ok the examples in the help files do have both sides of the distance matrix so I felt I had to do it… as you can see it is now working. I have ircam.descriptors~ and @a.harker’s cohabitating now to get more accurate spectral descriptions, until we have the fluid.descriptors (and/or updated ah ones) and I’m back to some normalisation I want to do on the full range (because Kurtosis is not very usefull our of the +/-3 (or 0-6) range but that is not playing ball with the 0-130 range of pitch and amp…