Ok, I finally got my hands on a copy of that paper in full.The contents are less clear than the abstract suggests. The main conclusion seems to be that for some reason they could not find out which parameters do make a person decide if a voice if female or male. They tested various parameters like resonance/formants, pitch and pitch shift. some of the XX women had very low pitch and still were perceived as female, they had very high frequency resonances/formants incomparion to all the MtFs. None of the MtFs could produce such high formants and XX women who failed to do so were perceived as male as well. However within the group pf MtFs, the ones with higher pitch and more pitch variation were on average perceived more likely as female than the others, even though they exceeded the XX women in both parameters to achieve that. It seems to me that increasing pitch and pitch variation can compensate for the lack in high frequency resonance which was not attainable by the MtFs. They did some resonance change and managed to elevate formants a bit, but still were far from the average female range with all the techniques they used.
My personal conclusion would be that resonance formants actually is more important in determining the gender of the voice (contrary to the suggestions in the abstract) but that there is no good way to reach that as all the attempts by MtFs to reach a higher formants were not bringing it up in the normal female range. There was one MtF with a VFS in the study, she reached a rather high pitch and formants but still was perceived as male for unknown reasons (maybe her voice sounding artificial). But when formants cannot be increased, pitch seems to actually be the most important secondary and more easily changeable parameter that can compensate for that, along with pitch variation. So a combination of changing pitch, resonance and other parametrs is needed and in that context if pitch fails, the other parameters cannot compensate - unlike in XX women who can compensate for a low Fo by other parameters, most likely resonance which is a parameter determined by biology (length of the vocal tract, changeably only to a small degree by control).
In respect to VFS I would say this article cautions a bit, as the VFS subject in it was not perceived as female despite hitting all the parametrs needed - but also the need for elevated pitch is demonstrated to compensate for the ability to increase formants beyond a certain range and thus I would say a well done VFS which does not make the voice sound restricted or artificial would be beneficial.
It would be interesting to see what would come from it if they looked as a Dr Thomas surgery which should increase formants by changing the length of the vocal tract...
