zrstr Thought I'd chime in on that paper and dynamics compression with most player systems. I've been interested in modern reproducing piano systems for decades, and really got into it when I started doing consulting work on the electronics systems for the Bosendorfer SE pianos in 1995.
Dynamics compression on both the high and low end is typical for player systems- as an engineering problem the softest notes are very, very difficult to reproduce with reliability without extremely expensive components and complicated design, and most people don't need or want to have a full concert-level 'fortissimo' sound in their living room. Full-power notes are also difficult to reproduce without extremely expensive components (powerful high-voltage solenoids), so almost all systems compromise there.
I happen to own both types of pianos that are described in the referenced paper- bought the Disklavier C1 MarkIIXG 24 years ago and it's been a fine instrument, but had the opportunity arise to buy one of the very rare original Bosendorfer SE pianos (a 225) early last year and did so. It's an utterly amazing instrument even at 37 years old and sounds wonderful!
To be fair to Yamaha, the newer Disklaviers, especially of the Pro variety are FAR more accurate in reproduction than the one I have.
The SE system was the product of an exceptionally exacting engineer (Wayne Stahnke). His goal was to make a reproducing piano that would EXACTLY reproduce any human performance, in a way that would satisfy the original artist that it was their own work, almost no matter the cost to make it. Mr. Stahnke was successful in that goal and his SE systems continue to be renowned and prized for that ability even 40+ years later.
Wayne Stahnke much later (2000's) engineered the Live Performance LX system to retrofit into existing pianos at a relatively consumer-level price point. It was based on much of the underlying SE technology, but with a number of compromises to meet the price target. One very large difference between the SE and LX is that the SE uses a custom 'closed-loop' solenoid drive system so it can much more exactly control the intensity and speed of the solenoids that actuate the keys and pedals at the time they are playing. As a result, even the quietest pianissimo notes can be reproduced accurately. The LX system was 'open-loop', meaning that more generalized parameters were used that approximate the desired intensity- still sounds great, but not quite as exact.
The non-pro Disklavier systems (such as mine) were also open-loop in addition to being lower-resolution. The open-loop systems (LX & Disklavier, most others) have to actuate the softest notes at a slightly higher level in order to guarantee that they actually sound, causing some dynamic range compression on the soft end. (Softest notes are played louder than the original performance). Smaller, less-powerful (and less-expensive) solenoids are used in these systems as well since very few customers want a full-power fortissimo, so that causes dynamic range compression at the high end too. (Loudest notes played softer than the original performance).
So now we come to Spirio. The original Spirio design was exactly the Stahnke LX system (he sold it to Steinway), with the same capabilities. Since then, it has been redesigned by a third-party engineering firm employed by Steinway and recording capabilities were added, making the Spirio R system. I don't know if the current Spirio systems use a closed-loop playback design, but the original was not.
The current Spirio systems retain the high-resolution recording pioneered by Wayne Stahnke for the SE systems, with 10 bits of precision (1020 levels) for hammer velocity and 256 levels of proportional pedal positions.
Because of the history and reasons above, I would expect that there is likely at least some dynamic range compression in the current Spirio design.
I should also mention that current Disklavier Enspire Pro models are similar in resolution and capability to the Spirio, so I'm not sure that either one is 'better' than the other for reproducing ability. I'm sure there could be arguments made either way.
Hopefully some of you are interested in this…
