My issues with Pianoteq. Am I alone? (Page 1 of 2)

Joshua, I agree with you on both of your points. 

First, Pianoteq does _not_ currently have the capacity, in _any_ model, to produce truly thunderous piano tones (which, of course, are accompanied by a rather percussive "striking" of the strings, especially in the bass).  I _do_ miss this feature, but I only rarely employed it when playing on my real piano (a Kawai upright with knock-you-on-your-@$$ power).  I certainly _would_ urge Modartt to explore this fervently, if they haven't done so already (or aren't currently doing it anyway ;^).  The sample libraries fail at this, too, which I would expect (simply capturing these tones via microphone just isn't going to cut it, because they will just sound _too_ bombastic), unless VSL's new mega-enormous-ridiculous "Imperial Grand" instrument says otherwise.  (I don't know and don't really want to pay the $$$ to try it out and be disappointed -- Pianoteq, despite its various shortcomings, does _not_ disappoint me!)

Second, I've always found the uppermost registers (i.e., the top two octaves) of _any_ piano to be rather annoying.  Modartt is definitely capable, I would imagine, of mellowing the sound in this area, but they probably haven't pursued it, as the reality I've experienced on acoustic pianos mirrors the models in Pianoteq fairly well.   I _do_ have to play very softly on a real piano to get delicate tones from those higher notes.

This leads me to my own complaint:  I think the presets really have _too_ much "hammer noise" -- 1.00 is too loud, IMHumO.  I constantly find myself creating new presets where I lower the hammer noise to 0.20 (and greatly reduce or eliminate the action altogether).  Even with these modifications, the hammer noise is _especially_ noticeable in the upper registers -- but this _does_ mirror the real pianos I've played!

(I have a few recordings of Soler's keyboard sonatas, played by Gilbert Rowland on Naxos, where the mechanical action of the harpsichord was _fully_ captured.  This is extremely unfortunate, because in addition to the beautiful music performance, I can hear the constant CLACKETY-CLACKETY-CLACK of the instrument.  Absolutely awful!  I find myself recalling these recordings when I hear the action noises in Pianoteq, though the impression isn't nearly so ruinous [unless I make it so ;^]. )

---"1) COMPLETE LACK OF FORTISSIMO SOUNDS. "---

Well, I agree that this was a anoying problem on Pianoteq, but version 3 implemented FF strikes for C3, and now in Yamaha C3 Add-On.

I was reporting the lack of fortissimos and fortississimos long ago.  There is a feeling that there is still no FFF, but only up to FF with the C3 and Yamaha add-on.  But The problem of FFF is that most digital controllers and digital pianos have no dynamic range up to suck strong hand strike.
Their maximun velocity for hand strik happnes a bit too short, and the FFF strike would be produced too easilly. The velocity curve can help avoid "way easy FFF strike", but will not correct the limited dynamic range of a controller.

This fine changes it's thanks a new modern specially developed tool  acquire specially to help sample the Fortissimos sounds:

:-)

Yamaha C5 it's now the most natural of all pianoteq's pianos, cause it's like Erard in good natural timbre, and have the fortissimos of C3. pianoteq major model.

I've spoke exactly of this previously.  But if I may indulge a repeat.  This is why my solution for getting fff is to have an additional pedal available at my feet.  This pedal would allow for fff at the higher velocities by switching in a MIDI controllable effect (could also be some model control). With the pedal down, fff would "fade in" for the higher velocities.  This is needed for the exact reason Beto says -the highest velocities are reached too easily on a MIDI controller.  Adjusting a curve on your controller is not going to make up for this.  I do hope the Pianoteqnicians are not such purists that they refuse to use any technologies other than modeling, because, frankly, modeling is a waste of CPU power in many cases.  EI:  for creating the fff woody, crunchy reverberation, delays and intermodulation distortion delivered by the physical construction of an acoustic piano (and all the metal wind-up toys sitting on top).  Also for creating that high metallic tines sound left -particularly high notes- when the damper comes off quickly -FM would be better for that, IMHO.  But the "Fortissimos Pedal" controlling an internal post-effect & model parameters would really work ...and also be an editable MIDI parameter when sequencing.  The only work around I've found is to use a "send" from your recording application under MIDI control out of your audio interface to a chain of stomp boxes and then back in to record on another channel.
"And that's the way it was, July 18, 2009..."

Have you tried to change the velocity sensibility from your controller (or digital piano) to Heavy?

In my Roland F-90, I get more dynamic  range if I set it to heavy. When it's standart (normal velocity) the maximun velocity happens with a bit low hand strike.  In Heavy it get fine, and I adjust the velocity sensibility of the sound response using pianoteq velocity graphic.

I recomend eveyone to try that, to check if can get more dynamic range from your controller.  Many people maybe did not notice cause they adjust sensibility just in the velocity graphic of pianoteq.

maybe we could compare mp3s?

http://www.forum-pianoteq.com/uploads.p...issimo.fxp

I like pianos to have a very *low* dynamic range, and let the variation in *timbre* provide the expression.  A softly played piano, when listened to loudly, sounds very beautiful to me, and then if we can also hear the loud timbres without blowing our heads off, the overall effect is wonderful. I think a good example of this is the piano in "Beautiful In My Eyes",  by Joshua Kadison.  I don't see any reason why this shouldn't be useful for just about any kind of music - even classical.

I'm not a recording engineer, but I know that one way this is achieved on a real piano is to use a compressor to process the signal. However, I also know that this can have adverse side effects.  Of course, we don't have to use a compressor in Pianoteq so we have the best of both worlds. 

I find that if I feel the need to increase the dynamics, it *can* mean that the piano sound simply isn't expressive enough.

Greg.

As it is simple enough to chain any number of effects to the output of PT with other software, in my humble opinion there is no need for Modartt to expend any extra effort in this direction.

One practical problem that a lot of users will encounter in the real world is that the majority of commercially available MIDI controller keyboards will produce a Note On Velocity of 127 very easily, if you just "play fairly hard"... you do not have to slam so hard on them so as to almost break the piano!

So, if Pianoteq made v=127 sound like "you are about to break the piano" by default, then most users would get this sound in normal playing!

I've made the suggestion that Pianoteq perhaps has some kind of "keyboard calibration wizard" which can produced a calibration curve to suit your controller keyboard

The new YC7 addon definitely has a good bit of roar in the low end.  Something I haven't heard in the other models.  I can't say that the mid range also creates "thunderous" sounds, but the bass octaves definitely have it.  Definitely worth a try.

Joshua, what do you think about this recording?
http://www.forum-pianoteq.com/uploads.p...g-test.mp3
The main things that I did were:
- Increasing the forte hammer hardness to 1.80
- modifying the volume balance between bass and treble + EQ
- modifying the midi file (issued from the 3.0 beta tests), as I noticed that the velocities of some notes of the last chord were around 100 or even lower, as well as some of the last octaves in the bass. As Pianoteq reacts differently to the 127 velocities, it is normal that velocity 100 does not provide the strongest fortissimo. As you pointed, things are different on a digital pianos where several velocities share the same sample. If you wish, I can send you the modified midi file and the fxp that I used.

Joshua, I just sent you the files. I'm not sure to understand your question (what means unprocessed?)… but in short, we built a model based on the equations of the mechanics, we solve the model with computing techniques issued from applied mathematics, and that gives you the sound. I don't know if you call this processed or unprocessed, but there is no other audio processing, apart the limiter that can be turned off. As you know, the EQ works in pre-processing mode, which means on the model itself.

Btw, the short recording in my previous post was not done with the YC5 but with the C3 solo recording. When I mentioned avoiding breaking the strings, I was referring to the recording that served for the YC5.

Jake, I uploaded it in the files section.

I wasn't sure about to post here or create a new thread but what I'm writing are my personal consideration about PQ's sound and thus could be fitted here.. It's since my first approach to PT2 demo that I've been feeling what I'm going to say. Maybe my thoughts could be of some help. I hope to explain myself quite clearly; it could be easier writing in italian

First of all, besides what I'm going to write, I want to carify that I'm very happy with ptq in general and it's my best sostitution to the acoustic thing and I appreciate the work done by developers and support found here in general.
Second I want to forget for a while about all differences we could have in listening experience due to different speakers, different way to play, different dynamic settings, different master keyboards, different room, different tastes and alike.
What seems to be common in this software (and in many romplers too) is the lack of presence in the midrange area of the keyboard (around C3). I mean a dull nasal sound that seems to come from neighbour's room. In last version is much less noticeable but still present.
Probably harmonics in that range of notes fall in the human most sensitive area of hearing and that could be why higher and lower notes seem much better (IMHO). The point is to understand why this happens..
As a sound engineer I know the difficulty to get a good overall sound miking a piano and because of it's big emission area every mic placing has good and bad consequences. BUT in reality even if placing is very important it's not as sensible as in PTQ (the minimum changes in placing results in too big sound change); moreover if I bypass mic amping in ptq using mono or stereo output mode the problem gets much worse.. In this direction miking simulation helps me to solve the problem but also lend me to a possible conclusion:
probably the sound generation equation is near perfect and doesn't need a '10000' core processor to get a convincing sound with relative resonance etc etc.. what is really difficult to calculate is the spatial diffusion model of the sound. In reality we have a complex combination of numerous sound sources quite distant one from another and this distance is as big as the piano is. So when we play a note the sound generating from a string(s) can be considerate a different sound source than the sound propagating from soundboard. Then (even if we do not consider the room and its acoustic) we must add other wood and metallic resonances coming from a couple of cube meters area minimum.. Thus the piano can be considered a 'multisource' sound generator. So even if the string, action, soundboard, resonator models are mathematically exact in the real world they combine together generating a big and gorgeous sound that we all know; but constrained in a stereo system phase cancellation is so present that the result is a tiny, suffocated  instrument (and obviously is true for bad acoustic recording too)..
That's the core of my humbling and, in two words, what I wanted to point out is that piano is a very complex instrument to record and reproduce because it is very big and can't be considered a single sound source instrument like the most of other musical ones. Here the real space plays the biggest role in sound quality.
I'm quite sure that developers know this but I could also be completely wrong in my considerations so.. just my two or three cents..

uh, thank you for your comment.
And keep up the good work, I have faith in this software.

IMHO *part* of the problem might also be one of frequency response. It simply sounds to me like Pianoteq isn't producing the very high frequencies that give a real piano the "shine", especially in the attacks.

As I've said before - I don't perceive this limitation in the CP80. I fell in love with the CP80 emulation immediately.

Greg.

Jake,
I'm sorry - I haven't tried it yet - will do so tomorrow.

But, I have experimented with EQ, and I haven't yet been able to produce the highs that I think are needed.  If I go too far, the sound becomes too harsh - not bright in the subtle way I want, and even then, I still don't hear the very high frequencies.   To me it sounds like the CP80 emulation has a special "metallic shine" engine which needs to be somehow grafted into the acoustic models.

Greg.

Have suggested more than once that Pianoteq let us run the model at 96k  
I am intrigued to know what would happen with that extra octave of sound in there. (And yes, I know my cochlea can't hear it by itself - but I suspect it can hear the interactions it has with the octave underneath)

(It's not an fxp. Just the preset called C3 Binaural. But you may want to experiment with very slightly increasing the room size in the Reverb pane and the freq cut-off in the Design pane. Somehow the Binaural presets, with the head over to the right of the piano, bring out the nice wash of high freqs that comes from the string sustain.)

Crikey! Thankyou Jake - yes, I think Pianoteq *is* already producing the very high metallic overtones for the acoustics.  The main thing was to increase cutoff all the way to max. Once I did that, I adjusted the other settings to get a somewhat playable sound.  I was very wrong indeed - I now have an extremely metallic rock C3 sound which I like a lot.   Btw, I did feel like I had to increase it a lot in order to get the very high frequencies. The next thing is to control all this power to make a more subtle realistic instrument. Btw, I have experimented with cutoff and Q etc in the past, but never to this extreme, and not really at all with C3 until now.

It's possible that the *density* of these high frequencies isn't as high as I'm hearing in some of my piano samples, though. By that,  I mean that the piano samples almost have a "hiss" sound, due to the sheer amount of high frequency energy. In Pianoteq, I think I can hear those same high frequencies now, but they're sort of spaced out more in frequency, if that makes any sense.

Here's a very metallic (and still very expressive IMHO) C3 preset:
http://www.forum-pianoteq.com/uploads.p...c64b83c828

Greg.
p.s Jake - sorry - I see that you meant an *existing* canned preset - I was in a rush.

Note that a piano doesn't have to be bright for me to enjoy it - far from it. This brightness thing is just one issue.  A very mellow piano recording that I love is "Imagine" (John Lennon), although that's helped by that great stereo ping pong reverb.

Greg.

I've noticed the large change with small movements of the mics (or the binaural head) - the change seems unrealistically large for the distance.

If/when you try the FXP, you'll probably think that the overtones don't decay in a natural way, but I still think it's a usable sound. (for a hyper-metallic kind of sound, anyway)  With the soundboard opened up all the way like that, Q-factor doesn't seem to have any effect (which is understandable), so I adjusted impedance and direct-sound-duration.  I suspect that I should have experimented some more with the cutoff at a lower frequency.

Yes, I do feel like I'm reaching a point of diminishing returns regarding adjustments. 

Just btw, when I first tried Pianoteq, I couldn't imagine ever wanting more in a piano sound. I'd never heard a digital piano "sing" like that, and I thought "they've done it".  But now, I do want more detail, more fidelity.

I've just listened to another V-Piano demo which an owner posted  over at Harmony-Central.  During that demo, the sound ranged from sounding worse than my first electronic piano (a Roland EP30 which I presume used a bank of oscillators and filters), to sounding very, very good. Due to the bad part sounding as bad as it does, I think Pianoteq is in front, and that's pretty amazing considering Roland's resources.  Interesting times indeed.

Greg.

Tweak all you want, it still sounds like they're holding back. The rock piano is a noticeable improvement, and they should at least upgrade the C and M pianos accordingly. But the soft/medium sounds in the middle register are still a bit off. Maybe they need to work on the equations so that the solutions pump out richer more complex tones.

Which brings me to another suggestion I made during beta testing... they should offer presets for the harmonics. They'd know far better than 99% of us how to adjust the harmonics to achieve certain sounds. We could still make our own adjustments.

Try playing harder than ff and you'll quickly exhaust their possibilities.