I read the PACE book about tuning. The term 'ghosting' is being mentioned but not explained.

Ghosting technique to help you hear beats ? How does it work ?

As I understand it, when you're tuning an interval, you are listening for beats in that interval. That is because their harmonics have a point (or more) where they fall on a common note. If you silently hold the interval (to lift the dampers off the strings, but don't actually sound the note (hammer doesn't hit the strings), then give a quick staccato jab to the common harmonic, you can hear the beats in the sympathetic vibration.

For example, let's say you wanted to ghost the F3-A3 minor third. The first common harmonic for this note is A5, so you would hold F3 and A3 silently, and do a quick stacatto on A5 and listen to the sympathetic vibrations. Also, the F3-D4 major 6th shares the same common first harmonic, but if I remember correctly the beat rates between the two intervals will not match on a tuned piano. As a couple more examples, for the F3-G#3 minor third, the first common harmonic is C6. Also for G3-C4 fourth, the first common harmonic is G5. For the C3-A#3 minor 7th, the first common harmonic is A#5, and for the F2-A3 10th, it is A4.
If your hand is large enough (assuming the piano doesn't have a sostenuto pedal) to reach a twelfth like F2-C4, for that one the upper note of the interval is also a harmonic of the lower note, so in that case you'd play the note an octave higher (C5) while holding down F2 and C4.

How do you establish the common harmonic ?

When you say "1st common harmonic for this note is A5", what does 'this note' mean ?

You basically go up the harmonic series for each note, to find the common note.

For example, the series of F3 is F3, (octave) F4, (5th) C4, (4th) F5, (maj 3rd) A5, (min 3rd) C6, (min 3rd) D#6, (maj 2nd) F6, (maj 2nd) G6, (maj 2nd) A6 and so on, and for A3 it's A3, A4, E5, A5, C#6, E6, G6, A6, B6, C#7 and continuing. For that interval and sequence of harmonics, the first common one is A5.

Of course those harmonics are not all "in tune".

"Common harmonic" = "coincident partial," non-rigorously.

If you're being rigorous, you use "harmonic" to mean the theoretical multiples, and "partial" to mean the actual string vibration frequencies, which includes... tada, inharmonicity.

Thanks for the correction, Jim. Must have gotten my terms switched or something, lol.... (maybe I must be listening to too many old recordings on which the piano is ~150-250¢ flat, and all the unisons are 15-40¢ out of tune lol)

Hi Jim, please don't mind explain this paragraph posted by 88key:

"For example, let's say you wanted to ghost the F3-A3 minor third. The first common harmonic for this note is A5, so you would hold F3 and A3 silently, and do a quick stacatto on A5 and listen to the sympathetic vibrations."

What is the purpose of 'ghosting' an interval ?

And why would A5 be in the picture of F3-A3 ?

To isolate a beat by itself. It makes it easier to hear than when an interval is struck normally and all of the partials are sounding at the same time.


Because the test interval is a M3. Can you name the 5th partial of F3? The 4th partial of A3?

Thank you for the clarifications. I believe Bill Bremmer in another thread of mine explained it quite clearly too. Surely now it has to be A5.

It is tricky to have to keep straight in your head that when someone writes a couple of numbers together like 6:5 that you have to determine from context whether it is:

• A ratio of vibration frequencies
• A nickname for an interval
• An octave width
• A locator for test note in an interval check
• An ETD style
• An aesthetic preference
• ...?

I was also someone who found it necessary to approach it from an analytic standpoint, though I had a personal mentor willing to field my torrent of questions, and who could keep track of where I was at so he could figure out what I meant by my questions, and could ignore what I actually said :-)

Thank you, Jim.

BTW, can I safely say that ghosting can only help you hear the harmonics on your right hand side. Someone did a demonstration on a digital keyboard and claim that he can hear the notes on the left when an interval is played on the right.

Yes.


Well, there's this, that Erus has brought up before:

http://en.wikipedia.org/wiki/Missing_fundamental

But I don't think it would have anything to do with ghosting, as a digital does not have a soundboard to produce the same effect.

Some high end digitals model string resonance, so I guess I can see the possibility of hearing the ghosting.

I don't see how the reverse process could work, but I'm intrigued enough to try it tonight. Any suggested intervals?

Ah! An interval is a different story. Any two notes played together produces an interference pattern at the sum and differences of their frequencies. We usually just think of this as beats in nearly coincident partials. Although I have not tried it (but will), playing any two consecutive partials of a fundamental could ghost the fundamental.

I had a Yamaha Clavinova before I became a piano technician, but then gave it to my brother (who then went and sold it!) Should've kept it so I could run out and try stuff like this! Maybe a foray over into one of the non-tech forums, and ask someone to try it.

I don't have a digital but I don't see how any of them could ever produce the ghosting effect. It doesn't involve the soundboard itself. To do the ghosting effect, you have to press down 2 keys without the hammers striking the strings in order to hold the dampers off of those strings. How would you do that on a digital? Then, you strike the note which corresponds to the coincident partial of the 2 keys you are holding open. That sound excites those two strings. There are no strings in a digital! The keys operate an electronic switch which causes current to flow and you hear an electronic sound.

The digital cannot be influenced by outside sounds such as when you press the damper pedal and "hoot" into the piano and hear the echo. That could not happen on a digital. That would be essentially the same phenomenon as the ghosting technique. The beat you are attempting to hear from the two strings whose dampers are held open is caused by an external source. If you could "hoot" the coincident partial pitch, you would hear the same effect as striking the coincident partial pitch.

But doesn't ghosting depend on the soundboard, or at least, the bridge/soundboard coupling? I don't think the vibration in the air caused by the note at the coincident partial location would be sufficient to cause the corresponding partials in the held notes to resonate.

And I think what Tim was referring to was the idea that in a sophisticated digital, the algorithms simulate the vibration of strings and the response of a soundboard. Were the simulation sufficiently complete, you might imagine that the vibration of the coincident partial note would cause those algorithms to impart that simulated vibration to the held notes (whether we agree on the mechanism) and produce "digital ghosting."

Unfortunately, I don't have any acquaintances with a Pianoteq or Garritan setup to try it out on. And it's a very particular effect that we look for, so I doubt just telling someone to try it and report back would satisfy my curiosity.

Yes, Jeff, absolutely - in my music theory classes, I bring the fundamentals alive by using intervals conformed of partials, to illustrate the overtone series. Makes a great impact, the students go "wow" every time



Jim: it depends on all the factors you stated above, and beyond that - the angle of the lid, the room acoustics, the temperature, the humidity, wether you are striking the key with a ten-foot pole or sitting by the piano...



Ladies and gentlemen... *drum roll*... finally a discussion where I can contribute and not just feed on your vast knowledge :-D

There can absolutely be no kind of ghosting whatsoever on a digital piano of any kind, for the exact reasons you state, Bill: there is no acoustic interaction between the tones. There is actually no acoustic laws AT ALL taking place in a digital piano (except for the mellow "thump" when the keys hit the keybed, and that information is never picked up the keys are just readers of velocity, speed, duration, and so on.

Soon after the (physical) reading, everything is converted to a digital format - binary 1/0, and a varying amount of bits (=possible combinations of 1/0).

• Thus, for example, a CD uses 16 bits, which gives 2^16= 65 536 different levels (reading resolution.) This feels like overwhelmingly much, but the human ear is marvelous at registering dynamics. 16 bits gives us only a 96 dB dynamic range. This is ok for 'compact' music (ie pop music etc.), but when you record classical music that ranges from pp to ff, you get into trouble in the "low bits" - that is, the quiet parts of the music. The remedy has been to apply noise [yes, noise] to the digital stream, to have something to attach the low bit data to. This is commonly known as "dithering".
  
  The preferred resolution in digital sound recording/reproducing has shifted to 24 bits (giving us a dynamic range of 144 dB), and continues to move beyond that. Still, we are stuck with 44.1 kHz/16 bit, until the CD goes out of fashion (which I predict will happen in less than ten years)

Well, the last paragraph was kind of off-topic, but I'll bring it back on track:

The modern digital pianos have two crucial improvements addressing the two limitations that steered me, as a piano player, away from digitals in the first place.

The first improvement is the sustain pedal, nowadays the high-quality digitals register 1/2 (and recently, even a 1/4 pedal). Thus, it's not just off/on anymore, which used to drive me crazy.

Secondly, they now include algorithms that mimic what happens when you press the sustain pedal on a "real" (so bite me, digi-freaks) piano. Previously, the only thing that happened was that the note you were sustaining was prolonged. Nowadays, there is a simulation of the sympathetic vibration of the whole piano (frame, soundboard, strings etc). The digital sustain pedal simulation calculate the data registered (ie keys pressed) and adds resonating harmonics (partials) into the sound, imitating the natural laws of acoustic, giving the digital more of a "real", alive sound.

This is, and will always remain, a substitute for the acoustic phenomenon of sympathetic vibration - the very essence of (and a crucial, mandatory requirement for) the tuning technique known as "ghosting". To try ghosting on a digital is somewhat like using a flight simulator - although the effect resembles real life, you're not really flying

Patrick

Part-time tuner, professional piano player, and former pro sound engineer - nowadays the sound engineering is more of a therapy form

If you have an accurate algorithmic model of string vibration, its interactions with a bridge, soundboard, and dampers, then what makes you think it could not produce ghosting?

Check this out:

http://www.youtube.com/watch?v=EjowZw0YjAk

Ghosting may seem like a special thing, but it is no different than the "normal" (pedal-up?) interaction between strings. It's just that we play dampening games such that we can isolate partials.

? there is no ghosting on a digital piano. only surrogates. sorry if i sound like BDB (no pun intended), but that's the case...

[ADDED GMT +3]: Jim - what, exactly, is "the string vibration, its interactions with a bridge, soundboard, and dampers" of a digital piano? It ends right there - or at least, it would steer you towards admitting that its a substitute for the real thing.

There is no physical symphatetic resonance of strings on a digital piano - it doesn't really matter how much we try to replicate it. It is very much like watching old cartoons where one could say "but it's red, i can see it". Yep. But the original was B/W, the color was added afterwards.

Jim, to very particular, this is what happens on a digital piano if you try to ghost notes (I still, for my life, can't figure WHY, but i leave the window open for new breezes...)

1) keys pressed and held down (ie C2/C3)
2) key struck G4
3) Comp Algorithm (simplified!):

if (C2 && C3 is pressed down) {
if G4 {
do this (mimic resonance (C2,C3))
}
}

It's all a replica. Sure, you might hear something resembling ghosted notes, but why bother?

I am typically amused by terse answers as well, no mis-interpretation of tone of voice from here.

I agree there is no use in ghosting on a digital, because we're not trying to tune it. So perhaps this discussion is of no use. But the subject, as brought up by Bill, was whether the excited partials exist at all in the two notes held while the test note is truck. And you stated there is no way it can ever exist. I claim, as TimR suggested, that in the newer digitals, it is possible for the same effect to have taken place (in the simulated world), and that it may be possible for a ghost tone to exist.


It is a mathematical model of how vibration interacts with a physical object. For this discussion, there is more detail here:

http://en.wikipedia.org/wiki/Physical_modelling_synthesis

It is not the same thing as simply playing back samples of a recorded piano.


It would be more accurate to say it is a simulation of the real thing :-) But yes, for someone who loves real pianos, I agree that is a substitute for the real thing. But it doesn't "end right there," if I may continue.


I think that is exactly what Roland claims the V-Piano is doing, simulating the sympathetic resonance. You might say "it doesn't matter how much we try to replicate it" about the entire effort to model real pianos, and yet they are getting better at it all the time.


And that would indeed be a ridiculous thing to do, since for music we don't care about ghosted notes. So why put a special case in your code to simulate it? (Don't worry about my understanding of software, it's been my profession for close to 25 years now, having taken up piano work only in the past few years.)

Or do we care? When we hold the damper pedal down, and play notes, we get sympathetic vibration in strings which we did not press. If we simulated this resonance by the aforementioned modeling technique, then we would be closer to making our synthesized sound more like a real piano, which is the goal of the people writing this software. It follows then, that ghosting would happen without any extra special case code, if we created the same conditions (two notes held, etc.) and if the sympathetic vibrations were modeled correctly. And if this is the case, then the above pseudocode you provided would not be needed, and we would hear the ghosted notes without there being special code written for it, only by virtue of more accurately simulating the piano in the first place.


Well, one WHY is that as piano technicians, our lives are being changed by digital technology. Why don't piano manufacturers make spinets any more? How does that affect our respective markets?

Perhaps my major point is this: it doesn't matter that there is no actual, physical sympathetic vibration. If the vibration characteristics have been accurately simulated, and the resulting signal can be faithfully transduced into waves in the air, then digital "ghost" tones exist. And that makes me curious to try it.

Jim,

great answer - my replies can be found below:



Aah, but that is not how I interpreted Bills text. The way I read it, he said that there could be no ghosting, and I agree 100% because ghosting is a physical phenomeon. I also agree 100% with you in that partials do exist in the simulation algorithm of, for example, a Roland V-series. Is this contradictory? I don't think so...



Let's agree that its both a simulation AND a substitute :-D


Sorry, I'll rephrase: There is no physical symphatetic resonance of strings on a digital piano - no matter how much we try to replicate it, there will be nothing acoustically happening.

I personally depend on digitals in a lot of situations, and I do like both what Roland is doing in the V-series, and the fact that they are getting better all the time.


well, what do you know - in that case, sorry for my pseudocode and being overly explaining!

I agree that that would be some rather dull and unnecessary programming, as there would be quite a few "if" statements :-D I still believe there might be some use of my example to illustrate what is basically happening in the modelling algorithm itself. This was not meant as a working example of code (as you with your background know very well ) it was just one more way to say that it's nothing really physical happening, just simulation.




Yes, the digital "ghost" notes exist as a simulation, but there is no ghosting happening... :-D
I think it's mainly a language knit-picking, and I also think we agree on mostly everything. But it's been a nice discussion - even though I faintly recall that I used to go out partying during weekends when I was younger, and now find myself in front of the computer, writing in forums.. well, there is a season for everything

Oh, and by the way, there's another developing going on, taking a slightly different path towards "real life" simulation: the Yamaha AvantGrand.
http://www.popsci.com/gear-amp-gadgets/article/2009-02/yamahas-digital-grand-piano

I played this for half an hour a month ago or so, and it's actually a totally reversed feeling to my previous encounters with digitals. The AvantGrand feels great when you play it (because of the physical simulation happening.) I'm not overwhelmed by the sound, though, when you just listen to directly lined recordings of it. The V-series sounds better, the AvantGrand feels better. Such a shame that these two keyboard giants can't combine their efforts!

So what you're saying is that if I do go find one of these digital pianos, perform the test, and hear a beat, I'm not really hearing ghosting happening?


That's neat you've been able to try both, I hope too soon as well.

I think one of the fundamental problems digital pianos always have is that they all require a very high quality amplification and speaker system to get good sound. People alreay spend the price of a grand piano in order to get "audiophile" quality sound, one that can reproduce a well-recorded piano, and that's without all the engineering required to produce the digital piano.

See? It's just a question of definition. We both seem to agree on what happens both in an acoustic and in the digital modelling. I always understood the term "ghosting" as the physical phenomenon itself. We might hear the partials in the digital, but to me that should be called something else.

Needless to say, I have no registered TM on the term, so feel free to let it include whatever feels right for you



I wonder if you'll have the same experience that I had, about the good feeling of playing the Yamaha. Funny that it feels so good that I got disappointed when listening to the sound files later!
BTW, the AvantGrand also boasts some kind of sympathetic vibration modelling, as well as at least a 1/2 pedal function.
Still, the (line-out) sound of the Roland V is much better in my opinion.

I gave it a try and it can work. A fundamental can be ghosted by playing two adjacent partials of the series. They do not even need to be played together, they can be played one at a time or back and forth. It sounded like the fundamental would increase in volume as the vibrations came into phase, or something. The 2nd and 3rd or 3rd and 4th partials seemed best. I can only think of one practical use, though. It could be used to determine the low end response of the soundboard. At some point the fundamental cannot be heard when going down the scale. This point might be determined by ghosting the fundamental.