From what ive gathered (doing interactive light/sound stuff and chatting with folks at CymaSpace) the issue is that its subjective. Synaesthetes are often frustrated with color to pitch mapping that claims it is "inherent" because the root color (this author picked blue because "it seemed settled") is the creators personal synaesthetic correlation. The issue gets complex when you consider what training set makes each synaesthete build the correlations between colors, moods, sounds, etc.
The article its-self is rad in that the circle of fifths is very much a "relative" thing to begin with so we are actually seeing some pretty "flavorful" results- but the search for the "true color of sound" is tough when light is 430-730 trillion hz and audio is 20-20 thousand hz. The factor that you then are dividing or transposing by is arbitrary.