Ah, Glenn, I've been more  delerious than that, come on   A few years ago I had the great pleasure of meeting Steve Vai and hearing him explain how his favourite guitar was picked from a line-up of identical, mass-produced CAD-CAM cut stock alder bodies because it just felt and sounded a little bit different from the others, had a different "voice."  I'm not really into the metaphysical but when I held "Evo" I "felt" something.

All that notwithstanding,  is there not sufficient (I hate to use the term and it's not quite the right one) quality control to prevent our Steinway test piano from being confused with the Bosendorfer one?  I would like to think so.  I wonder whether the "identical" models in your anecdote were consecutive numbers i.e. subject to the same scrutiny of raw materials as my test case?

...which still doesn't answer the question of what makes a Steinway sound like a Steinway and a Bosendorfer sound like one of those, apart from the "proprietary resonance technology" of the latter?

Does string gauge/tension come into Modartt's calculations?

Best//Neil

Neil:

There is no doubt that most of us (including me) are sure we can pick out the difference in sound between the major brands.

But I wouldn't bet on my ability to discern a Hamburg Steinway from a New York Steinway, and I'm not going to state that I could always do the same between a Steinway and a Bosendorfer.

No doubt there are differences, and these can most likely be noted in a side by side comparison, but if I led someone into a room blindfolded, and struck a few chords on one piano, I really wonder how many could say, "oh that's a so and so" for sure.  Without hearing both brands in the same room at the same time.

If the same test was conducted with two pianos (which were not previously identified) how many could accurately say, "piano A is a Steinway, and piano B is a Bosendorfer"?

I recall - when I was in my early thirties and still had good hearing - going to a performance by Ferranti and Teicher (duo pianists).  There were two nine foot concert grands set with their curved surfaces close together, lids completely removed.  The two pianists not only took turns playing melody and accompaniment, they switched pianos during the performance.

My wife and I had great seats (I observed that one was a Steinway, the other a Yamaha).  As I recall, I didn't notice the difference in the two pianos - most likely because when the Yamaha was playing runs up and down the keyboard, the Steinway was playing the melody.

Yet it is consistently said that Yamaha have a brighter (harsher) sound than other pianos.

Yes, I've always said that I prefer the Bosendorfer over the Steinway, but this may be because I've had several great life experiences on a Bosendorfer that I didn't have the fortune to have with a Steinway - I bonded with the Bosendorfers.

Glenn

PS - back from lunch.  What makes one brand sound different than another?  Looking at them doesn't seem to provide many clues.

Considering that the piano in its present form is a century old, and that there have been no major innovations I know of except Kawai's carbon fibre action, I suspect that "rules of thumb" have governed piano design for the most part.

With the advent of high tech analysis and testing procedures, acoustic speakers have improved dramatically over the past few years, but every time a new idea is to be tested on a grand piano, one must build a whole new piano - very time consuming and expensive.

Kawai made a major leap forward (in my opinion) by utilizing CF in their actions, but it hasn't caught on.  My piano rebuilding friend and I have discussed this and have concluded that conservatism rules supreme in piano building.  Compared to guitars say, pianos are very static in terms of new ideas.

All that notwithstanding,  is there not sufficient (I hate to use the term and it's not quite the right one) quality control to prevent our Steinway test piano from being confused with the Bosendorfer one?  I would like to think so.  I wonder whether the "identical" models in your anecdote were consecutive numbers i.e. subject to the same scrutiny of raw materials as my test case?

...which still doesn't answer the question of what makes a Steinway sound like a Steinway and a Bosendorfer sound like one of those, apart from the "proprietary resonance technology" of the latter?

Does string gauge/tension come into Modartt's calculations?

Best//Neil

Let's say we approached Yamaha, Steinway and Bosendorfer and asked them all to make a piano.  They have to use the same wood stock for the soundboard and case, and the string A0 sounding lengths have to be the same, but other than that, they can use their own methods.
Best//Neil

Well they'd definitely sound different, and the touch would not be the same.

But here's the kicker that may increase your delirium.

If each manufacturer made four seemingly identical models, there would be noticeable variation between the pianos of the same manufacturer.

I personally am aware of two instances where a purchaser traveled to New York (prearranged of course), and "test drove" a stable of four or five pianos.  The procedure lasted several days to a week.  The pianos tested were of the same make/model, and visually appeared identical.

The pianos were all in the same room.

After much listening and deliberation, one piano was selected as being the "best".

The cases were:  School of Music, University of Regina (where I lived in the '70's and '80's) and the Royal Theater, Victoria, BC (where I presently live).

Glenn

I use headphones too - an old Sennheiser HD430 and Beyer Dynamic Pro DT770 that to my ears seem to sound pretty good - then again sound seems to get very subjective - if I am to master some song I will have to finish at 2 hours tops or I won't hear the differences clearly and crisp enough....

Further it is actually very weird that in our pursuit for musical perfection we actually have to include all those imperfections (that we expect to hear) to get it perfect...

Hmm, Variax.  I have a PodXTLive and while it sounds better with my ESP H-1000 (with EMG actives) rather than with my Ibanez JS-1200 (which I sold this year and actually hated) I  still can't get on with it.  I think Native Instruments' Guitar Rig is better (especially the Bogner clone), although less portable, but still in another league entirely from the Fractal Audio AxeFX which no way could I afford to buy.  I  think the Variax falls down because it's  Piezo, and will always have the funny Piezo attack which is neither acoustic nor electric (you know what I mean, even though actually it's both).  It's certainly an interesting idea though.

Another complicating factor for realism (back to Piano  now)  is definitely the replaying system.  I once read that for convincing reproduction of  a flute and piano  performance, a system would require a minimum of  300W of electrical power to match the acoustic power generated in reality.  Then again, it never  ceases to amaze me how much noise acoustic instruments  generate, seemingly  effortlessly.

I mostly play Piano through high-quality headphones as I live in a terraced house, and I listen at a level which is comparable to sitting at a real grand.  I'd have to say that  certainly  since 3.5 came around, the experience is very close to that of a real piano, just not the exact piano I wish it was.

Best//Neil

Get well soon! you sound pretty delirious!!! 

As you, I would really like to know what it is that they do.

I would say that a whole bunch of parameters are possibly combined to mimic real life phenomena.
And probably a whole lot more parameters woud have to be added to get even more realistic behaviour.
What about humidity for instance?

Problem with modeled or simulated sounds is that they can sound very or even extremely good unless you have the real thing right next to it.... then there can be these little tiny things that are in the sound that give it this litlle extra... this.. life.
But as soon as you listen to the piano sound in an arrangement with other instruments you won't hear it or have to your utter best to hear it...

I have the same thing going on with guitar - I use a Variax modeling guitar and Variax bass - they can pose as quite a lot of different guitar brands.
While not being 100% perfect, they're pretty darn close and at a gig I can trash as the next heavy metal band or play jazzy or acoustic and in the total mix it will sound perfect...

[Warning:  I'm perhaps slightly delerious]

Conventional synthesis/sample manipulation software is too complex and the majority of controls bear no resemblance to the physical processes in the real world we're trying to replicate.  This is one reason that physical modeling appeals to me.  The former is like reverse-engineering nature.  Physical modeling is more like copying nature.

What bothers me is, how many of the presumed parameters are purely mathematical and have little or no relation to physical components/processes/construction in an instrument? 

As a sort of example from the current user interface, take the harmonics manipulation section:

We know that piano makers try to limit  a certain harmonic/overtone as it doesn't sound nice.  Fine.  We have a "harmonic equalizer" to deal with this.  Problem:  A piano maker doesn't have a harmonic equalizer.  He has a ruler.  Enter PTQ 3.5.  Now (and more so for Pro users) we can set the hammer strike position ourselves.

But we still have the "harmonic equalizer," and now some of us have it for every single note.  What is this control-set a substitute for?

I can only surmise that this is present because there are other piano construction details which affect harmonic developement, which we *don't* have access to. Maybe  Modartt doesn't have access to them either.  Maybe not all components of  piano tone production can be measured and therefore expressed mathematically.

Let's say we approached Yamaha, Steinway and Bosendorfer and asked them all to make a piano.  They have to use the same wood stock for the soundboard and case, and the string A0 sounding lengths have to be the same, but other than that, they can use their own methods.

We will assume that the 3 pianos sound quite different from each other and recognizable as the work of the particular builders.  What factors other than the ones we "fixed"  are responsible for the differences?

Can these parameters all be defined?  How much effect does the individual action mechanism of each piano affect the tone, and in what ways? 

Listen to Francois-Joel Thiollier's Debussy  recordings on Naxos and try to work out what causes the strange "buzzing" tone of his piano (also present on a couple of notes of the Bosendorfer in the East West library).  I don't know who made that particular piano, but I do know there is nothing in PTQ that sounds anything like it.

Do all piano makers use the same string length scaling charts?  What I mean is, if our 3 pianos had an A0 string length of 1.8m, would they all have the same string length at middle C?.  Let's say the lengths are the same, but the wire gauges aren't.  What happens now?

What about hammer shape and size, i.e. size of the contact patch with the strings?  How much influence does that have?

What about bridge construction?

What else could usefully be made available in PTQ Pro that I haven't considered?

When Modartt work on a new piano model, what do we think they do, actually?  Obviously the piano is recorded accurately, note-by-note (where possible) and I presume accurate physical measurements are made also, but what data really goes into the modeling process?  How much FFT analysis of the real piano is used to produce a PTQ file and in what way is that data used? 

Philippe has  said  before that due to current computer capabilities, certain components have to be pre-determined to save  CPU cycles.  I wonder what those components are and whether the offline calculations for them are the same for each piano model or computed separately?

Best//Neil

johnrule wrote:

It would be very interesting to be able to load any waveform in place of the default...

While true, this absolutely is _not_ the purpose of Pianoteq.  You _really_ want a synth based on complex piano mechanics, but we're all SOL on that one for now.  ;^)

As Neil said above, the complexity of the Pianoteq "master laboratory" is, most likely, STAGGERING.  The _next_ step (or, honestly, giant leap) for our dream-team to make is the creation of "Pianoteq Factory" (or the like), but this involves simplifying and streamlining an _incredible_ number of components.  We shall not see such a beast for some time...

And I don't mind waiting, because I'd rather have MODARTT focus on further improving/refining the current software _while_ providing additional, high-quality add-ons, provided to us with their collective years of experience and know-how.  (Personally, I'm hoping for a whole _crop_ of new add-ons sometime in the near future -- not because I'm bored with the vibes or the YC5, because I love them _both_, but because I know that they'll do a complete, sincere job.)

(The likelihood of making a _bad_ instrument from manipulating the "deep" parameters is probably very easy;  what we definitely _don't_ need are a bunch of ham-fisted attempts at new piano intstruments, concocted by users who would be fumbling about blindly [at first, at least].  This would only damage Pianoteq's reputation.)

I think you make some good points.

It took over a one century for the piano to evolve into its present state.  There were many failed experiments, but it gradually evolved into its present form, which hasn't significantly changed for 100 years or so.

If one were to start from scratch, and attempt to "build" a piano in Pianoteq, it would take some considerable knowledge and effort - and most likely much time.

And as Neil pointed out, how many of us have the mathematics combined with the piano technician's skills and knowledge to be able to come up with something useful?  I suspect none of us.

Glenn

It would be very interesting to be able to load any waveform in place of the default...

While true, this absolutely is _not_ the purpose of Pianoteq.  You _really_ want a synth based on complex piano mechanics, but we're all SOL on that one for now.  ;^)

As Neil said above, the complexity of the Pianoteq "master laboratory" is, most likely, STAGGERING.  The _next_ step (or, honestly, giant leap) for our dream-team to make is the creation of "Pianoteq Factory" (or the like), but this involves simplifying and streamlining an _incredible_ number of components.  We shall not see such a beast for some time...

And I don't mind waiting, because I'd rather have MODARTT focus on further improving/refining the current software _while_ providing additional, high-quality add-ons, provided to us with their collective years of experience and know-how.  (Personally, I'm hoping for a whole _crop_ of new add-ons sometime in the near future -- not because I'm bored with the vibes or the YC5, because I love them _both_, but because I know that they'll do a complete, sincere job.)

(The likelihood of making a _bad_ instrument from manipulating the "deep" parameters is probably very easy;  what we definitely _don't_ need are a bunch of ham-fisted attempts at new piano intstruments, concocted by users who would be fumbling about blindly [at first, at least].  This would only damage Pianoteq's reputation.)

If Pianoteq were an Open Source project, every single parameter would be made accessible to everyone.

Not necessarily.  The information required to produce a piano "from scratch" is in a binary.  For obvious reasons.  Even if it wasn't, we'd probably be in a worse position than I was as a Hauptwerk 2 beta-tester, having  to read a 250+ page manual just to understand (ha!) the file format.  And that's just a fancy sample player.  Without a solid grounding in applied mathematics and the acoustics of pianos, the "747 Cockpit" of equation values would probably be meaningless.

Best//Neil

JR

I for one absolutely love Pianoteq, it's playability and editability.
But apart from that I'm still curious about what else they can do to a model ... buy yes I know... curiosity killed the cat!

cheers
Hans

No, I have asked the same question some time ago and was told that they themselves use a bunch of parameters the size of a Jumboject cockpit!
I did not get a definite answer on ho to be able to make a ptq-file...
They only talked about fxp-files..
Pianoteq Pro itself does not have any extra export options - only fxp can be exported.

If Pianoteq were an Open Source project, every single parameter would be made accessible to everyone.

But Modartt is a small company who wants to make some money developing and selling the (IMO) best, most innovative, software piano on the market.

I think, that if they would make EVERY single parameter of the Piano Models editable by the user, it would be easier for the competition to make a Pianoteq-like instrument or software using reverse-engineering.

Even though I'm a great fan of Open Source (I'm a Linux user) I can also understand Modartt's wish to make some money. If it helps them to make a superb product, everyone is happy - including me.

G

... tutorials, advanced tutorials step by step, and also some tutorials about the piano behavior.

We must realise that few people knows how to tune a piano and the acoustic behavior of a piano.  So why not teach that art?

I agree that more tutorials would be good.