Virtual piano comparison

dore_m I guess what I'm saying is that any MIDI file worth rendering to test a VI is going to have alot of dynamics

We are on the same page on this….

dore_m I can make a mockup very quickly, within about 20-30 minutes nowadays. But that is one layer,

I think here we are going on a different route. Even if quoting you, I was really responding to

Deleted It would have been nice having a real acoustic recording thrown into the mix, while recording the virtual instruments through microphones as well. Would be amusing to see how many people dislike an original and go for the digital fake instead.

Unless we find somebody with a good player piano and good microphones, we can't do that. However we can do "the other way around", namely, take both the Audio recordings of the acoustic pianos played in the competition mentioned below, and the corresponding MIDI files, trying to match the original acoustic with various virtual instruments (on PTQ's website they already do that for some of their demos). It'll be interesting to see what really people prefer. Probably @dore_m is the only one with good enough microphones to do exactly what @Deleted asked, namely "acoustically" record the virtual instruments too (to add some room ambience and background noise to not make it obvious?) -- but perhaps even without that step it could be an interesting exercise, and we can start/stop the audio recording taking most noises out (after the piano playing starts, most noises are inaudible).

Anybody has the spare time to do that? Note that in some years they have only the audio, and in some others they have only the MIDI, but luckily there are plenty for which they have both.

Here are some links for a couple of years in which they had both MIDI and Audio:

https://www.piano-e-competition.com/videos_2018.asp
https://www.piano-e-competition.com/midi_2018.asp

https://www.piano-e-competition.com/videos_2017.asp
https://www.piano-e-competition.com/midi_2017.asp

@RIP Wow, so many variables to control. I suppose if you just use a VST preset and render the midi from the competition link, you might get something decently comparable. But if a producer took the midi, chose the right piano library, and then "produced" it (with EQ, reverb, etc), he could make it sound better than the video.

I enjoyed both ptq and ravenscroft.

@HZPiano Modern U render please? 🙂

Hello,

navindra, 😀

I thought of contributing a render of NI Noire as well. (And didn't think of Modern U, how silly is that…😲).

But… I have no clue how to do such a thing (nor do I have any software installed other than standalone players). I never record, edit, save MIDI or anything like that. I just play Modern U and, since last December's sales, have my toe in the water with a few others (Noire, Hammond B5v3).

This is an interesting thread, though 🙂.

Cheers and happy comparisons,

HZ

I have Modern U. I’ll do an export later.

to @TomLC and others that may be interested, these were the settings used in the Ravenscroft render. The velocity curve was set at linear.

Did the upload limit go back down to 2048 KB? Tried to upload a couple of mp3s and hit that error.

navindra yes, I brought it down to 2MB. What’s the size of the files you were trying to upload? I can increase it to a bit more but I’d prefer not much more.

It’s 3MB now.

David B I'm still curious to understand the source of the shrillness you mention, so here are some experimental renders. Are either or both of these an improvement for you? (I feel like an optometrist.)

Hi navindra : Obviously I cannot speak for @David B so I don't really know what does he refer by "shrillness", but I know what I call "Pianoteq shrillness" and I complain from, and maybe it is the same thing: it is an overdone resonance that is really loud and almost continuous in volume and pitch. In my opinion, Pianoteq' resonance sounds realistic and very complex if you listen to separate notes, but when the notes are played close together, the resonances piggy-back on each other and becomes a continuous "shrill/ringness" that buries the individual notes. In most pieces, that continuous metallic ring gives me ear-fatigue very rapidly. Your tamale-a example is a very good example of that, while the resonance in the tamale-z sounds more tamed but it is still tiring. In the tamale-a listen to the section that starts at the 55 seconds: the fundamental of the notes are buried beneath their own "halo"... The problem is that if you reduce too much that resonance, then pianoteq sound becomes very thin. To me, Pianoteq resonance is at the same time the best and the worst feature of Pianoteq...
I am always hopeful that I can find or modify a preset that tames the resonance and keeps the spatial spread using reverb. But I was never capable of achieving that in the demos (I confess to give up easily).

The Modern U with my custom preset.

God Bless,
David

David B Thanks for the efforts, David. Sadly it seems like the Modern U disappoints here.

The sound is strangely thin and artificial. Weirdly, even though you cranked up the sympathetic resonance, it seems to be quite lacking.

VSL Bösendorfer 280VC with my custom preset. I've been enjoying this 280VC lately. I like the intimate rendering in the smaller stage B.

God Bless,
David

David B I'm not a huge fan of 280VC demos on the Internet but the ones you exported here sound pretty good. It's still rather bright but it suits the music I played. I think I prefer the darker and more distant character of Garritan CFX but for more jazzy type of music (what I actually played here), a closer, brighter and more open character as the 280VC is also excellent 👍🏻

CyberGene It's still rather bright but it suits the music I played.

Yeah, I've been working to try and tame some of (what Bösendorfer calls) the "sparkle" in the 280VC. Particularly in the upper register. I started by reducing the gain in the 1300 - 8000 frequency range on all the mics I'm using.

and then working on some very specific frequencies that I found too shimmery.

and of course adjusting the velocity curve editor for my N1X.

I attenuated the volume slightly up to 80 and then began a gradual increase so that my N1X can hit 115, which is the highest I want to go with this sample library. I think the highest velocities in your pieces were between 104-110.

It's still a brighter piano but a nice contrast to my VSL Bösenderfer Imperial that is darker and warmer. It's nice to have variety and switch things up occasionally. My ears get board hearing the same thing everyday. I'm happy with this current 280VC preset.

God Bless,
David

CyberGene I don't know if you've seen this, but this guy crushes it on the 280VC. Evidently, according to VSL, we are hearing VSL's 280VC sample library being controlled via midi by this real 280VC/disklavier.

I like my preset with the cut highs a bit better, but this guy's playing is off the hook. 😀 Listen to his various arrangements in this video.

God Bless,
David

vagfilm In my opinion, Pianoteq' resonance sounds realistic and very complex if you listen to separate notes, but when the notes are played close together, the resonances piggy-back on each other and becomes a continuous "shrill/ringness" that buries the individual notes. In most pieces, that continuous metallic ring gives me ear-fatigue very rapidly. Your tamale-a example is a very good example of that, while the resonance in the tamale-z sounds more tamed but it is still tiring. In the tamale-a listen to the section that starts at the 55 seconds: the fundamental of the notes are buried beneath their own "halo"... The problem is that if you reduce too much that resonance, then pianoteq sound becomes very thin. To me, Pianoteq resonance is at the same time the best and the worst feature of Pianoteq...

Of course, the one you like best is Pianoteq with the resonance dialed down. I did leave a little bit in. The one you didn't like isn't Pianoteq at all but a sampled Steinway.

My goal with the experiment was to find out whether the objection was to the basic sound/tone of the Steinway or to the resonance. Particularly, because the top samplers like VSL seem to do such a bad job at resonance that everyone sets it to negative infinity anyway.

To me, having no resonance modeling sounds completely lifeless... I'm used to gobs of resonance from my acoustic.

cg1.mp3

cg2.mp3

God Bless,
David