Roland usually have their own real pianos, to get sample to prepare their digital pianos. So we have no Steinway or Bosendorfer in Roland digital pianos.

In the case of V-Piano they keep the same behavior, and avoit emulation of Steinway or Bosendorfer, or any trademark real pianos.

Do you think that modelled technology must avoid trademark pianos?

I don't know of any Roland acoustic grand piano, but I'm willing to learn differently.

Modeled pianos are no different than current DP's - as far as I know, none of the current DP's claim to be modeled after any specific brand of piano.  I've been to a very large Yamaha dealer a number of times, and know one salesperson very well - he's never claimed that the top of the line CLP's sound like the Yamaha concert grand (this may be because it doesn't sound line one either).  LOL.

They may want to avoid using the names of Bosendorfer, Yamaha, Fazioli, Steinway, etc. but it would be difficult for anyone to prove that a modeled piano copied an acoustic piano.

Also, consider this - there has never been a perfect acoustic piano - they are all compromises in design.

Then, there is the manufacturing process which introduces variations.  The soundboard is made of Sitka spruce (or some other spruce) which is a tree.  Wood from trees is never perfect.

My piano-restorer friend says that copper wound strings have variations from string to string which affects the sound - which string is "correct"?

It is impossible to make two identical Steinway pianos.  Or any other brand for that matter.

So if Pianoteq models a German Steinway, which one would they copy?  I think the obvious answer is that they pick the "best" features of all of them.

Can Steinway sue Pianoteq for copying the Steinway sound?  Steinway would have to prove which piano was copied.  And Steinway doesn't make digital pianos.  What would be the basis of their lawsuit?

If I had a voice exactly like one of the Three Tenors, could one of them sue me?  I doubt it.

In the case of V-Piano they keep the same behavior, and avoit emulation of Steinway or Bosendorfer, or any trademark real pianos.

Do you think that modelled technology must avoid trademark pianos?

Previous posters suggest PTQ has problems in the bass register, but I am quite happy with my personal configuration there. The area for improvement I see is in the treble area, where I find the sound "glassy" or "too-digital"; I'd say this is connected with bandwidth/sampling rates. I'd be tempted to say the 44.1Khz is totally inadequate for this purpose, but on the other hand I have heard CD's with lovely treble piano sound and they have no greater bandwidth.

Also of course, we have to consider the full end-to-end audio chain and it's possible that such HF problems are not in PTQ at all but in the follow-on audio devices. In a previous system I had a EMU 0404 PCI card which has amazing specs (192Khz!) but the HF problem was still there.

With my current E8600 dual core CPU is not a problem. So, any tips on optimising treble sounds?

I've been quite happy with the "stock" C2 Chamber sound, but I do two things; I increase the Piano Size to 4.81 metres (about 15'-9"), and increase Sympathetic Resonance to +5dB (controller 17 = 80), or even a bit higher.  But I like Sympathetic Resonance (and really big pianos!!!).

With my limited exposure to maybe six or eight concert grands, the first thing I've always noticed is that the bass is proportionally stronger in a concert grand than the bass in say a six or seven foot grand.  The high treble strings in a concert grand aren't much longer than those of a six foot, but the bass strings sure are, and I can hear it.  In fact, it takes me quite a while to adapt to the stronger bass of the nine footer.

This is my reasoning for increasing the Piano Size (maybe I'm imagining it's different).

I also turn the Pianoteq reverb off and add it later in processing with my EMU 1820M (if I want it - but often I like the sound without any reverb at all).

This may help demonstrate how much we differ in taste from individual to individual.

Also of course, we have to consider the full end-to-end audio chain and it's possible that such HF problems are not in PTQ at all but in the follow-on audio devices. In a previous system I had a EMU 0404 PCI card which has amazing specs (192Khz!) but the HF problem was still there.

With my current E8600 dual core CPU is not a problem. So, any tips on optimising treble sounds?

It might even be possible to have two versions of Pianoteq - one that will run on a laptop and generate "acceptable" sounds, and a more complex one that can generate the sounds far more realistically for rendering a recorded midi file to a wave file.

I wouldn't mind if Pianoteq took longer to render a midi file to a wave, if it produced a much better sound.

I would also prefer that!

I have been waiting since the 80ties to play a piano-type INSTRUMENT on a modern keyboard. And that's what pianoteq is for me an instrument for playing with a piano-type sound.(and for the first reactive computer model really good sounding!) I'm absolutely shure that it will be getting better as long computer power is rising-and also will never reach the "real thing". And it's not important! 24 pictures/sec is not the real thing.96Khz is not the real thing. Maybe we will reach a point where the simulation is making us happy for a while. And when we get used to it, we have learned more about it and we will find another weakness in it. That's why we are here. expand the universe of experience. So this is how we can help the modart team to increase their product. No guarantee that we will ever be satisfied.
Every comment is a point of view. Sometimes it is hard to keep al the emotions out of the discussion.  I would also like to have a model which sounds similar good to a real Bösendorfer/Steinway/... I would dance a real long and happy dance if it would cost only 425€ and maby I would buy 3 licences and 5 computers with all the money I've saved, not buying a Steinway concert grand.
Actually I don't have the money for a Steinway at all. I can't buy 3 Licences and 5 copmuters. I can sit down to my pianoteq and express all my feelings and I know it's all I can ever do with each typ of instrument.
AND i totally agree with all of the comments about the lack of sound when I've played a copmplete mist.(although when I 've played a real Piano)

My conclusion: There is a lack of sound and I love to play that wonderfull instrument.

New version?-download

cheers
Hans

In my academic experience as an analyst and an engineer, my instructors used to say "all models are wrong, but some are useful."  I think that saying applies to Pianoteq as well as any other software modelled instrument.  I agree that any individual note played in Pianoteq does not sound 100% like a real piano and lacks something, which I am not sure how to identify (others in this post seem better able to articulate that better than I can).  However, I love the playability and the resonance capabilities of Pianoteq (which is also been noted in many posts).  The instrument feels alive when I play it.  This is why I use the product - because it acts much like a real piano.  I have not experimented enough with other sampled pianos to know if they do this all that well, but my impression is they have shortcomings in this area.  Other people have other things they look for in a software model and I don't see anything wrong with identifying those things.  It is a first step in making an existing model better.

So, I believe Pianoteq is wrong as a model, but I think it is useful, so I use it a lot . . . I also look forward to any improvements that Modart makes to help make the Pianoteq model less wrong.  I believe they have the potential to make the model better.  I note that no model will ever be perfect.  I think this forum is constructive in that it provides the developers alternate view points and ideas.  That makes it easier, in my opinion, to improve a good product, making it more useful than it is now.  So, I think it should be O.K. to constructively critisize the model.  It is also useful to point out what is right about a model, since the developers will than know what users think is important and not accidently program out what users like.  It is hard to say what is right and wrong about a musical instrument through the imperfect method of language since writing about sound seems kind of like trying to smell something you see or taste something you hear.  However, writing is what we have, so it will have to suffice.

As an engineer also, I tend to agree with your comment about models being "wrong", although perhaps "not quite correct" may be closer.  Having said that, a model is never completely accurate (in structural we often model for testing but realize that scaling isn't always perfect).

I happen to firmly believe that the present model is a compromise, and that's why it could be better than it is.

Why compromise, one might ask, and what would the compromise be?

I'll guess - the mathematician/technician/developer knows what's missing - the criticism has been "thinness" or lacking something that doesn't quite emulate a piano.

I find the treble smooth (a friend says it's 'silky smooth'), but still not quite full enough.

Many on the forum have identified the bass (as has my pianist friend) as being weak.  I won't disagree.

So why hasn't MOdartt fixed it?

As an engineer, I long ago realized that every design is a compromise - whether it's conflicting requirements, or the limitation of available resources that can be thrown at a project.

So why can't Pianoteq be better?  Because not all of us have or can afford the computer that could handle the computations that are required to more accurately emulate the bass of a concert grand piano.

It was said for years that a computer would never reach the ability of a grand master at the game of chess, and for the longest time I watched as Spassky and others beat the computer.

Then a few years ago, the computer programmers "got even" - they knew how to program the software, but the computers available couldn't handle the computations - and their break came with "Big Blue" which has (to my knowledge) beaten the world's best chess masters.

Back to P/T; it's been known for quite a few years that when at least twelve or so partials plus the fundamental are included in a generated piano tone, it is difficult to distinguish from an acoustic tone.  Bass tones are the most complex and require more computations than high notes.

I suspect that if Pianteq was programmed for "Big Blue", the sound would shock us.

It might even be possible to have two versions of Pianoteq - one that will run on a laptop and generate "acceptable" sounds, and a more complex one that can generate the sounds far more realistically for rendering a recorded midi file to a wave file.

I wouldn't mind if Pianoteq took longer to render a midi file to a wave, if it produced a much better sound.

So, I believe Pianoteq is wrong as a model, but I think it is useful, so I use it a lot . . . I also look forward to any improvements that Modart makes to help make the Pianoteq model less wrong.  I believe they have the potential to make the model better.  I note that no model will ever be perfect.  I think this forum is constructive in that it provides the developers alternate view points and ideas.  That makes it easier, in my opinion, to improve a good product, making it more useful than it is now.  So, I think it should be O.K. to constructively critisize the model.  It is also useful to point out what is right about a model, since the developers will than know what users think is important and not accidently program out what users like.  It is hard to say what is right and wrong about a musical instrument through the imperfect method of language since writing about sound seems kind of like trying to smell something you see or taste something you hear.  However, writing is what we have, so it will have to suffice.

I'd love to have the realistic feel that Pianoteq users describe, with the response/playability much better than what's achieved by the sampled pianos I've played.  But when I listen to the demos, I think they don't sound enough like they're played on a real piano.

I've played some (VI) synthesiser sounds and (physically modelled) electric piano sounds that respond wonderfully to touch - more like I expect from a piano, and not like I've had from sampled pianos - so I can appreciate that Pianoteq could offer a similar playing experience.  But, enjoyable though they are to play, those instruments don't, of course, sound like a piano.  Unfortunately, I don't think Pianoteq really does either, yet.

---

I'm sorry to drop in un-invited only to say something critical.  But IMHO, those posters here who find no fault with Pianoteq's sound aren't listening critically enough.  Perhaps the playing experience is so good that it overwhelms the perception of the sound(?) - ie in the way that you can still enjoy good music when its played on a hi-fi system that doesn't fully do the music justice.

And apart of this all... what is perfect? If someone would consider some type of Steinway to be perfect, put a couple of them together and see what differences are between them all... it might show that 'perfect' can be pretty relative.

I am not that much of a 'purist' myself - I want a sound that I like and one I can use in my recordings and live play and is flexible... so to me Pianoteq fulfilled that needs.

cheers
Hans

Although pianoteq is a physically modeled piano, it doesn't mean it is 100% physically correct.  A more complete model is much more computationally intensive, and so I'm assuming pianoteq uses a much, much more simplified model.

Hopefully the model will get updated in future versions...  It just doesn't make any sense to me, how you guys can be all praising and defending output from an approximated physical model, saying it's superior compared to an actual recording.  Just by the way pianoteq is built, that's not one of its strengths.  If you just so happened to like the output from a simplified model?  Good for you.

It's probably the biggest reason holding people back from making pianoteq their main piano.  It's the resulting sound.  But again, these things will probably all change with updates so perhaps it's a little pointless to be poking at it.

I had meant to add that I long for the sort of 'randomness' in key to key characteristics preserved in the fully chromatic-sampled instruments, and which I understand may be made available in a future 'Pro' version of Pianoteq.  That will, I expect, be a minor revelation for this user.

Helpful criticism is good! 

Cheers,

Stephen

This is a good point - I rendered a friend's midi file for him because he wasn't at all happy with the Giga results.

His comments:

1.  Silky smooth in the upper registers.

2.  Bass doesn't have enough "edge" to the sound when played hard.   I won't disagree, because it doesn't - it's too smooth.

3.  His most interesting comment is just what you've pointed out (he is an excellent pianist AND a piano technician that rebuilds pianos); he said:  "It's almost too perfect from note to note - real pianos aren't that way - the copper wound strings in particular can have large tonal variations".

I wouldn't disagree with this third comment, but I replied to him, "if you were designing and building a piano and could control absolutely everything, would you purposely put in some notes that were a bit off?"  He laughed and said "no".

I stand corrected on the number of layers in the larger sample libraries.  Ten or twelve layers?  they must weigh a few kilos.

Should I be holding my breath for someone to direct me to a better sound that comes out of my computer?  PM me; I won't be offended.

A valuable and stimulating post.  I have found some of the same disappointments with sustain characteristics and, of course, sympathetic resonances in otherwise fine-sounding samplesets.  Probably my biggest surprise was with NI's Akoustik Piano library.  But it does have a useful upright.

A technical point: for a long time now, the bigger libraries have sported more than 3 velocity layers (switching or, more, usually cross-fading bewtween layers).  6-8 is typical, occasionally you will see 10 or even 12.  There are at least two Sampletekk instruments that have 16 (pedal up AND pedal down!)  Yes, this makes for massive hard drive usage.

I was always infuriated with those crime shows on TV where the parents are defending their child's innocence 'at any cost', not the brightest ultimate choice and of course morally bankrupt when knowledge of wrongdoing is present; the protective instinct needs to be tempered with higher considerations.  So it is here.  The greatest value of the forum (apart from its undeniable value as a social and creative vehicle) is to establish consensus on performance issues.  No one ever sits at an actual piano and says "it's not real enough"; such a thought would be an absurdity.  But there can be any number of other things to react to, both positive and negative.  Some of Pianoteq's behaviour is startlingly — and seductively — 'real', while other areas generate that 'hmm' reaction many have honestly reported (and I am sure are all too evident to the clever modelling artists at the company).  Overall, wonderful progress in a program of engineering I believe we all wholeheartedly support, and do hope will realise a potential which would have it truly conquer these various frontiers of resistance.