A pianist's innate tone quality- an embarrassing myth of the past, or the innocent victim of bad science? Part i

Before going any further,  I'm going to lay my position on the table. Whether a pianist can vary tone (independently of volume) is a subject that generates such extremist fervour that if some people get any sense that I won't be stating exactly what they want to hear, they'll be off at once. If you're the sort of person who is only interested in hearing validations for your existing belief system, then by all means go elsewhere to cherry pick whatever you can find to agree with you. However, I'm actually here to disagree with any definitive position about a still uncertain issue. When someone is overly certain (either for or against absolute tone) then you should probably be suspicious. There's simply not enough evidence either way for a truly informed final stance yet.

What the science does show is that innate tone is definitely not as big an issue as some might believe (with the majority best explained by illusions of contrast). However, it also shows that it may be audible in particular conditions. It's certainly not a proven impossibility (either in theory or practice) and not enough conditions have been properly tested. After all, the fact that no search operation succeeded in finding flight MH 370 didn't prove that the plane doesn't exist. It showed that they didn't look in the right place yet. Experiments have been adequate to say that if tone exists it's not anywhere near as substantial or prevalent a factor as some have imagined. What they haven't done is provide a basis to dismiss it. In a future post I'll likely go into more detail about why the common theory (of only a single hammer speed being relevant) is unfit as a model for analysis. Recorded data has definitively proven that there's more going on. Accurate theory involves issues of flexion and vibration (which are variable based on how smoothly the hammer gets accelerated). For now, however, I'm going to concentrate on some of the problems with previous investigations and also show where I believe any future research should be directed, for the best chance of getting more meaningful data.

Relative effects proven

Firstly, lets start with what relationships can offer us. There are three main variables here- timing of key depression/release, intensity of each note and pedalling. I doubt whether anyone would wish to dispute that these are an issue. Many deny absolute tone outright, but I've never seen anyone who denies the fact that you could generate plenty via those elements alone. Today, proof of how much is possible without absolute tone is simple enough. Record yourself on a decent digital piano (which merely monitors the energy in the key strikes) and listen back. You can do a pretty good job of a great many things. If I try to play with what I consider "coarse" movements, to some degree I will hear that reflected in the sound of a digital. Styles of movement can still have corresponding effects on perceived tone, through intensities alone. However, I don't consider that quite everything is there. I feel both that awkward movements are punished far less than on a real piano, and that the most expressive playing comes to life far less than on an acoustic. See the following video:

https://youtu.be/Hway_tTFOzc?t=58

Hearing a pianist of Katsaris' calibre has a somewhat perverse effect. In a sense he is able to show off how much is possible from the instrument. In another sense, however, such a fine artist emphasises all the more how much is missing, compared to his sound on an acoustic. That film is now rather old, but I still find even the latest digitals suffer from a constant degree of neutrality. Absolute tone could indeed be the missing factor here, but that belief is not adequate to draw a scientifically worthy conclusion. Technology is advancing, but there are still plenty of other issues that could explain what is missing from the current level of simulation. We can only use it as clear evidence that a good deal of illusion is possible without innate tone quality.

The search for absolute effects

In this previous example, it was incredibly easy to isolate those variables from a possibility for absolute tone. Here, we have much a bigger problem- to isolate potential changes in tone quality from all other variables. The moment anything else is included, you risk making illusions via relationships, therefore no definitive conclusions can be drawn. In testing for tone, scientists have always reduced to a single note at a time (looking to pair notes at similar volume and see if people can h ear any other differences). The problem here is that some things will register clearly only via a build up, rather than in small quantities  Imagine a situation where someone shows you a single piano key. Let's say that key has a single mm shaved off its width. On its own you may have no idea at all that anything is amiss. However, build a whole keyboard out of the similar keys and now you have a piano keyboard that is missing 5.2 cm of width (if you're wondering- it's not 8.8 cm as I first thought, because only the number of white keys affects the total width). Still too small to notice? You could say the same about ingesting a trace quantity of poison, vs a number of trace quantities in close succession. 200 or so apple pips contain enough cyanide to kill a person (check it out, it's honestly not a myth!). A single pip wouldn't even cause a headache, however. It's another situation where effects can only be readily distinguished through accumulation.

In cases like this, the result does not even need to be more than the sum of its parts. The sum of its parts alone may register clearly where a single part does not. Just because scientific protocol can only draw conclusions from isolated notes, it does not automatically follow that isolated notes will therefore be adequate to judge an effect from. This is especially true when testing whether humans can hear a difference. Even when analysing sound waves with computers, no two strikes will ever be literally identical to start with. There's no definitive way to determine what traces of difference are meaningful on paper and which are not. If a scientist looks at a number of sound waves and says they show no evidence of tone, that does not mean they are the same waves, but merely ones that have been deemed close enough. This is an especially big concern when it comes to older research, such as Otto Ortmann's from the 1930s. With the extraordinarily primitive audio technology of the day, how could they possibly determine with any confidence what constitutes "close enough" for such narrow conclusions?

When many notes are played loudly into an open pedal, there is every reason to suspect that (should absolute tone quality be possible) the cumulative effect would be many times more distinct. That fact that we cannot segregate variables in any alternative way does not mean that failure to detect major differences provides a conclusive dismissal of tone in general. By extension, neither can we reasonably assume that when many notes are being struck, the relationships between those notes fully accounts for perceived tone quality. All we can say is that both variables could be in play and we have no meaningful way to isolate them (during a genuine musical performance) in order to assess whether there is a cumulative effect. This is not on a par with some "psychic" claiming that their powers would be stifled by testing under laboratory conditions, because the spirits of the ether would not approve. There's a vast and genuinely problematic discrepancy between those circumstances that are convenient to test and those circumstances that occur during real life music making.

Noise effects and the pedal as amplifier

In spite of this problem, I believe that current testing trends have missed a trick that could easily be added without compromising the standard of scientific controls. I have a proposal that I would encourage anyone involved in future empirical testing to pursue. Basically, I believe that the primary issue in perception of bad tone is the thud of the key into the keybed. I have heard for myself how loud this can be, after breaking a bass string on a few different pianos. Play the key with blunt force and you'll hear both the noise of the piano action (sending the hammer into space left by the missing string) and the noise of the key thudding into the keybed. The action noise was relatively superficial, creating more of a clattery sound. However, the key bed thud was deep and far from minor. In fact, on digital pianos there will tend to be a whole lot of noise. The pianist Frederic Chiu told me that when he was practising with headphones, not only could his wife hear the thuds from another room but she could feel the floor shaking in response. I've heard a lot of similar stories about people in apartment blocks unwittingly sending these substantial thuds into neighbouring properties during late night "silent" practise.

On a digital, the thud can occur simultaneously with the sound of the playing and thus compete against it. However, it cannot directly alter the musical sound at source. Back on the acoustic, there is a notable difference. However, it's specific to when having the sustain pedal depressed. When I did so, not only did the key itself thud but it also passed notable overtones into the other undamped strings. Play multiple notes in succession and you will cause a series of similar noises to start building up. To get some idea of this effect for yourself, try pressing very quickly and suddenly on the sustain pedal. It tends to cause something relatively similar. Alternatively, hold the pedal down and "knock" on the piano or even shout into it (open the lid first). You'll get an idea of how freely the undamped strings can be excited- not merely by other musical sounds but also by general noise effects.

Now, at this point I will add that I've seen many scientific papers that both referenced noise effects within sound waves and also tried to casually dismiss them on the grounds that they are not part of tone anyway. Even if we were to assume that outside noises could not interact directly with the sound coming from the piano's strings (in spite of having now proven that they both can and do), this would still be a completely inappropriate technicality upon which to exclude the relevance of this data. What human ear can knowingly distinguish between musical sounds and noises and then segregate them out into isolated components, in order to judge the music independently from the noises? Remember that it's specifically because the human ear cannot be trusted to separate out variables, that scientific protocol will only accept data based on the artificial limitation of an isolated note at a time. Yet we are going to casually assume it can make such impossible distinctions here?!!! Let's be blunt- that double standard is simply terrible scientific practise.

If you were to hear a recording of a pianist in which the start of every note triggered the corresponding sound of somebody breaking wind (with real vigour), do you imagine that you'd be able to easily separate out their musical performance in order to get a good measure of their perceived quality of sound? Yes, it's a ludicrous piece of hyperbole, but it shows that we can't simply decide that something is functionally irrelevant for the sake of convenience alone. It's for good reason that concert-goers are not encouraged to have a chat throughout a piano recital. Try telling security that it won't affect anybody's impression of the pianist's tone anyway... However, the things that we can't even consciously pick out might be more meaningful to perception than those which we know aren't part of the musical performance- and especially when those noise effects are also causing sympathetic reverberations directly within the piano itself. That cannot be considered "separate" from tone on any objectively valid level.

Searching in the right place

So, building on that premise, the most meaningful place to look for tonal variation (both via analysis of the sound waves and blind testing on humans ears) is when the pedal remains open and when notes are played loudly. While I can't say with certainty that nobody has ever tried to test under these conditions, I've never seen the pedal directly mentioned in any scientific paper. Why not? I'm scratching my head here, because nothing would make the role of that thud clearer or more detectable. Now, I already described pedal as a variable, but it can also be used as a tightly controlled constant. If you were to play two successive notes into the same open pedal, they would merge together and ruin the control. But there's no need to do that. Every time the pedal needs to start by being depressed. There then needs to be some delay, to allow any potential noise caused by the dampers to fully decay. If each note is played into these same conditions (always with a full reset to silence, before each recorded note) no controls are compromised whatsoever. All it means is that you have the ripest possible conditions in which to try to determine whether the effect of the keybed thud can be varied independently of the level of volume produced by the musical sound. To exclude the pedal from all testing is irrational.  It's the most probable means of literally amplifying a variable which both creates its own sound and has the ability to exert at least some effect on the musical sound.

Theory certainly suggests that you should be able to vary the volume with some degree of independence from the thud. In this post I have given explanations for how you can affect how you can pass acceleration to the key in ways that either increase or decrease the levels of mass and momentum which is sent into the keybed collision. Look at the diagrams early on, which show a finger either lengthening or being squashed while moving a key. When it's crushed inward, the arm will land with more energy. When it lengthens there will be a sense of  instead bouncing the mass up and out of the point of contact (with the greatest acceleration focused into the finger tip and key). A pianist also has different contact options, from a very pointed finger to a very flat finger (with larger surface area) that should logically affect the noise effect to at least some degree. One study even claimed hitting the key with an umbrella makes no difference. It makes the methods look extraordinarily dubious, if their recordings were unable to capture any difference between the noise effect of a skilled hand and that of an umbrella. A literal absence of difference does not even seem plausible within appropriately sensitive methods for capture of data.

Let's say two pianists are both playing at a similarly extreme volume, but one manages to reduce the severity of keybed noise by some 10% or so, compared to the other player. While I'd expect the difference to be most audible in thick chords, it's feasible that such a relatively moderate difference could affect the sense of sound even on an isolated note, when the pedal is used to boost overtone relationships. Experiments have traditionally either checked notes struck from above/direct contact, or compared strikes made with a "stiff" or "relaxed" arm. Technique is a whole lot broader though. Most relevant to the severity of thud would be whether the hand and arm stop abruptly into impact (thus dissipating the bulk of the energy directly into the instant of collision), or continue out from the point of contact to send momentum away. The difference is basically similar to that between stamping into the ground with a stop, vs pushing out of the ground while walking/running. Identical sounds would certainly not be expected from the different styles of contact.

Those who rule out tone generally say that "ugly" sound is just the result of playing louder than an instrument's limit. I don't find that credible. I once heard a concert where a rather slender pianist seemed to hit the piano with all her might (sending a lot of arm energy into hard impacts) without getting a notably large volume of sound out. However, the noise effect of the keys was so strong that I believe I could literally distinguish it from the direct musical sounds. The next pianist to play the piano had what seemed to be a much "purer" loud sound, without that same percussive thud. For the reasons mentioned earlier, I cannot prove that my hearing was accurate. However, alongside all of the additional variables, it's still perfectly logical to ask whether the superior sounding pianist might genuinely have achieved a better ratio of direct sound to noise effect. Have a listen to Cziffra here.

https://youtu.be/3zGx98fuR4I?t=33

Most of this is owed to contrasts within the most widely accepted variables, but I do not believe that it's an adequate explanation for that something extra which distinguishes his spiky sound from other players. If you listen carefully, there's a particularly audible thud at 36 seconds in. I suspect this may be caused by a slight stamp on the pedal. However, with all the variation in Cziffra's sound, I believe he also exploited noise effects when moving the keys. Much of the effect is down to wild dynamic contrasts, as well as variety between longer sounds and explosively short ones. But I think it's a hell of a stretch to argue that noise effects are not actively contributing to the perceived sound. I believe he contrasts technique that results in "purer" sound with less noise effect, against deliberately sharper attacks, that excite more chaotic frequencies responses into the open pedal. "Ugly" tone is applied to selective moments, but to great effect. When I listen to other performers play the same arrangement, they do not get that same "bite" to their sound. I can't say for sure whether the sound is more about key bed thuds or attacking the key from a height. Coming from above has been proven here to have an audible effect, that actually precedes the musical tone (note that this study didn't try to downplay the noise as irrelevant to perception of tone, like many other had). However, I'm convinced that the most probable explanation is some kind of noise effect that is both heard directly and through sympathetic vibrations within open strings (nb. even his short explosive sounds often involve a very brief pedal during the attack, which is instantly released after the explosion of overtones).

Coming back to an optimal experiment, notes must also be tested in many different registers of the piano and at many different levels and articulations. While I think it's probable that loud playing produces the most audible thuds, it could also be possible that loud playing might serve to conceal them more beneath the intensity of string vibrations. All angles must be pursued. It's also feasible that different registers of the piano would vary in how likely it is for the thudding sound to show up clearly. Very high notes would both carry a lower energy of their own (due to the lighter hammer and shorter strings) and also be expected to offer less direct competition between their high frequencies and the lower pitched thud of the key. There is every reason to think this would be a good place to test, but there's also plenty of reason to look at extreme bass registers. The point is that even when working on an individual note, we need to look at extremes of dynamic, extremes of register and also (above all for me) use the pedal for the purpose it was actually designed for. It's also logical to try a range of different pianos in case some offer more scope for exaggeration/reduction of thuds than others. I tend to feel that older pianos offer more scope for the kinds of edgy sounds we heard from Cziffra, so they would be a good place to test various attacks into the pedal.

Any test which failed to detect differences in tone (without having checked these kinds of options) has not only failed to disprove tone. It has also failed to contribute anything worthwhile to the pursuit of either science or music. Narrow testing no more "disproves" the possibility of tone than the timing of a school sport's day 100 m parents' race "disproves" Usain Bolt's official world record time. Not having observed a given result is not sufficient to declare it impossible. If you get a negative in narrow conditions, you need to ask what alternative conditions might show a clear result- not go out and tell the world you just proved tone to be scientifically "impossible" under all circumstances.

Evidence for tone in existing research and why belief in at least some possibility matters

I already linked the paper which showed that a preliminary noise effect was sufficient to allow many listeners to distinguish struck vs pressed notes by ear. This study has also given an indication that some listeners could distinguish differences on some notes and also shown differences in the resulting tonal balance, even without pedal. It's not a body of evidence that is adequate to start proclaiming that absolute tone is now proven to be a huge issue. They state in the study itself that it wasn't as much as the pianist expected. 

However, there is a world of difference between recognising some possibility (with more research required) and asserting a conclusive possibility of absolute zero. Moving away from the scientific stance and towards an artistic one now- even if it were truly impossible to exert any influence over absolute tone whatsoever under any circumstances, it would still be a bad idea to work based on cold pragmatism first and artistic ideals last. At face value, the piano should be most deadpan and least expressive instrument imaginable. However, when it really does sound that way then nobody is to blame other than the pianist- for playing it so poorly. Deciding that you should think merely of pressing keys faster or slower, in order to make notes merely louder or quieter, is no way to succeed in producing sophisticated illusions of something greater. In the earlier days of the study of tone, people had grown up in a culture founded upon phrasing and musical ideals. The idea that notes can only be louder or quieter was nothing but an intellectual novelty for those artists who were swayed by the research. They might have believed it away from the instrument, but you don't literally see them treating piano playing as founded upon nothing more than attempting to find the right speed to move each key at. The playing was not defined by a strict belief in tonal limitations but by artistic goals. 

Today there are some pianists who are so obsessed with working from (what they believe is) pragmatism, that they have neither learned to control absolute tone nor even to produce any worthy level of illusion. They are so caught up in a premise that is probably not even entirely true, that they have forgotten to be an artist first and a thinker second. If you want to make even an illusion, you're going to need to do a lot more than think of pecking at piano keys at different speeds, in the hope of getting a worthy phrase. I recently saw a post by a pianist who was trying to ridicule his teachers for telling him to think about his tone quality. The joke was on him, however. Whatever else you might believe, tone quality does exist. We can debate whether it can exist independently from volume, but it certainly varies in proportion with volume. If you punch into every note needlessly loud then you will indeed have an uglier sound than a pianist who shapes phrases according to musical principles. Sure enough, this pianist's sound was noisy and brutal- filled with percussion and lacking in close control. Ruling out the very important advice of a teacher (based on dubious science) had done him no favours. He was too busy calling tone impossible to realise how poorly he was controlling even relative dynamics. If you want to make a musical phrase, then you can forget ramming the arm on to every note (and then validating a bad approach by saying science has proved that the piano doesn't care how you move the keys). A musician needs to learn how to move the arm smoothly from side to side- helping to organise the movement of the fingers in better relationships, while also helping to avoid the volume of excessive downward pressures.  I don't think anyone has ever been worse off for striving for tone quality at the piano. However, at least some of those who insist that tone is impossible will be putting themselves at a genuine artistic disadvantage. 

There's something incredibly absurd about the fact that many people are willing to believe that basic musical techniques for expression are on a par with occultism, while basing their view on a religious style faith in what is very low quality and poorly tested science. At the science end, it's about time that archaic assertions were tested with better quality methods- to help prevent this movement of elevating flawed science ahead of artistic goals. Most importantly though, is how to look at things from an artistic point of view. You must at least believe that musical illusions exist- and that you can't hit the piano keys any old way if you want to make them. Especially not with an umbrella.