Full description of timbre vector dimensions

[vjosap]

Hi,

I am analysing and extracting the timbre of tracks based on Spotify's low-level feature analysis. In the documentation, it is stated as follows:

"The timbre feature is a vector that includes 12 unbounded values roughly centered around 0. Those values are high level abstractions of the spectral surface, ordered by degree of importance. For completeness however, the first dimension represents the average loudness of the segment; second emphasizes brightness; third is more closely correlated to the flatness of a sound; fourth to sounds with a stronger attack; etc"

So in the documentation, we have only explanations for the 4 first dimensions, but what about the rest of them? Does anyone know a good article, paper or blog post that explains all of the 12 dimensions?

Thank you.