Tone production: Doing the right things for the right reasons
As a sophomore in college, I performed in a master class given by a former Van Cliburn Competition medalist. At one point, I was asked to play certain chords so that my fingers moved toward the fallboard as they depressed the keys, and this was supposed to change the timbre of these loud chords without actually changing their volume (providing a "richer" sound). It took all of my willpower to quietly follow this advice and not bring up the fact that the piano escapement mechanism makes the basis for the advice completely fallacious. I didn't want to be disrespectful, so I followed the advice.
This "pushing in" technique caused the hammers to strike the strings more slowly, as I predicted it would, producing a softer sound, giving the master of the class (and the audience) a most glorious false positive. I felt violated, because I had just been used as a tool to advance an illusory belief system I did not share. Asking me to simply play softer would have been equally as effective and a lot simpler. I've received instruction like this in more than a few lessons and master classes, and I have encountered many teachers and pianists who subscribe to various misguided beliefs about what can change the tone of a given note at the piano. This has led to my interest in researching the topic.
The heart and soul of misguided ideas about piano tone is a belief that one can play a single isolated note twice, each time at the exact same dynamic level, but with a different timbre by virtue of key-stroke style. This is impossible, because the hammer coasts freely during the last eighth of an inch on its way to the string. Without this mechanism of "letting go" (referred to as escapement, let-off or set-off), hammers would not be free to rebound naturally away from strings, creating a nightmare of having to manually rebound hammers away from strings with our fingers. Hammers are only in contact with strings for one to four milliseconds (click here to see Figure 1). Even when we play a short staccato note at a forte level, the key touches the bottom of the keybed for a whopping forty milliseconds (click here to see Figure 2, "key bottom" timing). For softer notes, the contact time is even longer. We do not possess the physical capability to manually rebound hammers away from strings quickly enough to excite the wide range of frequencies and overtones that are generated by freely-coasting hammers. Even if we did and the piano were built to allow for manual rebound, we would then have no way to control articulation, since all keystrokes would have to be staccato.
Escapement prevents the hammer from conveying certain information to the string, such as how much weight/tension we use in depressing the key or how quickly the hammer accelerates to its set velocity. In the end, the only pianist-controlled factor that is transferred to the string is hammer velocity.
Because of this escapement mechanism, piano timbre corresponds directly with the hammer's velocity at the point of escapement. The pianist is therefore powerless to change the tone quality of a single note without also changing its dynamic volume. What this implies is that the techniques we teach to achieve certain tone qualities cannot be considered "necessary" in order to produce certain sounds. At best, various techniques are most likely to produce these sounds, because there is always more than one way to depress a key so that the desired hammer velocity is achieved in the end. We can ask a student to stroke the key like a paintbrush, but we could also ask the student to play the note a little bit softer.
If this is all true, then what makes one pianist sound so radically different from another when comparing performances of the same piece of music? It is tonal wash, which I define to be the overall musical effect that is created by manipulating tonal properties of a series or combination of musical tones. These tonal properties are:
Pitch (which string is struck),
Rhythm (when the string is struck),
Volume (velocity of the hammer when it strikes the string, also affected by the soft pedal on upright pianos), and
Duration (when and how quickly the damper dampens the string's vibrations, also affected via damper and sostenuto pedals).
There are also things that the feet can do, pushing this already infinite four-dimensional audio canvas into another two dimensions:
Damper and sostenuto pedals affect the strings' ability to pick up sympathetic vibrations of the tone (manipulating both volume and timbre), and
Una corda pedal affects how many strings are struck and what part of the hammer strikes the string (changing volume and timbre).
We can also silently grab keys to lift certain dampers off of strings in order to manipulate sympathetic vibrations, but this is part of the fifth dimension of damper/sostenuto pedaling—whether pedaling with the fingers or with the feet, it is still pedaling.
Are these six dimensions of piano playing really not enough to explain beautiful, colorful piano playing? They should be, and they are. But without fully imagining the magnitude of what six dimensions truly offer to the pianist, the accomplished artist's psyche feels a strong (but as it turns out, simplistic) impulse to invent a seventh dimension while playing the piano—a dimension of manipulating timbre independently of volume.
In search of the seventh dimension
There has been some surprisingly thorough research in search of a seventh dimension of piano playing, especially over the last thirty years. Here are what seem to be the three most significant questions that the strongest research in piano acoustics has answered with regard to this seventh dimension.
1. When playing loud from above the keys (the technique most commonly associated with "harshness"), is it possible that the clicking of the finger or fingernail against the key ("touch precursor," "noise," "early noise," or "transient noise") is responsible for the difference between a harsh tone and a pleasing tone?
In 1929, Otto Ortmann published The Physiological Mechanics of Piano Technique (republished in 1962), which distinguishes between a percussive and non-percussive tone according to this precursor noise (which he just calls "noise"). In his chapter on tone quality, he wrote, "It is now definitely known through both theory and experiment that all qualitative differences [in tone quality], excepting the variations in the noise-element, are quantitative differences [directly related to velocity of the hammer]." (p. 337)
More recently, two Swedish researchers (Askenfelt & Jannson, 1988) sought to find out if listeners could identify "struck" notes (attacked from above) from "pressed" notes (finger resting on key before stroke begins) with and without finger-key (FK) noise. When FK noises were removed from recordings, listeners could not identify struck tones any better than random chance—they tended to give "struck" ratings for louder tones. When FK noises were present, only half of the listeners could identify struck tones most of the time, while the other half could not identify struck tones more than random chance. Another study (Goebl, Werner, Bresin, & Fujinaga, 2014) produced virtually identical results. When finger-key noises were removed from recordings, again listeners were unable to identify struck tones any better than random chance, and when FK noises were intact, again only half of the listeners could identify struck tones most of the time.
In real piano playing, we almost never have a situation where a pianist plays a sequence of notes that are all preceded by perfect silence, which greatly facilitates the distinguishing between pressed and struck tones. If it was "quite difficult" or "demanding" to identify these tones in the midst of perfect silence, it stands to reason it would be virtually impossible to do so in a real musical context. This statement is compounded by the fact that the microphone was less than a foot away from the keys in the second experiment. Not even the performing pianist can hear finger-key noises that well, and yet despite this unnatural advantage, listeners in both experiments still found the test to be "quite difficult" or "demanding." It is difficult to imagine that those sitting in the front row (let alone a middle or back row) of the concert hall could use finger-key noise to identify struck vs. pressed tones any better than random chance. The microphone was ten centimeters away from the strings in the first experiment, but the testers thought the finger-key noises were still "salient" in the recording. As evidenced by the data, the listeners didn't think the finger-key noises were so salient. Furthermore, these studies both show that FK noise does not resonate within the sound of the struck tone in any humanly discernable way, because when FK noises were removed from the recording, pressed and struck tones sounded the same. This confirms Ortmann's findings, showing that FK noise is a cue preceding tone rather than a part of the tone itself.
2. When a key suddenly stops at the bottom of its descent, it thumps against the wood of the keybed. Even with felt between the key and keybed, this noise is still considerable enough to resonate inside the piano. In fact, research has shown that this thumping noise is actually a significant and necessary part of the piano's desired timbre, and piano manufacturers are quite aware of this phenomenon, as they are very careful about the wood they select for the keybed. Can this noise be manipulated independently of hammer velocity?
In the same 2014 study mentioned above (Goebl, Bresin, & Fujinaga), testers found that musicians could hear the difference between tones that included key-bottom (KB) sounds and tones that did not. Testers chose notes near the top of the piano (E7 and F7) to give listeners an advantage in discerning between the lower-frequency KB thump and higher-frequency pitch. To produce a tone without the KB noise, pianists had to play with a staccatissimo technique. In this type of stroke, the finger is pulled quickly toward the palm, both depressing and releasing the key in one motion of the finger (I call this "key-dusting"). The authors of the study correctly acknowledge that this type of touch is used "infrequently" at the piano.
This implies that whenever a master pianist asks us to play "less harshly," we are to play staccatissimo every time, which of course would be silly. The opposite ends up being true in the context of real pianism: whenever master teachers ask for less harsh playing by using a certain technique, the technique always increases the amount of time the key is in contact with the keybed (e.g., "use relaxed arm weight to make each tone sing beautifully").
The testers also acknowledge this musical impracticality: "It might be that key-bottom sounds are also perceptually relevant in real-world settings, but this has to be studied in future research." This statement is scientifically generous to the chance that FK sounds could play any role at all in the context of judging "good tone" without the help of visual cues. There is a name for when visual cues create the illusion of changed sound: it's called synesthesia (when stimulation of one sensory pathway leads to automatic, involuntary experiences in a second sensory pathway). For example, in a competition a judge might comment that the pianist played "harshly" when in fact such a comment would not have been made if the pianist had played behind a curtain. There is a strong tendency for this to happen especially when those judges have strong opinions about correct vs. incorrect technique.
Furthermore, going back to Figure 2, observe that even when playing staccato, the finger is in contact with the key for a full 100 milliseconds. Thirty milliseconds are spent accelerating the key to the bottom of the keybed, and the key rests at the bottom of the keybed for a full forty milliseconds. These time intervals might look significant on a chart, but they are imperceptibly small to human experience. To produce a loud tone on the piano while minimizing KB sound, we would have to accelerate the key downward and then pull back well before the key hits bottom with robotic quickness and suddenness that no human would be capable of executing. We cannot manipulate key-bottom independently from loudness of the tone in the space of five, ten, or twenty milli-seconds without the extreme key-dusting staccatissimo technique.
To demonstrate what happens when real humans play the piano, see Figure 3, which measures the length of time it takes for the key to hit key-bottom in comparison with when the string is struck. While the implications of these timings are fascinating (play loud enough and the key hits bottom before the string is even struck!), neither trained nor untrained pianists display any difference in timing of when the key hits key-bottom when playing at certain dynamic levels.
Even without this data, we can all conduct our own simple experiment. Go to a piano and play a tone of moderate length at the dynamic level of forte. Play the same tone again at the same dynamic level, except this time, make sure the key's thump at the bottom of the keybed is softer. You will not be able to do this—every time you achieve a softer thump, it will be accompanied with a softer dynamic tone.
The researchers also stated, "...we selected a total of twelve tone pairs with two patches, three loudness categories (ranging from 104 to 122 on a scale from 0, silent, to 255, maximally loud), and with and without key-bottom sound, respectively." This small dynamic range of 104 to 122 is to be expected from the fact that the key-dusting staccatissimo technique is the only way to minimize KB sound, and it even more severely limits the musical application of these results. One cannot achieve a ff dynamic level via key dusting, and one cannot hear KB sound at any less than a mf dynamic level.
3. A "striking from above" keystroke has been shown to cause additional vibrations in the hammer shank (the wooden stick that a hammer is attached to), which in turn causes vibrations in the hammer. This vibration continues beyond the escapement point (and escapement itself even causes additional vibrations in the shank which translate to the hammer). Would these vibrations empower the pianist to control timbre independently of loudness?
There are two types of vibrations that are created in the hammer shank: 1) a slow "backwash" effect somewhere around 50Hz, and 2) a faster "ripple" effect that is around 250Hz (see Figure 4). The backwash effect causes the hammer to move toward and away from the string as it travels to the string, which means that the hammer could possibly strike the string with more or less velocity, depending on the timing of the strike. The ripple effect causes the hammer head to move in a direction parallel to the length of the string, which would negligibly change where the hammer hits the string and could possibly affect how long the hammer is in contact with the string.
Unfortunately, the wood length and density used for hammer shanks varies from piano to piano, making the backwash and ripple frequencies vary in ways that would be impossible for any pianist to predict. Even more importantly, supposing the 50Hz and 250Hz frequencies were held perfectly constant among all pianos, even the slower backwash frequency of 50Hz is still too rapid for human perception to be able to deliberately manipulate in the service of tone, either consciously or subconsciously. Rather than being a useful tool, hammer shank vibrations have a randomizing effect upon the piano's tone.
These three concepts, finger-key noise, key-bottom noise, and hammer shank vibration, represented the three most promising explanations for the claim that piano tone can be manipulated independently of volume. These explanations do not pan out, whether because they're not audible enough to make a difference to a listener sitting at a reasonable distance from the keys, because they're not part of the resonating tone itself, because their manipulation requires a technique that is never useful when dealing with the issue of improving tone, or because pianists do not possess the perceptive ability to be able to use it deliberately.
Tone production: Doing the right things for the right reasons
Tone production is a very real part of pianism; it's just that its basis isn't what many pianists think it is. There is nothing wrong with telling a student to play with more "weight" if a less harsh tone is desired. Slapping the keys with the hand (pivoting at the wrist, the same way in which we might knock on a door without moving our arm) can produce a percussive sound, especially when landing on just one or two notes, while dropping the entire forearm with a sense of "weight" (and loose wrist) helps to moderate that sound. While the hand can jerk downward very suddenly, the forearm (which is controlled by larger and therefore slower-moving muscles) has a harder time moving with the same degree of suddenness. Using additional "weight" moderates key (and therefore hammer) acceleration and nothing more. To tell a student that involvement of the arm and wrist achieves an equal dynamic level as involvement of the wrist/hand alone—except "without the harshness"—is nonsense.
As another example, it is still useful to play a note less staccato while the pedal is down in order to achieve a "warmer" effect. But it can only enhance a student's musicianship if we also explain that 1) a staccato touch is more likely to produce a greater hammer velocity because of the quickness of the stroke, or 2) seeing the pianist play with a less staccato touch gives the audience a greater illusion of smooth/warm playing (using synesthesia to our advantage).
What about flat-fingered strokes vs. curved-fingered strokes? Flat-fingered strokes tend to produce blurred articulation (which we call "legatissimo" when it is done purposefully). Curved-fingers, producing strokes that are as energy- and time-efficient as possible, are great for playing with clarity (or even hyper-clarity, such as the leggiero touch), since the fingers move up with so much more ease and swiftness. Flat-fingered playing also tends to moderate dynamic levels (not too soft, not too loud). Given these parameters, it is no surprise why flat fingers are used for a Chopin nocturne, while curved fingers are necessary for a Bach gigue.
Words like "warm", "harsh" and "transparent," as useful as we all find these words to be in our teaching, are nothing more than mere descriptions of certain combinations of note velocities, timings, articulations, and pedaling (together creating tonal wash). "Warm" might represent a generally soft and legatissimo sound, louder velocities in the lower note registers, or a phrase with a quiet dynamic peak. "Harsh" would be generally too loud. "Sparkly" might represent louder velocities in the upper registers with a staccato touch. "Transparent" might come to mind when we hear an unpedaled, pianissimo Alberti bass.
Acoustically, harshness is an especially interesting feature of piano playing. We know that a harder hammer on the piano produces a harsher, bright tone. But exactly how does this happen? A harder hammer will bounce off of a string more quickly than a soft hammer. This allows higher partial frequencies to be excited before they are dampened by the hammer itself. Yes, that's right: hammers actually act as dampers when you watch a hammer hit a string in slow motion. This hammer-dampening effect is more pronounced with softer hammers. Imagine a pillow hitting a string—the pillow would be in contact with the string for such a long amount of time that even the fundamental frequency of the string would be dampened long before the pillow finally "bounces" off of the string. If a hammer is too hard, it can even excite partials (overtones) to louder amplitudes than the fundamental frequency itself, creating an unpleasantly bright tone. Indeed, harshness is caused by hammer velocity or hardness, not by some meta-velocity technique that the pianist failed to use. On the piano, volume and timbre are inseparable.
In The Physiological Mechanics of Piano Technique, Ortmann gives a similar list of tone qualities (sparkling, velvety, crisp, bell-like, dry, brittle, singing, pearly—p. 339-352), giving similar explanations of why these qualities are all valid despite the one-to-one correspondence between loudness and timbre. It is also worth noting that these words don't mean the same to every-one. For example, I think a tone sounds more "bell-like" when the damper pedal is used, while Ortmann argues that use of the damper pedal will weaken the bell-like effect. (p. 342)
Ortmann warns of the danger of synesthesia—the great need to get away "...from eye-impressions in any analysis of auditory tonal qualities. So many qualities are read into the piano tone by the eye, and this is so often done, that it is very difficult for even an experienced listener to dissociate the two sense-impressions." (p. 341) Science has overwhelmingly confirmed the legitimacy of this warning. Research has shown that what we imagine influences what we see and hear (Berger, Ehrsson, 2013), what we see influences what we hear (Shams, et al., 2000; also note the "McGurk Effect" as coined in 1976 by Harry McGurk and John MacDonald), and what we feel (tactile sense) influences what we hear (Gick & Derrick, 2009).
Ortmann also notes that with a "singing" tone, muscle contraction often peaks after the tone has already sounded:
"All this increase is wasted effort, of course, so far as its effect upon the tone is concerned. Nor do we find good pianists normally using much of it. It is a favourite device of the emotional amateurs who read all sorts of pathos and romance into the long-suffering piano tone." (p. 346)
A slightly different hand position can drastically change the timing and velocity of notes (as well as timing of release), and therefore the tonal wash. When you have thirty or forty notes in a passage, all approached with one technique as opposed to another, the combination of all these velocity and timing variations might sound so different from one pianist to another that we can sometimes even consider this tonal wash to be as unique as a signature or fingerprint. One simple change in hammer velocity within a chord (on just one note) can affect the overtones present to such a degree that we perceive a change in "piano timbre," when actually it was caused by the volume of a single tone.
Those who believe timbre can be manipulated independently of hammer velocity sometimes object to considering tones in isolation. Ortmann points out a double-standard here:
Objection is often made, and somewhat justifiably so, to the assignment of tonal qualities in the single tone. And in the experiments made, this objection was frequently voiced by the pianists making the records. However, if the qualities assigned by the player to a tone passage do not exist in the single tone, that very fact is definite proof that the so-called tonal-qualities are not inherent qualities at all, but result from dynamic and agogic variations among a succession of tones." (p. 355)
Pianists and teachers should be able to agree that musical beauty doesn't come from playing a single tone with the fingers, hands, wrists, elbows or shoulders pointing or moving this or that way. Nor does it even come from relaxation, tension, weight, or levity. Musical beauty comes from intense listening on the part of the pianist, constantly adapting and calibrating every tone in response to tones that have passed and in anticipation of tones to come, regardless of what physical mechanics are used in service of this beauty.
Adapt and embrace
All great pedagogues understand that teachers must adapt their teaching to students. Part of this adaptation should include being able to teach in concrete terms to students who do not work well with illusions. We should be able to explain the "why" behind anything we teach clearly and effectively. If not, we should be transparent about our ignorance (there is nothing wrong with not knowing everything!) and take some time to reflect upon what is being taught. Is it the wrong principle, or is it the right principle for the wrong reason? This reflection ensures that our most curious, intelligent, and engaged students are not violated, insulted, and/or silenced for their mindfulness. As a student, I would have been much more open to following illusory advice if it were delivered with more humility or transparency. ("I am not sure why this works, but I find it works well for me.")
It is truly wondrous and awe-inspiring to realize how much is possible within these deceptively-complex six dimensions of piano playing. In fact, embracing this opens up a new door of artistry when you consider the words of Orson Wells: "The absence of limitations is the enemy of art." While I don't believe it would actually be a vice to invent a piano that would be capable of allowing fingers to manipulate timbre independently of volume, I think this new mindset makes us stand in even greater awe after witnessing a colorful performance of a Haydn sonata, such as what I heard from Olga Kern at the 2009 MTNA Conference. To think that Kern was able to produce her magical sounds without the use of a seventh piano dimension is all the more inspiring.
By examining and confronting the best science behind tone production that we can find, we are empowered to satisfy our most curious students and to enlighten the others, and we embody the idea that in piano pedagogy, the art and science of teaching should be inseparable. This enlightenment does not put ourselves in shackles of realism. We are instead liberated from the prison of illusion and artistic dogma.
The last temptation is the greatest treason: To do the right deed for the wrong reason.—T.S. Eliot, Murder in the Cathedral
Askenfelt, A. (1991). "Measuring the motion of the piano ham-mer during string contact." STL-QPSR (Department for Speech, Music and Hearing KTH Sweden: Quarterly Progress and Status Report), 1991.
Askenfelt, A., & Jansson, E. (1988). "From touch to string vibration." Five Lectures on the Acoustics of the Piano. Östen Häggmark, ed., May 27, 1988. http://www.speech.kth.se/mu-sic/5_lectures/askenflt/askenflt.html
Berger, C. C., & Ehrsson, H. H. (2013). "Mental imagery changes multisensory perception." Current Biology 23, (14), 1367-1372.
Burred, J (2004). "The acoustics of the Piano." PhD diss., Ph. D. thesis, Professional Conservatory of Music Arturo Soria.
Galembo, A. (2001). "Perception of musical instrument by performer and listener (with application to the piano)." Proceedings of the International Workshop on Human Supervision and Control in Engineering and Music, Kassel, Germany. 2001.
Gick, B, & Derrick, D. (2009). "Aero-tactile integration in speech perception." Nature 462 (7272), 502-504.
Goebl, W., Bresin, R., & Galembo, A. (2004). "Once again: The perception of piano touch and tone. Can touch audibly change piano sound independently of intensity." Proceedings of the International Symposium on Musical Acoustics. 2004.
Goebl, W., Bresin, R., & Fujinaga, I. (2014). "Perception of touch quality in piano tones." Acoustical Society of America, 136, p. 2839-2850.
Ortmann, O. (1929). The physiological mechanics of piano technique; an experimental study of the nature of muscular action as used in piano playing and of the effects thereof upon the piano key and the piano tone. New York: Dutton.
Rosen, C. (1999, Oct 21). "On playing the piano." The New York Review of Books, 46 (16).
Rosen, C., Wolf, K., & Ryser, M. (1999, Dec 16). "Playing the piano." The New York Review of Books, 46 (20).
Rosen, C., Ross, D., & Kunin, C. (2000, Feb 24) "Playing the piano." The New York Review of Books, 47 (3).
Shams, L., Kamitani, Y., & Shimojo, S. (2000). "Illusions: What you see is what you hear." Nature 408 (6814), 788.
Twedt, C. (2010, Jul 8). "The piano voodoo of tone production." Cerebroom, http://blog.twedt.com/archives/222
Twedt, C. (2010, Oct 18). "Studies addressing piano voodoo of tone production." Cerebroom, http://blog.twedt.com/ar-chives/657