Recently I had a Jazz pianist who knew nothing about sampled VST pianos or modeled pianos play on my setup. The gentleman was not PC nor tech literate, strictly acoustic. He said that all the pianos (Ivory II, Pianoteq D4, Bluthner etc) sounded just fine to him. He said they were excellent. He was stunned by exploding notes (velocity jumps) on the Galaxy Vienna Grand but still was quite fine with all the offerings, expressed no preferences, and really didn't care which one I selected.

I did try to explain about velocity layers and modeling. He nodded and got back to making music.

How good is Pianoteq? I think Pianoteq is great. To my friend the pianist, it's just another excellent piano (his words).

SteveLy wrote:

Best I could describe it is that there is a hint of a sound as if a metal bar was being struck rather than a string.

Hmm.. I don't notice such things I talked about other thing: it's like flatness, weak brightness, harshness... ohh... I can't describe it even in my native language, and moreover can't in English, and I think I'm the only man who notice this, so I won't even try

Between F3 and C5 is where I notice this the most on the D4 Daily preset (Pianoteq 5 Stage version, default model settings), F3 to F5 on the K2. The higher notes sound excellent on both, and the low notes as I expect (I'm a newbie to pianos though). I'm not sure how to describe it either, but maybe something like a bandstop filter is applied, like the sound is missing some brightness, maybe? (edit: sorry, I meant bandstop filter .. or, inverse bandpass ... where a certain frequency range is filtered, leaving most of the spectrum untouched)

Great discussion guys, thanks.

No need to post any concrete MP3, all demos from official sites are very good (for my taste). Or, for example, this amazing composition of Shaun Choo (here comparsion of Ravenscroft 275 and True Keys American/German):

https://www.youtube.com/watch?v=hkGpSO_FNqc

And this he played on acoustic grand:

https://www.youtube.com/watch?v=3UMbBdy42q0

I had no idea this discussion thread would become so long.

Best I could describe it is that there is a hint of a sound as if a metal bar was being struck rather than a string.

Hmm.. I don't notice such things I talked about other thing: it's like flatness, weak brightness, harshness... ohh... I can't describe it even in my native language, and moreover can't in English, and I think I'm the only man who notice this, so I won't even try

Undoubtedly "B" is Pianoteq. This is what I talked about, it has characteristic pianoteq-ish sound, slightly synthetic and your sampled piano for me sounds more natural.

I think I know what you mean. It's a subtle thing but occasionally it jumps out at you. Best I could describe it is that there is a hint of a sound as if a metal bar was being struck rather than a string. Because it's fairly subtle it must be difficult to quantify. A large part of ongoing development of Pianoteq I imagine would be quantifying these differences, which can then be used to refine the physical/computational model.

If you listen to or play earlier instruments from the Legacy collection the improvements from one version to the next are very apparent. No doubt that having to do the calculations on-the-fly in real time is a big bottleneck as well and increasing processor speeds are helping things along too. But most of it seems to be about more refined modelling and more clever programming because the latest and probably the most refined instrument to date, the Model B, is very much playable on a 10+ y.o. single core Dell D410 (with a few compromises of course).

I'm fairly confident that we can extrapolate from the improvements to date and expect similar for at the very least a couple of major Pianoteq versions into the future. But even now, when playing the instrument, the subtle differences between the modelled and acoustic pianos are very easily forgotten because it's a living/breathing instrument; which is my way of repeating of what you (Ross) were saying earlier about playability being superior to sampled DPs. (In fact, Ptq v2 or v3, I forget which one I first tried, was a joy to play, despite being more synthetic sounding and all.)

Even the best acoustic pianos have their quirks and unpleasant characteristics as well (little noises, buzzes, funny resonances etc). But one forgives and forgets and just plays because it feels good: you and the instrument have a "thing" going on.

Sampled pianos just don't cut the mustard for me. There is no "thing" going on. At least not up to the best $US5000 or so can buy these days, which is probably about the most expensive I've played. But I don't see the point of paying even that much when I could get a pretty damn good Yamaha U1 for that kind of money.

from audience point of view Pianoteq is weak (IMHO)

Thankfully it's just your opinion, the audience I played Pianoteq to was just as amazed as myself in the quality of sound Pianoteq puts out, and realism of it.

@Beto-Music, I hope to do another comparison as you suggested sometime, trying to keep as many parameters as possible nearly the same (such as reverb, equalization, volume and other characteristics). I believe that if I had instead used one of PianoTeq's "Sound Recording" presets (instead of "Stereophonic" output), the sound might have had some nice additional character. I should have done that anyway, since the microphones used to sample pianos always imbue the samples with their own particular characteristics.

I personally prefer the modeled sounds, which I find complex, dynamic and rich, to sampled sounds, however pristine they may be, which seem relatively flat and sterile to me, like a dead acoustic piano. :-)

Yes, Pianoteq produce too ideal sound, too flat and even. Condition slider helps not so much, "bad condition" Pianoteq just produce ideal "bad condition" sound. Another side of Pianoteq's sound that upset my ears is too mellow nature of it, it really hard to get bright but not harsh sound in treble; if you just increase hammers hardness you get more harsh sound, but sampled pianos can produce very warm non-harsh bright sound. This is just my IMHO. I personally hardly believe that anybody consider "B" example more realistic than "A", probably it's just my ears.

But I want say again that Pianoteq has no alternative for playbility, for response. No velocity layers (ok, for humans 127 is infinity), very realistic pedal (even top sampled pianos has troubles in repedalling and non-realistic damping, partial pedaling), great dynamics. So, from pianist point of view Pianoteq is f*cking unbelievable realistic, from audience point of view Pianoteq is weak (IMHO).

OK, sorry. If anybody will give such a comparison later, let me know please.

Stephen, what about try to find a sampler os s Steinway B, to make a guess test with the latest pianoteq model B ?
It should be more fair, since the model B represent the lastest achievement by Modartt.

Well, the tone may be not 99,9999999% perfect yet, but it's near 99%   
Enough for most people.  :-)

The remaining few people with critcs will help to improve it even futher.

I compare sampled and modelled like compare Stop Motion and CGI.
Stop motion, the best ones, have very detailed "puppy" model, with good texture and looked organic, but in motion it Always looked like stop motion, no trail of moviment duging esposure.  CGI started and looked somewhat artificial, for many years, until it improve to standarts that make it look very natural today. As result stop motion it's no longer used.

Undoubtedly "B" is Pianoteq. This is what I talked about, it has characteristic pianoteq-ish sound, slightly synthetic and your sampled piano for me sounds more natural.

Well, you are correct, file B is PianoTeq, although I did mention that in my last post above.

But about the sound, I disagree. I personally prefer the modeled sounds, which I find complex, dynamic and rich, whereas sampled sounds, however pristine they may be, often seem relatively flat and sterile to me, like a dead acoustic piano. :-) Pianos are sampled with microphones, and in the file comparison above I did not include PianoTeq's microphone modeling in file B, choosing the raw stereophonic PianoTeq model instead, which I thought might better approximate the stereo output of my Yamaha CP5 keyboard (I may be wrong about that, however).

But perhaps I'm just expressing a personal preference, and everyone has their own preferences for their own reasons.

Finally, I uploaded those two audio files to the PianoTeq Forum, along with the original MIDI file I used to make the comparison, and all three of those files appear below.

Undoubtedly "B" is Pianoteq. This is what I talked about, it has characteristic pianoteq-ish sound, slightly synthetic and your sampled piano for me sounds more natural.

Thanks for doing that again so carefully. The most obvious difference I hear between these latest two is that the stereo field is wider for A than for B.

Darn. That was an oversight (one of several I had to correct). OK, I've uploaded a new B MP3 file (under the same filename, so it should replace the one in the previous post) with a (hopefully) wider stereo field. I was trying to keep most parameters the same or similar (velocity curve (flat), reverb room size (fairly small), equalization (flat), volume normalization the same (-2 dBfs), no additional sound processing for one file that the other didn't have, everything as similar and as simple as possible, etc.).

Thanks, Steve.

I like both the sampled and the modeled Yamaha grand pianos, but I like PianoTeq's modeled pianos better than any of the sampled or Roland "SuperNatural" pianos I have in my external hardware. At any rate, I thought that others might find this comparison interesting.

Added later-- For the curious, in both cases above, MP3 A is the sampled Yamaha grand piano in my Yamaha CP5 MIDI keyboard, and MP3 B is PianoTeq's modeled Yamaha grand piano (the YC5). In the case of the second comparison pair of MP3 files, PianoTeq's output in file B was changed from "Sound Recording" to "Stereophonic" to try to more closely match the stereo audio output of the CP5, although I am more fond of the sound variations imparted by PianoTeq's various microphone types/brands and placements in its "Sound Recording" models.

As a bonus, the same MIDI file using the PianoTeq Steinway Model B, unmodified Close Mic preset--

http://www.forum-pianoteq.com/uploads.p...se-Mic.mp3

---

Now this test, trasnposed to the world of pianoteq tone , becomes harder and very welcome.