I've acquired a copy of the paper referenced in the article: Motor origins of timbre in piano performance by Kuromiya et al. There's a lot of statistical information, but many questions remain unanswered. I'll concentrate on the second part of the study, which is introduced thus:
A professional pianist participated in a performance experiment designed to produce two distinct types
of single-tone keystrokes. The two keystrokes were characterized by a marked difference in the acceleration at the escapement point of the key, while the other features—including key velocity at escapement, mute velocity and acceleration, duration, and both onset and offset noise —were kept approximately constant. Specifically, the pianist was instructed to manipulate only the acceleration at the escapement point while minimizing changes in the other variables in a single-note keystroke.
I have bolded the last phrase of that paragraph, because I have never encountered a pianist who could knowingly influence only the acceleration at the escapement point. I would like to have more information on how the pianist achieved this result.
Other points of interest:
The study took into account the noises when the finger hits the key, when the key reaches the keybed and when the note stops. The relationship between these noises and the main sound of the note as transmitted by the soundboard is dependent on the placement of the listener, or in the case of this study, of the microphone(s). The only information given on the recording process is that the piano was a Shigeru SK2, recorded in a soundproof room with a Neumann U87 Ai microphone. It is not stated where this microphone was placed.
The pianist played a single note many times, apparently for a duration of about 400 ms (less that 1/2 a second). I couldn't find out if this was always the same note, or if notes in different ranges were tested. The scientists identified keystrokes that differed primarily in terms of the acceleration of the key at the moment of escapement
(more than 12%), whereas the other factors, in particular the velocity at escapement, had much smaller differences. The listeners (16 professional pianists) were able, to a degree better than that of chance, to class the lower acceleration tones more often as "lighter" or "clearer" than the higher acceleration ones.
The magnitude of this perceived difference was not discussed at all. Since the listeners were not all in agreement - on average, they were just more likely to class the low acceleration tones as clearer - I imagine that the effect is very subtle. I doubt that it is particularly relevant for practical purposes, as I am sure that other factors that we pianists strive to control - dynamic shaping of a phrase, dynamic balance, pedalling, articulation and rubato - have much more influence on the music we communicate to our listeners.