Yes, that's getting closer, but even that isn't the test I'm interested in. The test I want is to give examples of excellent sounding piano recordings (either from real, or sampled pianos)  to Pianoteq experts, and see if they can get Pianoteq to make the same sound, and THEN for me to not be able to tell the difference in a blind test.

Also a good idea - but to get a close match, the same instrument would have to be implemented in PT, so it would have to be a similar sounding Steinway, Bluethner etc.  Even in the Pro version you can't fundamentally change the character of the piano beyond a certain point, there are basic parts of each instrument that are built-in.

(personally I would love to have even deeper access to that stuff in the Pro version, eg. abilty to change hammer sounds, sound board/cabinet character somehow etc).

The perfect test would be if you could find a good recording from the actual piano Modartt used for their model.  Of course Modartt should have these recordings ...

I respect your effort here, but I'm not sure this test is interesting. Well if you asked me if it was pianoteq, I would be inclined to answer "yes" even before listening, because it is obvious to me that one would want to see how many people are fooled by thinking pianoteq is a real piano.

Not necessarily.  I have been critical of PT's sound over the years even though I love it in all other respects.  I've always struggled to get a sound that I find realistic, even with extensive programming.  I could have been pointing out that PT can't yet match the 'real' sound I like.  When I started programming the sound, that's actually what I felt like.

Sure, my test was not trying to be definitive, just a fun experiment in how much I could push the new engine to sound more complex than the patches (imo).  But it was my patch, what sounds realistic to me, someone else may have a different thing they look for or find real (usually related to the real pianos they know and play).

I'd like to see the kind of test you propose.

ReBased wrote:

Thanks for the votes & comments guys.  Final result:

Real:    6    28.57%
PT5:    15    71.43%

That's misleading. Only 1 person voted real. A few said they think it's not Pianoteq, but rather another software piano.

No, there was an actual poll at the top of the page, and this was the result (it was removed when the poll was closed).  People may have voted without commenting.

A more interesting "Pepsi challenge" or rather "Turing test" would be to use two recordings, one of pianoteq, one of a real grand (recording not manipulated to sound "synthetic"), and have people spot the fake...

Yes, that's getting closer, but even that isn't the test I'm interested in. The test I want is to give examples of excellent sounding piano recordings (either from real, or sampled pianos)  to Pianoteq experts, and see if they can get Pianoteq to make the same sound, and THEN for me to not be able to tell the difference in a blind test. 

I'm not interested in Pianoteq sounding the same as a boring real piano.

Greg.

I respect your effort here, but I'm not sure this test is interesting. Well if you asked me if it was pianoteq, I would be inclined to answer "yes" even before listening, because it is obvious to me that one would want to see how many people are fooled by thinking pianoteq is a real piano.

A more interesting "Pepsi challenge" or rather "Turing test" would be to use two recordings, one of pianoteq, one of a real grand (recording not manipulated to sound "synthetic"), and have people spot the fake...

I adore Pianoteq from the day I discovered it, and version 5 is amazing. But I don't think it passes the Pepsi challenge yet.

- Eran.

Thanks for the votes & comments guys.  Final result:

Real:    6    28.57%
PT5:    15    71.43%

That's misleading. Only 1 person voted real. A few said they think it's not Pianoteq, but rather another software piano.

What tipped you off?  Did you just recognize the Bluethner timbre?

Not immediately. I tried playing along with the chords - one of them has a very distinctive ringing harmonic that I heard in the Blüthner model with the same chord.

My midi controller has uneven velocities and in desperate need of repair. So I think the uneven thing is overrated.

.

You get to feel good about yourself .  What tipped you off?  Did you just recognize the Bluethner timbre?

delt wrote:

Just curious, why not?

I will probably use this in a recording and I like to have unique sounds.

ReBased wrote:

Once again just curious, randomization of which parameters? and just curious yet again detuning of which parameters

I randomised most of the parameters, per-note (you need the Pro version for that).  It's really about getting an 'uneven' piano.  Same with detuning.

If you don't have the Pro version, you can also get some of that using the 'Condition' slider just above the virtual keyboard (which I also used).  You can deteriorate the sound with this, and it will detune the keys too.

My midi controller has uneven velocities and in desperate need of repair. So I think the uneven thing is overrated.

Just curious, why not?

I will probably use this in a recording and I like to have unique sounds.

I randomised most of the parameters, per-note (you need the Pro version for that).  It's really about getting an 'uneven' piano.  Same with detuning.

If you don't have the Pro version, you can also get some of that using the 'Condition' slider just above the virtual keyboard (which I also used).  You can deteriorate the sound with this, and it will detune the keys too.

I can't give you the FXP

Just curious, why not?

Once again just curious, randomization of which parameters? and just curious yet again detuning of which parameters

Real:    6    28.57%
PT5:    15    71.43%

Votes: 21

It was ...

... PT5 Bluethner (custom patch).

When I first tried the PT5 trial, I really didn't like the sound.  The new string model took me a day to get used to (I heard all this metallic stuff I hadn't heard before).  I like it now, it's more realistic - but then I still have the problem that PT presets are (for my tastes) a bit 'scientific'/sterile, too accurate, too precise.  I don't think it's an engine problem, but more a different style of piano.  I prefer imperfect, 'character' sounds, and lots of resonance/ringing.

So I spent a day re-programming an old PT4.5 Bluethner patch, trying to get the most organic, warm, complex and (to my preferences) realistic sound I could.  Sort of a bit upright-ish.

I can't give you the FXP, but basically it uses a ton of randomisation + detuning to make the sound more varied, richer, less obviously PT.  It adds very subtle PT effects to mix things up more (there's almost no audible Reverb btw, it's all strings). Mic positions matter too (love the new mics).

And finally I tried to trick you by running the rendered file through analogue console and reel-to-reel tape simulations, to make it sound like it could have been real mics (there's even a little hiss in there).  That's pretty subtle though (easier to hear on the 96/24 source files, or good headphones).

This is the original file (as posted):
http://www.forum-pianoteq.com/uploads.p...tom%29.mp3

And same, but without any randomisations, effects or console/tape sims:
http://www.forum-pianoteq.com/uploads.p...%5D%29.mp3

On the first day I wasn't even sure I would upgrade my PT Pro! (I never thought that would happen!).  Now I've figured out how to get my sound, I'm v. happy I upgraded .