My hearing is not pitch perfect but I have yet to hear an analog device (turntable or cassette) that can faithfully sustain a piano note. I always hear a variation in speed and often can hear it on CD recordings made from an analog source. Personally I hear digital reproducing the decay of piano notes better too.
Sustaining a piano tone is perhaps the hardest thing for the analog. I have not heard a cassette recorder that can do it perfectly. There is one that might be capable of pulling it off - Eumig FL-1000
https://www.hifiengine.com/manual_library/eumig/fl-1000.shtml - but, I have newver even seen one in flesh.
Turntables - and RECORDS, in particular - are a very hard nut to crack regarding speed stability. Not impossible, but so far not available in a single commercially available device having all the required technical features. Separately, they USED TO be available - in several different products, none of which has been in production for several decades.
Here, the least of the problem is the turntable drive system itself; all three major drive principles have managed to produce designs capable of wow and flutter levels well below audibility decades ago.
The biggest offender in speed stability in turntables is the fundamental resonance due to phono cartridge stylus suspension interacting with the combined mass of the cartridge and effective mass of the tonearm. This should be, ideally, placed around 10 Hz - and damped best possible . By far the best paper on this is
https://www.theanalogdept.com/images/spp6_pics/TT_Design/MechanicalResonances.pdf There have been tonearm models available in the past catering to this problem really well; the only two designs remaining in current production are the latest incarnations of Dynavector and Well Tempered tonearms.
The hardest nut to crack are the records themselves. To be precise - their eccentricity. Even in the specifications for the physical properties of the pressed phonograph record, there is too high tolerance for as basic spec as centre hole diameter - even IF we assume the record to have perfectly centered hole. The diameter of the spindle of the turntable is also too loosely specified - and those tolerances are usually further exceeded in practice, either to manufacturing tolerances or deliberate decision on the part of turntable manufacturer. There is no arguing that decision to make the center spindle on the turntable of a bit smaller diameter than spec IS a viable decision, arrived at with practical use of real world records and turntables; with centre hole on records being towards the lower (or below.... ) specified diameter, it is hard/inconvinient/impossible to place the record on the turntable with nominally specified spindle diameter. There are manufacturewrs who deliberately chose to make centre spindle towards the larger upper tolerance of the spindle dimension spec; in order to have record as tightly centered as possible, with minumal or no free play between the spindle and record centre hole. It allows for always having the record centered on the platter always THE SAME - but centering still being at the mercy of the precision of the centering of the record itself. There are cases in real life one will be forced to enlarge the hole of the record just to be able to place it on the turntable with deliberately "fat" spindle.
The only practical solution to the problem of record eccentricity came from Nakamichi - their record centering turntables:
https://positive-feedback.com/Issue33/tx1000.htm They have been discontinued ages ago and today they are, for all practical purposes, unobtainium. If they ever turn up for sale, usually they are in not fully functional condition; the lack of spare parts and indespensable alignment kits will stop all but the absolutely most determined.
Turntables have, unfortunately, been understood for what they actually ARE, by hard to believe extremely small percentage of the manufacturers. Historically, the first turntable design that went into the right direction has been the original Oracle in the late 70s. Only recently there is a descirnible trend for the turntables to focus on their real issues ( once wow and flutter have been sucessfully brought below audible levels ) - vibration control. Even one of the biggest offenders from the past ( certainly not alone, but numeriacally BY FAR the biggest - in history ) giulty of poor vibration control, Technics, has finally realized "something" has to be done. Just look at the vibration damping characteristics of their best model from the past (SP-10MK3) vs the new SP-1200GR and SP-10R :
https://www.technics.com/us/products/reference-class/direct-drive-turntable-sp-10R.html If the turntable rings for more than about 0.1 second with anything above approx -60dB, it can not reproduce the decay properly - either piano, or anything else. Turntables successfully dealing with this have been available for decades; trouble is, they tend to be, by default and by nature, rather expensive.
Where does all of that lead ? Towards learning the lessons from past mistakes and applying all of that into a single design with all the required features - including some I have deliberately left out above. Can be done ... - but requires lots of $, first to be done at all, second to be done at "reasonable" price.
Just to be perfectly clear; I do not consider CD to be ever capable of doing it. Although the amount of sound produced by the piano above 20 kHz is extremely small, for the sense of immediacy and that of the acoustics of the venue in which the piano is being played/recorded will always require frequency response extended at least to 40 kHz. Getting that with an analog turntable is doable - albeit, as shown above, at a not inconsiderable cost.
Infinitely less expensive way is high resolution digital. Anything from 96 kHz/24 bit PCM and up - but preferably DSD.