Why spectrum profile values are relative?

Maybe you could use the specialized software I wrote to automate this task instead of Audacity:
http://www.forum-pianoteq.com/viewtopic.php?id=1634

You seem to be aware of it.

With the automatic method, the whole keyboard can be modified by copy/paste in about 30 minutes, and then each note further adjusted with the scaling function in Note Edit if needed.

Also you have to take into account that the spectrum is dynamic, changing all the time, so what can be done is only specifying relative changes to the overtones as set in the model. There is no fixed value in dB.

EDIT: Rereading your post, I see you use Standard version...so that explains your use of Audacity. My method is for the Pro version in Note Edit. You can't really do comparisons using only the 8 sliders in the GUI. A modified value there is applied to all notes, not only the one you analyze.

I understand the cost problem, especially if it's the only thing you need it for and want to do it only once...

PtqSpecProf works in conjunction with the Pro version and generates for all harmonics of each note the difference string to be pasted from the clipboard into the Note Edit pane, and, at the end the fxp is saved, but the Pro version is needed all along to do this.

I have posted in the past some examples in the FXP Corner such as this one:

http://www.forum-pianoteq.com/uploads.php?id=1642

While it's not a perfect emulation (of course! that's impossible) comparing the fxp with the original Bluethner preset shows the timbre has been drawn more toward the sound of the Bösendorfer samples I started with.

If you use a dynamic spectrum analyser like this one: http://www.voxengo.com/product/span/ you will see that the relative amplitude of each overtone varies a lot with time and differently for each note! This is the complexity of the piano sound, and the only thing that my software can attain is a snaphot at a fixed velocity.

Comparing compatible fixed spectra works partially since the model makes them vary with time in its own way, but it does change the timbre in a meaningful way. The pasted spectra differences vary wildly from note to note. This was a surprise to me when I first saw the result...But consecutive notes vary smoothly in timbre despite that.

It's possible to do interesting global changes with only the 8 sliders in the GUI (Rachel Jimenez produced some great D4 presets this way) but it's mostly done by ear and in conjunction with other parameters. If you want to try using Audacity for comparison with a favorite note from your goal piano it's only 8 substractions to do by hand! Not such a big deal. But don't expect too much this way...

I'm of the opinion that whatever sound reinforcement combination you are using will greatly affect the final sound.  Most speakes seem to have dead spots. Obviously two 10" speakers would be different from a pair of 15".