Interesting new modelled piano plugin

[burningbusch]

Originally posted by keyman_sam:

I wish i could play like that Miles Black guy. And thats an improvisation? wow.

Go to http://www.pgmusic.com

 

Miles has a Master Jazz Piano course available at that site. Very nice stuff.

 

Busch.

[Jazzed]

This is actually amazing. I'll set up the demo this tommorow; see how it plays controlled by a PX-310. Exciting.

 

Now all I'd like is a Mr. Ray 73 AudioUnit (which I'd pay for) and my physical modelling needs would be totally satisfied.

[Jazzwee]

I've just finished doing a quick study of the demo. I don't have a strong opinion yet.

 

When I first loaded it, it didn't have that character of the timbre that excited me. But definitely it was very easy to control so the dynamics and expression is a clear winner.

 

After I tested some of the other presets of grands, I found some to my liking and it actually sounds pretty good.

 

Then I did a comparison against my S90ES and I couldn't make up my mind. This, I don't think will be a hardware killer. There is no doubt that the S90ES sounds good. This sounds good too but there isn't sufficient difference to say 100% that it kills the S90ES, because it does not.

 

However, considering that this is a softsynth, and I'm playing this on an EMU-Xboard, it is pretty amazing and playable. It does consume a lot of CPU however. If I use a lot of polyphony, I get clicks on my 2.16GHZ Core Duo.

 

I have a feeling the the character/timbre issue is something that they can control. It just may be that the sample used in the S90ES is a particularly impressive piano.

 

Note however that although this is modeled, it still is not a real piano and I can still tell the difference. One of my tests is the legato action and the effect of the release of the key on a real piano (discussed extensively in another thread). It does not behave like a real piano when releasing the key. This is a limitation of the keyboard/key action that cannot be solved without redesign. There's no measurement of the release that can be sent as a MIDI signal.

 

I have not tried Ivory but in the absence of my S90ES, this is definitely something to consider when played by a good weighted controller (my Radio Shack/Casio CDP-100).

 

I will keep testing this to firm up my opinion. This is just my preliminary gut reaction.

[Les Mizzell]

In comparing it against the RD700SX piano, the Pianoteq sounds more like there's a real piano in the room.

Just my opinion, but between this and Synful, we may be witnessing the beginning of the death of sample libraries for certain instrments. Sure, Synful and Pianoteq ain't perfect, but they're getting there and both will so some things that just can't be done with samples.

 

I'm already using Synful for around 85% of my orchestral strings, winds and brass. Some of my sample library is now collecting dust.

[Les Mizzell]

Crap - how do you DELETE a post you made?

[Jazzwee]

You can't delete but you can edit.

[Fogwall]

Hi everyone,

 

I'm Niclas Fogwall, responsible for sales and support of Modartt Pianoteq.

 

Very nice to read your posts, we always listen to our users and take into account all constructive remarks that can be useful for our further development of Pianoteq.

 

Originally posted by Jazzwee:

It does consume a lot of CPU however. If I use a lot of polyphony, I get clicks on my 2.16GHZ Core Duo.

This does not sound normal. Can you please provide us more information about your system? I suggest that you use the support form on our site and we can investigate further.

 

Regards,

Niclas Fogwall

Sales & support

Pianoteq

www.pianoteq.com

[Jazzwee]

Nothing unusual. I have a Dell Latitude D620 2.16GHZ Core Duo, 2 Gig Ram, 7200 RPM Hard Drive. Audio Interface is an EMU 1616.

 

I don't get an issue with normal playing, and I do have it set at a latency of 10 seconds so maybe that's the issue. CPU goes to about 35% when I start hearing the clicks at this latency. It only comes up when I'm trying to sound 20-30 notes multiple times. Obviously not an issue for normal use.

 

I will test further. Thanks for responding here Niclas. That is very encouraging.

 

My comments are preliminary. I have to spend a lot more time with it to be firm in my opinions. For one, I still need to test it with a fully weighted controller to see how that reacts.

 

But I am certainly a potential customer.

[marino]

Originally posted by Jazzwee:

There's no measurement of the release that can be sent as a MIDI signal.

I'm not sure if you mean that *your* controller doesn't send it, perhaps - but such measurement has been part of the MIDI spec almost from the beginning. It's called Release Velocity, and I don't think anybody has ever used it on a digital/sampled piano yet. One possible reason for this could be that very few keyboards implement it.

It would certainly add to the realism of a sampled or modeled piano.

[bg]

I listened to most of the mp3 examples. The tone is lovely, kind of distant, a little covered. The pieces chosen and the tone are staid--not a criticsim in itself. I'd like to hear something a little more aggressive, like with some forte staccatos in the bass.

[Jazzwee]

Originally posted by marino:

Originally posted by Jazzwee:

There's no measurement of the release that can be sent as a MIDI signal.

I'm not sure if you mean that *your* controller doesn't send it, perhaps - but such measurement has been part of the MIDI spec almost from the beginning. It's called Release Velocity, and I don't think anybody has ever used it on a digital/sampled piano yet. One possible reason for this could be that very few keyboards implement it.

It would certainly add to the realism of a sampled or modeled piano.

Marino, the problem is the switch off point is too low. A real piano does not stop the sound until you are all almost all the way to the top of the key range. And the sound reduces (from the effect of gradual damper application). All my keyboards and controllers stop the sound at about midpoint and do it abrubtly.

 

And yes, there is no hardware/software utilizing release velocity apparently (that I know off). Maybe it is part of the spec but are keyboard/key action makers actually sending a value for this? To me it requires an application of half damper logic if I release the key very very slowly so I would imagine a combination of velocity + switching off. This would make for a very realistic legato, which should be even more doable on a modeled piano.

[marino]

Good points, Jazzwee. So perhaps the better way to replicate a piano would be a combination of hardware (a keyboard especially designed for that kind of attack and release response) and software designed to take advantage of that. Of course, that would rule out plugins or even modules.

(Hey Pianoteq, what about contacting Casio... )

[burningbusch]

Niclas,

 

Any chance of this running on Receptor? I tried the demo but could not get it to install.

 

I'd be happy to beta this for you.

 

Busch.

[Jazzwee]

Originally posted by marino:

Good points, Jazzwee. So perhaps the better way to replicate a piano would be a combination of hardware (a keyboard especially designed for that kind of attack and release response) and software designed to take advantage of that. Of course, that would rule out plugins or even modules.

(Hey Pianoteq, what about contacting Casio... )

Marino, Until the keyboard action makers start supporting these kinds of details, obviously Pianoteq will have no data to model. What I do wonder about is if it is even discussed. I don't claim any genius to this yet I've never heard of 'release' logic being discussed before (other than by Cydonia and me). Although I have seen references of people talking about "realistic legato".

 

I think that modelling will make this a doable concept. I would think it would be particularly hard to implement in sample based pianos.

 

I have to say that I was surprised about well a modeled piano sounds.

 

In any case, this will be the beginning of a new era. It will be exciting.

[Darren Landrum]

I never thought about how a modelled piano plugin might influence hardware manufacturers. It's entirely possible that some of them may start supporting release velocity if there's something popular out there that will actually use it.

[orangefunk]

Darren, I reckon you ought to think about putting it in a self contained hardware module. The market is crying out for a module. I can't even remember the last one released... perhaps the Kurzweil Micropiano from 1994 or so?

[marino]

Well, release velocity in itself can be useful for a lot of differen things beside realistic note ends on a piano. It can control release of the amp and/or filter envelopes in sustained sounds, it could be changing the release characteristics of percussive sounds (using FM, perhaps), it could add a bit of grit to organ sounds when you release the note abruptly, not to speak of modulating the 'ending noise' on woodwind and brass sounds...

 

Etc. etc.

[Les Mizzell]

I never thought about how a modelled piano plugin might influence hardware manufacturers.

Oh really? And how many folks have been screaming for a good 76 key master keyboard for how long? Seen one yet? Me neither!

 

Well, maybe the CME VX7, but it's not shipping yet so we'll have to wait and see, huh?

[K K]

Originally posted by LesMizzell:

Oh really? And how many folks have been screaming for a good 76 key master keyboard for how long? Seen one yet?

Clickee here

 

O RLY

 

http://media.urbandictionary.com/image/large/orly-37424.jpg

 

http://img.photobucket.com/albums/v348/BPMdotEXE/Random/O_RLY.jpg

[Jazzwee]

Back to serious business (ya right...). Heres a more formal review of the Pianoteq VST.

 

I did a head to head comparison of my Yamaha S90ES against the Pianoteq. Heres my setup. My Radio Shack/Casio CDP-100 is connected to USB using an M-Audio Uno Midi/USB interface. The computer is a Dell Latitude D620 2.16GHZ Core Duo, 2 Gig Ram, 7200 RPM Hard Drive. Audio interface is an EMU 1616 (used here for output only). The VST Host is Ableton Live. Both the Yamaha S90ES and the EMU 1616 outs are fed into a mixer and then go into two monitors (stereo). I also had a headphone monitor handy. Overall, the Pianoteq setup is on one of the highest grade laptops available.

 

Findings: Overall, I am very impressed. Mind you the S90ES was no slouch. But to even discuss this comparison is already telling. Which is the winner? Does software beat hardware? I dont know. Were talking close here.

 

First of all, I tried to pick a preset from Pianoteq that was closest to the S90ES. When I was using headphones, I tended to go to Grand C1 Deep. Now I realize that I was compensating for the headphones. When comparing using a monitor, the closest was Grand C1 Bright. Another test anomaly earlier was that when testing on headphones, the Pianoteq has ZERO reverb. The S90ES has some reverb in the sample, as well as a small reverb in the setup. On headphones, this made the Pianoteq sound stark. Anyway, you have to choose a Reverb setting like Concert Audience and it brought it to life. I tweaked some other things like a little higher Unison value, a little more hardness, and a little EQ biasing the high end. I came really close to the character of the S90ES.

 

There is still something missing that makes the S90ES seem more alive. But this is very subtle now. Comparing individual notes, the S90ES has a slight tinny sound to some notes in the C3-C4 range. This is very realistic sounding but is not necessarily pleasant. The Pianoteq has zero of these tinny sounds. Maybe it makes it more unrealistic but it certainly sounds like a perfect Grand piano though.

 

Would I gig with this setup? No doubt, I would leave my S90ES at home, take my laptop + CDP-100 and do a jazz gig. I would not miss anything there. It is very playable. Now the S90ES is also quite playable so Im saying that it is very close. Not enough for an audience to notice. Just picky musos will detect the difference.

 

Would I record with this setup? Again, there is not enough difference that would make me be adamant about any particular choice.

 

Issues: CPU is something to keep an eye on. As I mentioned in my pre-testing, CPU use is high. Compared to another sample based VST Im using, the CPU use is more intensive. It is probably similar to a sample based VST with lots of reverb and effects on top of the sample. My CPU level was typically around 30-35%, going to over 50% with more notes depressed. It appears to use the CPU in the after effects sounds (resonance). The computations appear to go on after you release the key so if you overlap, there seems to be more calculations being done. However, it does not increase after a certain point. Given my Platform (2.16GHZ Core Duo), This might be an issue for anything slower than my prior laptop, a 1.6GHZ Pentium Centrino M. Some benchmarks compare this CPU as about half the power of my current laptop.

 

Im using a latency on the output of 10 seconds. I did not try to lower the latency any further as it was sufficient for my purposes. I imagine a slower computer might need higher latency to make this work. The documentation does say that a computer no older than 2 years would be necessary. It does not appear to use a lot of ram and certainly disk space is not an issue as the 8 Meg footprint is a measly number to fit in 2 Gig of Ram.

 

I wish I had Ivory to compare against. Maybe someone can compare Ivory to the S90ES so at least theres some benchmark.

 

Conclusion: As a software based solution, this is definitely recommended. In some ways it is a little better than hardware, in some ways it is not. Minor differences. Perhaps this can be tweaked using the parameters and someone could make it even closer. Or someone could compare against another piece of hardware like an RD700SX in great detail.

[bg]

Just FYI, I tried the demo version on an iBook running OS X 10.3 and got an installation error. The requirements say 10.4, so I guess they mean it.

[Phil Aiken]

Originally posted by burningbusch:

Niclas,

 

Any chance of this running on Receptor? I tried the demo but could not get it to install.

 

I'd be happy to beta this for you.

 

Busch.

Busch,

The demo installs just fine on my Receptor. Run the .exe on your PC, and then just copy the resulting folder to Receptor as per usual.

 

A couple of quirks thus far on Receptor....you can't edit via the Gui page or Receptor freezes.

You have to go to the detailed edit window with the sliders. Not a big deal to me...Also, regardless of the state you save a patch in, it comes up with the Reverb on. I assigned my mod wheel to toggle it off.

 

For Receptor users, this is a potentially a great thing, as it requires no samples, so can be used easily in snapshots with other sample intensive programs. It is also simply fun to play and interacts with the player in a way that most sampled pianos don't. It is also very very tweakable - and simply changing a parameter like "Unison String Width" makes the piano sound less perfect and more real. On the other hand, it can't match something like Ivory for pure realism of tone.

Lastly - and this is important to me - it has a mono output button. This button actually works, and makes the piano sound good in mono. My pet peeve with sampled/digital pianos is always how they sound in mono. I have thus far only heard it through ny studio speakers, but play on A/B'ing it with both of my digital pianos, plus Ivory and White Grand Jr. through my stage amp later this week.

I suspect that with a little tweaking, this could be great for rock band use. I also suspect that the developers will continue to make improvements in the tone of the instrument as time goes on.

 

-Phil

[DafDuc]

For recording use, there's always the option of recording with something else, like a soundfont for example, and then rendering with the the pianoteq. I do that kind of thing all the time, as even a 5-7ms delay can throw me off. Well, that and the new Acid Pro 6.0b bug that sometimes causes VSTi notes not to release at record.

 

Daf (love my Acid, but she's a bit flighty)

[Jazzwee]

When I record in Live, I have the ability to move the sample to compensate for latency. So I'm more concerned about latency my ears can detect. That's a good idea with the soundfont, but even with that (I use soundfonts too), I don't usually go below 10 ms (on my old computer).

[burningbusch]

Thanks BeeThree for the tips. I got the demo to load. The next test will be whether the full version will work with any copy protection issues.

 

Busch.

[Mark Schmieder]

Did the Yamaha KX88 support release velocity? My understanding is that it remains the most fully-spec'ed MIDI controller ever made -- even including the new CME VX series, if discounting the motorised sliders.

[midinut]

I downloaded the demo to try it out. It installs just fine but it locks up inside Cubase every time I try it. I'm running an AMD Athlon XP 2500+ (1.82Ghz) with 512Mb RAM. Maybe it doesn't have enough oomph to run it. Anybody else try it in Cubase. I checked the website and didn't see any mention of system requirements other than a "computer that's less than two years old" and checked the troubleshooting FAQ and didn't see any issues. Maybe it just don't like me?

[steadyb]

I like Ahmad Jamal.

[jfhyde]

Did the Yamaha KX88 support release velocity?

Nope, just aftertouch and breath cntrl. I think the old Oberheim matrix had release velocity, the Alesis QS series, not sure what else.

 

I've got pianoteq demo working fine in sonar on my 2 Ghz intel machine. Also, I emailed pianoteq twice now concerning pricing and copyprotection and received a reply within hours. Midinut: you might want to email them with your Cubase problem.

 

John

[cd8]

Originally posted by marino:

Good points, Jazzwee. So perhaps the better way to replicate a piano would be a combination of hardware (a keyboard especially designed for that kind of attack and release response) and software designed to take advantage of that. Of course, that would rule out plugins or even modules.

(Hey Pianoteq, what about contacting Casio... )

Mmmmmm...a 76 key 400 dollar Privia with release response and pianoteq (and Jennifer Love-Hewitt maybe, but I'll have to think that better)

 

Bg, I also tried to install on 10.3 and it didn't work. Then I runned it in a G4 1,6 Ghz and 10.4 and it worked perfectly. I used it normally though and canceled some resonance, I found it too much on the lower keys.