anushkakhebade99-spec/spotify-audio-clustering ? reverse-engineered prompt

Build me a Streamlit app for Spotify track clustering that lets me explore how songs group together based on their audio features, not genres.

I want it to load a trained K Means model, a standard scaler, and a PCA transformer, then show an interactive dashboard where I can see the clusters in 2D, compare their typical audio profiles with radar charts, and try my own custom track values to get assigned to the nearest cluster in real time.

Please make it feel polished and easy to use, with clear labels and simple explanations of what the clusters mean, like high energy pop or acoustic folk. If the model files or data already exist, wire everything up to use them, and if anything is missing, recreate the pipeline cleanly from the dataset. Look up current Streamlit and scikit learn docs online if you need to.