I've worked on one or two guitars with 'sleeves' in the soundholes, and also ran some experiments with a removable 'tornavoz' in one of my 'test mule' guitars. The tornavoz was of cardboard, but patterned after the one illustrated by Romanillos in his book on Torres.
In all cases, the effect of the tornavoz/sleeve has been of the sort that theory would predict: as you make the soundhole into a longer effective tube, the pitch of the 'main air' resonance drops, and it becomes less powerful. This shows up quite clearly on a spectrum chart. With the full depth tornavoz the 'main air' resonance practically dissappears, and the spectrum can closely resemble that of a small 'Baroque' style of guitar with a parchment rose. In the latter case the drag of all those edges cuts the power of the main air mode 'way down, although the pitch of what remains is high.
Playing the 'test mule' with the tornavoz in it confirmed what the spectrum suggested: it has the timbre of a small, early style of guitar, but since it's bigger it's a bit more powerful. Like the small bodied guitar it tends to 'project' well, but the lowest bass notes lack fundamental. This may not come across in all cases as a lack of power in the low end, but simply as a difference in timbre.
The folks making guitars equiped with tornavoz' say that there are certain design differences that need to be incorporated to get the best results. Obviously, my 'test mule'; did not have those alterations made, and that will no doubt be cited as the reason for the 'failure' of the experiment Perhaps that's so. However, I'll note that, for all their complexity, these are physical objects that must obey physical laws. I saw the sorts of changes that one could predict from an understanding of the physics of the thing, and would expect to see the same sorts of changes no matter what the starting point was. Wth a different starting point the outcme might be 'better', but I can't see any way to avoid the lowering of the 'main air' pitch, and the loss of power.
I will also note that most of Torres' guitars that were originally equiped with tornavoz' have had them removed. They are still useful instruments without the sleeve in the soundhole. This suggests that a similar guitar with a usable sound should change in similar ways with the insertion of a tornavoz. It can't be magic.
The other objection will be that I did not make my tornavoz from copper or brass, as Torres did. It was impossible on this guitar to modify the soundhole in such a way as to allow for the insertion of such a rigid device of the 'correct' size. I hope at some point to make another test guitar that will allow for this. Meanwhile, I'll stand on the experimental results I have, and remain a skeptic.