Ithaca I think people who say “physics says” aren’t modeling the mechanics accurately enough. An idealized model will likely fail to explain/predict some real-world behaviors.

I am in agreement with you here. Almost all models are approximation. We like to start simple, and think that it's not necessary to include things that don't affect the result meaningfully or perceptibly. But a critical part of scientific training is as soon as models (theory) and reality (experimental results) deviate, more rigorous investigation is warranted, and minds/models/theory should be ready to be changed.

Good luck obtaining grants for this kind of research on piano sound, though! 🙄

      iternabe I am in agreement with you here. Almost all models are approximation. We like to start simple, and think that it's not necessary to include things that don't affect the result meaningfully or perceptibly. But a critical part of scientific training is as soon as models (theory) and reality (experimental results) deviate, more rigorous investigation is warranted, and minds/models/theory should be ready to be changed.

      Yeah, OK, but if you want to propose a different model you have to at least present an argument of why the simple model fails and a tentative explanation. I'm not getting what exactly it is about impulse that changes the way the hammer hits the string. Some people have proposed that the elasticity of the hammer stores some angular momentum but I'm very sceptical that this effect is non-negligible compared to the total momentum of the hammer. I'm too lazy to do actual calculations but I'm open to arguments (with math) if you think you can convince me otherwise.

        iternabe Good luck obtaining grants for this kind of research on piano sound, though! 🙄

        That might be a good challenge for the science Youtubers. I don't know if you watched it but a few years ago there was a wager between Steve Mould and ElectroBOOM where each attempted to give an explanation of the Mould effect when dropping a chain of beads from a height. There were a lot of experiments and back and forth arguments and other Youtubers like Veritassium also put in their 2 cents. It was very entertaining to watch. Maybe someone is up for a piano tone challenge. 😄

          A lot of comments about the beautiful, rich and expansive tonal color palette of acoustic piano are making me longing for one! Weather, tuning and regulation, and just the analog and infinite grades of motion of the piano action is just fascinating. Though, some of these factors are affecting the tonal quality over time, or after mechanical adjustment.

          I was contemplating more along the line of what a pianist can do with their interaction with the keys at a particular moment to affect the tonal quality.

          In my opinion it’s more about controlling the hammer better up to the point of no return. Since that’s all we can really do. I found an interesting university study that seems to touch on this discussion. It’s lengthy and I just started reading it:

          https://public.websites.umich.edu/~brentg/Publications/Thesis/Chapter2.pdf

            danno858 that's interesting. It seems to explain @iternabe's observation with human perception:

            By carefully governing the timing overlap of notes as they follow one another, the musician can evoke a certain percept in listeners which will not be labeled in terms of timing at all; it will be labeled by the ear and auditory perceptual centers in terms of the frequency domain, that is, timbre. Although it is not, in physical terms, independent control of timbre and intensity, it is perceived (and labeled) as such.

            I was going to suggest it might be an auditory illusion. That article seems to go to similar direction. But it also suggest that the pianist have some control over the perception. I will read the rest of it now 🙂

            Human brain is complicated. I suspect brains of different individuals process the same auditory input differently. Not only that, the input data is different too. Because our "hardware" is different. We all have different physical ears. Some people even have different left & right ear canals/structures.

              iternabe careful now. You will start hearing weirdness in pianoteq which will ruin it for you 🙂

                  hebele iternabe careful now. You will start hearing weirdness in pianoteq which will ruin it for you 🙂

                  Hey, maybe that will be the day I run out of excuse not to buy a 7-foot grand! 😜

                    danno858 I found an interesting university study that seems to touch on this discussion. It’s lengthy and I just started reading it:

                    https://public.websites.umich.edu/brentg/Publications/Thesis/Chapter2.pdf

                    @danno858 I cannot thank you enough for find this chapter, as it is truly enlightening. I have yet to read the second half of the chapter on synthesizer. This paragraph has already made me want to get the entire book.

                      More thoughts of the moment as I am reading the chapter posted by @danno858:

                      The author pointed out a major difference between synthesizer/digital piano’s method of determining key velocity vs acoustic piano. In acoustic piano, the velocity that matters is at the instant the hammer escape from the action. In digital piano, the velocity is the average velocity of the key passing through the distance between two sensors. Needless to day, the same gesture (i.e. accelerating the key from stationary to motion at the let-off point) can produce very different velocity value between acoustic action and digital action. Moreover, effect of gestures that produces non-linear acceleration of the key/action can get lost by DP actions simplified averaging algorithm.

                      This also makes wonder about the effectiveness of the best hybrid actions, like Kawai’s NV10S. If its sensors is designed to detected the velocity of the traveling hammer before it strike the string, then it is a much better approximation of acoustic action in that the velocity detected is almost the instant velocity of the hammer escaping the action. I still say almost because if the velocity sensing mechanism employs a pair of sensors placed apart, then it is still an average speed over distance, which might still be a bit off from the final string striking velocity if the hammer does not maintain constant speed due to friction or gravity.

                      I found another published study on this topic:

                      Once again: The perception of piano touch and tone.
                      Can touch audibly change piano sound independently of intensity?

                      The study concludes: Our results suggest that only some musicians are able to distinguish between a struck and a pressed touch using the touch noises as cue, especially the finger–key noise that characterises a struck attack, whereas others could not tell any difference. Without those touch noises none of them could tell a difference anymore. When they could not hear the touch differences, they tend to rate louder tones as being struck, and soft tones as being pressed.

                      Ithaca Randomly pulling stuff out of my hat (sorry, I'm tired - 3 nights of very short sleep) - think about two mechanisms which can make contact at a single point, each of which has a squishy balloon-like or somewhat elastic section somewhere inside. (Say, the finger pad and compressible joints in one mechanism, and a soft hammer + leather accoutrements in the other.) Those soft/elastic sections absorb and then dissipate energy over time, and probably in a non-linear way. So if you whack the first mechanism at the far end in one way, the way that force over time gets transmitted through the first mechanism and then through the second mechanism to produce a hammer taking vertical flight isn't so obvious. If you've modeled systems with multiple springs with different K constants, that might give you a better sense of what you're looking at. But even then, typically when people look at these systems, they're not looking at the time factor, unless they're studying something with a constant-period oscillation, so again, in steady state.

                      It's certainly interesting to look at the mechanics of your hand and finger but that's not what the contention is about.

                      There is a cut off point where the escapement mechanism throws the hammer free like a projectile. It doesn't matter how the key action was moved before the escapement point. After this point the hammer is in free fall not touching anything until it hits the string. The only thing that can physically affect the sound at this point is the momentum of the hammer. Unless you can show that it's possible for the action (and all that comes before the escapement) to impart some angular momentum such that the hammer hits the string at a different angle that is controllable by the player then there isn't any physical way to produce a sound that has a different quality at the same volume. If the piano had a mechanism where there is a physical connection between the player and the strings at the point of sound production, like in the harpsichord or clavichord, then I would believe you. But the piano has this air gap which disconects the cause from the effect.

                        Ithaca I propose that what the musician hears as control over timbre is effected by careful control over
                        timing. For a particular tone, the pianist is indeed able to select only two parameters beyond pitch:
                        the hammer strike speed and the hammer strike time (although the time of tone damping is also
                        important). I propose that by selecting these two parameters, a pianist can produce a percept in
                        his or her listener of timbre being controlled independently of intensity

                        That doesn't make sense. How does the player control the strike time? The travel distance of the hammer is constant so the time it takes for the hammer to travel depends on the initial speed. How do you get two independent variables?

                            BartK my understanding is authors’s proposal is not about controlling the timing during the hammer free travel, but just the absolute timing of hammer strikes the string. And this control is performed by the pianist’s chosen gesture.

                            Since the author did not elaborate further, my own thoughts is the pianist is choosing a gesture not necessarily with the goal to affect timing - rather the choice is aimed to affect timber (independent of loudness). However, variations in gestures started at the same time may inadvertently reach the letoff point at different time, the result is variation in timing. Then, if variation in timing can be psychoacousticly perceived as variation in timber, then both pianist and audience can be lead to believe that timber and loudness was independently controlled.

                              Ithaca Although the author did not specifically say, I do believe it is variation of timing in relation to other notes (before, after, or simultaneous).

                              I still do not believe a note played in isolation can have its timber controlled independent of its loudness. The other paper I posted today provides experimental proof.

                                Ithaca I'm saying that (1) I think finger velocity alone (measured at the relevant location/point in time) is insufficient to completely determine the final hammer velocity

                                Yes, I can agree with that. I'm just saying in the end the result is control over the individual note volume and that is linked to timbre not independent of it.

                                Ithaca (2) varying finger "gestures" on a key that seem similar in volume/force can affect the timbre of the note, because the gesture can affect how long the note is allowed to sound before it's damped. In my own piano, with the wonky D note, it's very clear: the extra metallic/nasal/brassy/blaring resonance has a much faster decay than the primary resonance of the note at certain volumes, so at those volumes, when you sound the D, the timbre of the note changes over time, if you allow it to ring for long enough. At the worst (seemingly) constant volume, letting the note sound for about a full second makes most of that annoying higher-order resonance go away, whereas a maybe 250 ms press will give you maximum obnoxiousness. For this one note, it's like you've combined a piano with an out-of-breath kazoo player, but for the little bit when the kazoo is going full force, the timbre is decidedly unpleasant.

                                I see what you're getting at but I don't think this gives you as much control as you might think. The time until releasing the note is basically what pianists call articulation and depending on how fast the music is going you don't have that much choice in when to release a note. Maybe you have a bit more leeway in slow music.

                                    In this discussion it may sound as though I don't believe you can control piano timbre at all and that I'm against any notions of doing extra motions to affect it but that is not true.

                                    The evocative descriptions that pianists use to describe music such as "shimmering like the wind" or "horn call" are useful to help you imagine the sound you want, which with enough experience a pianist can express by making the appropriate movements. The process is mostly a psychological one where the inner ear and imagination combine to control the playing aparatus in a subconscious manner. At a more basic level using specific movements like flatter fingers or arm dampening gives you more control over the nuances and makes it easier to achieve the sound. I'm not against descriptions like that if they help achieve the sound you want.

                                    What I object to is the "play it loud but not harsh" kind of descriptions, as if the two could be separated. A more realistic description is "play it as loud as it needs to be but not louder".

                                    I think the reason some pianists think like that is psycho-accoustic. From their training they are used to moving a certain way to produce loud chords because it gives more control and is more natural. When they play in unnatural ways, such as rigid fingers and no arm dampening, the sound comes out much louder (and thus harsher) than it would using the same force, so they have the impression of playing equally loud but with a harsher tone. But I think that if you recorded the spectrum of the sound and the speed of the hammers you would in fact find that the volume is different. It's all an illusion as with everything on the piano.

                                      BartK In this discussion it may sound as though I don't believe you can control piano timbre at all and that I'm against any notions of doing extra motions to affect it but that is not true.

                                      Ditto. In fact, this whole discussion, together with the scientific research articles, makes me believe even more in the pianists' ability to create the perception of varying timber while maintaining the loudness. I also see more clearly the importance of find the gesture(s) that allow one to create the tone quality desired.

                                      BartK What I object to is the "play it loud but not harsh" kind of descriptions, as if the two could be separated. A more realistic description is "play it as loud as it needs to be but not louder".

                                      I really like this example. But even here, my new understanding is that direction of "play it loud but not harsh" is in fact referring to what the player and ultimately the audience should perceive. Therefore a student may actually benefit by understanding such description is not to be taken as a strict and simplistic direction in the magnitude of physical execution.

                                          I looked a bit more into Brent Gillespie. He is a professor of mechanical engineering. The chapter on piano was part of his PhD thesis on how to create a digital synthesizer (or digital piano) that is truly faithful to acoustic piano not only in the capability to produce whole palette of sound but also give the pianist the exact same control and feedback. His analysis of how pianist make the sounds from the instrument is a mechanical engineer's view. To me, that is not only interesting but also rigorous, probably even more so than you can get from many of the YT pop scientists.

                                          One of his concepts is to reduce the mechanism of playing the piano to "mapping". In beginner piano learner's term, I understand this as "if I do this move, I will get that sound". He further breaks down the mapping to two stages: the mapping from hammer's escape velocity and timing to the final sound (this is one-to-one); and the mapping from the player's gesture (movement of all involved body parts over a short time and distance) to the hammer's escape velocity and timing (this is many-to-one).

                                          This model of thinking excites me with some new insights:

                                          First, if many gestures can map to one hammer escape velocity & timing (thus the same tone), then one can claim a gesture makes this particular tone but cannot claim that particular gesture is the only way. It may very well be that certain gesture works better for the anatomy of the majority. But when it doesn't work for a few, there exists other gestures to be explored for the same effect. Furthermore, for one desired tone, with a number of gestures to choose from, the decision can be made on other criteria, such as which will give the player more control, more consistency, and less stress.

                                          Second, how we learn to manipulate the piano is in fact how we learn the mapping. It can be done empirically, by trying and listening as many variations as we can. This is why it takes time and there is no shortcut. Then, a good teacher is the shortcut by leading the student to try first the gestures that's most effective, and not wasting time on those gestures that's not (or even harmful). But then, can a pianist help themselves by having a better ability or sense on what new variation in gesture to try that has a better chance to work? Here, I think having the correct understanding of underlying mechanical and physical model can give one the edge.