancy plots, but if you would like to include a quick visual sanity check—say, a centroid trend plot—that is welcome.

Deliver everything within the next few days and keep the code portable (Python 3.9+, Librosa 0.10 preferred). Let me know if any clarifications are needed before you dive in.