4 ms·

The sensor wouldn't wait for the whole travel and hit of the hammer, instead it would measure hammer position/velocity at all times and replicate that, so the l

The sensor wouldn't wait for the whole travel and hit of the hammer, instead it would measure hammer position/velocity at all times and replicate that, so the latency could be fairly small, I would be curious if small enough for a musician. In the article they mentioned "[...] activated by solenoids, a proven technology even used on some Spirio Steinways.". On Spirio Steinways's website they advertise a player piano model called Spirio [0], and in the description it they mention their capabilities as "[...] measure hammer velocity (up to 1020 dynamic levels at a rate of up to 800 samples per second) [...]", a sensor at 800Hz would mean a sensing delay of 1.25ms as a starting point. To that we would need to add the mechanical delay of the solenoid, unfortunately I couldn't find a source for how fast are solenoids (the fastest I found was a solenoid hydraulic valve fully closes in 13ms, but that is the full travel, I couldn't find info on response time for first movement), so I'll make an out-of-my-ass guess that a fast solenoid starts acting in two or three ms, then round up and estimate the total delay to 4ms. So would my 4ms guess be too much for a musician? I don't know, not my fields, my guess (I've been doing that a lot today) is that if we are talking tempo it is probably too little to matter. So my only worry would be phase, if a musician's ear is sensible/trained enough to detect phase (I don't know if it is, probably isn't right?), then a 4ms delay would mean being in antiphase to the expected response at ~125Hz, that is the B2 key. [0] https://br.steinway.com/pianos/steinway/todays-steinway/spirio https://br.steinway.com/pianos/steinway/todays-steinway/spir...

jacquesm 3y ago
The Spirio only uses the solenoids during midi playback, not during 'normal' operation.
scatters 3y ago
4ms is 1.4 meters (4'6") of sound in air. That's less than half the length of a concert grand, so at most it would double the latency as experienced by the pianist's ears.
seabass-labrax 3y ago
Thank you for looking into Steinway's technology in more detail. However, there are a couple of things that don't quite feel right about it. Firstly, even if the sensing delay could be reduced to a couple of milliseconds, which I believe is perfectly plausible, I doubt whether the velocity of traditional solenoids could be controlled with enough accuracy to replicate the velocity exactly. They are on/off devices, so the only way to moderate the velocity is by rapidly changing the voltage by extremely small amounts. This I am supposing must need a feedback loop with a very complex function to account for different types of mechanical friction, electromagnetic impedance along the solenoid's coil and other factors that associate the voltage with the instantaneous velocity. Alternatively, one could use a stepper-motor style linear actuator, which would certainly be very precise and could theoretically be powerful and fast enough to emulate a traditional piano's hammer. However, I have never seen such specialised stepper motors advertised, nor one small enough to fit in an unmodified grand piano action. All in all, I don't believe that Steinway's really do have that capability, and I suspect that, whatever their measuring accuracy is, the precision of the actuator itself is more like a crude correlation to a few dozen levels of force than the claimed hundreds of levels. Of course, they do have a very slick-looking player piano, and I suspect that it what really matters for their clientèle. On to your second point, I'm pretty sure that 4ms would indeed not be detectable. By 10ms I think I would be able to hear it; maybe there are people who could hear a delay shorter than that. As for phase, you can definitely detect the relative phase of two identical tones, but since an acoustic piano has up to three different strings each with complex harmonics I'm not sure you could pick out the phase of any individual tone that easily. However, I will admit that I am tempted to try now! :)
- ZiiS 3y ago
  I expect they mean they have 10bit open loop control. Analog has lots of smoothing and things that cancel out you can get away with simple. The speekers and headphones that can reproduce 16bit 44kHz music don't have feedback.

jacquesm 3y ago

The Spirio only uses the solenoids during midi playback, not during 'normal' operation.

scatters 3y ago

4ms is 1.4 meters (4'6") of sound in air. That's less than half the length of a concert grand, so at most it would double the latency as experienced by the pianist's ears.

seabass-labrax 3y ago

Thank you for looking into Steinway's technology in more detail. However, there are a couple of things that don't quite feel right about it. Firstly, even if the sensing delay could be reduced to a couple of milliseconds, which I believe is perfectly plausible, I doubt whether the velocity of traditional solenoids could be controlled with enough accuracy to replicate the velocity exactly. They are on/off devices, so the only way to moderate the velocity is by rapidly changing the voltage by extremely small amounts. This I am supposing must need a feedback loop with a very complex function to account for different types of mechanical friction, electromagnetic impedance along the solenoid's coil and other factors that associate the voltage with the instantaneous velocity. Alternatively, one could use a stepper-motor style linear actuator, which would certainly be very precise and could theoretically be powerful and fast enough to emulate a traditional piano's hammer. However, I have never seen such specialised stepper motors advertised, nor one small enough to fit in an unmodified grand piano action. All in all, I don't believe that Steinway's really do have that capability, and I suspect that, whatever their measuring accuracy is, the precision of the actuator itself is more like a crude correlation to a few dozen levels of force than the claimed hundreds of levels. Of course, they do have a very slick-looking player piano, and I suspect that it what really matters for their clientèle. On to your second point, I'm pretty sure that 4ms would indeed not be detectable. By 10ms I think I would be able to hear it; maybe there are people who could hear a delay shorter than that. As for phase, you can definitely detect the relative phase of two identical tones, but since an acoustic piano has up to three different strings each with complex harmonics I'm not sure you could pick out the phase of any individual tone that easily. However, I will admit that I am tempted to try now! :)

ZiiS 3y ago
I expect they mean they have 10bit open loop control. Analog has lots of smoothing and things that cancel out you can get away with simple. The speekers and headphones that can reproduce 16bit 44kHz music don't have feedback.