BDMO,GS,QLP,Ivory,Pianoteq,Galaxy vs recording artist perfomance/tone

[Ernest]

The demos can be found at http://www.proaudiovault.com/bdmo-comparison.htm

This is a comparison of two passages one pianissimo and one forte from Chopin's Scherzi in Bb Minor No. 2 Op 3 as performed by 7 well know recording artists Alexander Paley, Artur Pizarro, Artur Rubenstein, Mikael Pletnev, Yundi Li, Jennifer Lim and Martin Kasik.
The same passages were also played back on 6 digital piano sampled libraries The Bluthner Digital Model One, Garritan's Steinway, Quantum Leap Pianos, Ivory (Bosendorfer and German D), Pianoteq, and Galaxy (Bosendorfer & Steinway).

These demos were mean to illustrate how close sonically a sampled piano library approaches the sonic qualities of real world acoustic recordings. To accurately hear the differences in each of these recordings one should listen on high quality headphones if you only have speakers then listen to the recording fairly loud so that you can hear the subtle tonal variations. I also find the various interpretations of the pianists playing the same passages fascinating - timing is everything -:)

Suggestions on how to listen
First listen to either the pianissimo or forte acoustic recording in its entirety - this helps you build a sonic reference in your mind. Then listen to the matching sampled piano library recording in its entirety.

Next consider use a sound editor - if you do not have one download Audacity (a freeware sound editor application that is available for either the Mac and PC). Listen to each recording comparing individual sampled piano libraries with each other or comparing an individual acoustic recording with a particular piano library. Sometimes it makes sense to listen to 2 or 3 acoustic recordings first before listening to a single sampled piano library - this helps in focusing the ear to subtle differences between the acoustic versus the sampled domains. Fast switch between various recordings only confuses the ear and makes it very difficult to hear exactly what is going on in an individual recording when compared to another group of recordings.

What to listen for in the pianissimo passages
In the acoustic recordings the overall tone heard is soft full or round tone with a prominent low end in other words it should have a tranquil or very calm tone. The piano timbre should not should not have a thin, cool or nasal quality. The attacks of the piano should also not be prominent instead they should blend in with the ringing or singing element of each piano note. Prominent piano attack in soft passages is a common problem heard in many sampled libraries one can clearly hear in real world piano recordings that this should not be the case if tonal accuracy is a goal.

What to listen for in the forte passages
In the acoustic recordings the bass part theme is very strong in most of the interpretations even thought there are fast runs in the right hand they do not dominate the overall sound instead they sonically sit further back in the overall sound. In many of the sampled piano libraries the right hand runs are often much more prominent than they should be when one references it to these acoustic recording.

Ernest Cholakis
Numerical Sound/ProAudioVault
www.numericalsound.com
www.proaudiovault.com

[Jake Johnson]

Were the libraries\vsti's played live, or was a midi file used?

[Ernest]

A midi file was used for the sampled piano libraries. The Garritan Steinway and Quantum Leap Pianos recordings were taken from there respective web sites - but I would assume that they were also rendered midi files.

I would think that some overdubs/punch ins were done by the recording artist. To record a live perfect take as challenging as this Scherzo would be difficult.

Ernest

[scope4live]

It's an excellent sounding instrument, the variable sustain is also a great feature. I would buy this in a heartbeat and would pay twice the asking price is you could create a sostenuto set with it.

I am not familiar with all of the Kontackt scripting features, etc. But surely there should be a way to provide that overlooked feature.

Is there a remote chance an upgrade in the future could have this capability?

I am sure the decision to go w/o that was a wise choice as many composers here would simply use multiple tracks to achieve some of the basic uses. But there are still some live performers who would die for that feature. Dave Gruisin's recordings in the Movie Score The Firm are fine examples of sos work.

You Piano Library is excellent, you should be very proud. ;)

[Ernest]

JimmyV

Thanks. Regarding the sostenuto feature. It is impossible to create an accurate sostenuto with any kind of scripting regardless of the Kontakt scripts that claim to add this feature.

You cannot recycle any part of a piano's tone and have it sound like a realistic sostenuto. The main reason is that there are many non-harmonic components are present that also change depending on what other pedal up notes are played. The only way to do it is to have a different impulse for every piano note. The set of impulses would change depending on what note is selected a very big job that would max out any 4, 8 or 16 core computer.

On this forum there has been much discussion of this feature. Respectfully I think that it is too subtle of an effect and it will easily be masked as soon as another instruments are mixed in. We included three other areas for convolution that will clearly stand out in a mix - timbral, reverberation and sustain impulses.

In fact there were timbral impulses in the BDMO that were inspired by the Dave Grusin's piano sound in The Firm as well as several of the acoustic recording Alexander Paley (Bluthner), Artur Pizarro (Bluthner), and Mikael Pletnev (Steinway). These impulses total change the overall sound especially in the bass region of the piano all are easily discernible with even casual listening.

One area that has not been discussed is dynamic range and the problem that samplers have in reproducing it - that is a far bigger problem. We addressed this in the BDMO - a real challenge because Kontact 2 and 3 does not have a proper dynamic range engine (should be linear decibel adjustable for each note). Some information on this subject can be found at http://www.numericalsound.com/sample-velocity.html

Note that the only sampler that can reproduce dynamic range correctly is Digidesign's Structure.


Ernest

[electone2007]

It is impossible to create an accurate sostenuto with any kind of scripting regardless of the Kontakt scripts that claim to add this feature.

You cannot recycle any part of a piano's tone and have it sound like a realistic sostenuto. The main reason is that there are many non-harmonic components are present that also change depending on what other pedal up notes are played. The only way to do it is to have a different impulse for every piano note. The set of impulses would change depending on what note is selected a very big job that would max out any 4, 8 or 16 core computer.


Ernest,

Are you talking about SOSTENUTO here (middle pedal)? Or am I the one who's confused?

[electone2007]

It's an excellent sounding instrument, the variable sustain is also a great feature.


BDMO has variable sustain? Is that the same as proportional sustain? Ernest, please?

[Ernest]

Electrotone2007

Sorry about the confusion it was not intentional :-) Yes I am speaking about the middle pedal I assumed that was what JimmyV was speaking about since he knew that BDMO has variable pedaling.

Yes the BDMO has proportional sustain.

Ernest

[electone2007]

Hi, Ernest!

I am all the more confused.

Can you please define 1) Sostenuto, 2) variable sustain, and 3)proportional pedaling?

I am already wary about developers saying a product has a feature when in reality it doesn't. Sometimes, unintentionally, the definitions of terms may vary. You say BDMO has proportional pedaling? And sostenuto is impossible to script in Kontakt? I just want to clarify.

If another thread is necessary for this I may start one.

Jun

[scope4live]

Thanks for your reply that many choose to never answer. That's a pity as I really miss using one onstage. I haven't used a real Piano in our show for a long time. Since we had hardware sequencers automating lights and vocal effects, I use to have a QX-1 cover the sostenuto drones, but even w/ backing tracks I could never get the colors of the 3-450KHz range of notes to sound passable. The low Bosendorfer type Drones sounded really nice though.

Perhaps the PhysMod/DSP approach would be a good alternative in the future. A group of 4 x ADI 400MHz Black Fins doing real time calculations.

Your variable pedaling is a big step forward towards these real world issues, and I commend you on that. :heart:

Do you plan on looking into the MIDI CC #88 Prefix of 2007 for future dynamics? It seems this would really help w/ current controllers that seem hardly capable of triggering Velocity Layers w/ any real accuracy. It might be possible to use the new spec on current libraries, but I am really just a performer who lacks the technological knowledge of developing.

Brotha' Man Electone2007,

It's variable pedaling, and proportional sustain. It sounds like the BDMO can emulate the way a sustain pedal on a real Piano functions. This is not easy, but is really great for adding DSR transients, meaning decay, sustain, and release, to phrases. which sound much better than the full on/off options that are most common w/ Romplers and Libraries.

The Sostenuto Pedal acts exactly as a sustain pedal would, but only sustains the notes played while it is depressed. For example, in Ragtime there are several peices that have chords being used in the mid section, while walking up or down w/ both hands above and below those notes sustaining, while the walking notes have zero sustain, a luxury modern composers using Romplers don't really need. And for good reason. It would require extreme amounts of resources as Ernest explained above.

The soft pedal ( Una Corde ) on a real Piano shifts the hammers over from striking 3 strings, to only striking a single string, which is another feature that there are work arounds for. The lowest 2 Velocity Layers can handle that sound by the controller sending out very low ( 001-035 ) values. Those lowest Layers could have very quiet samples, or actually have samples of the real soft pedal being used. I doubt many could tell the difference. However on an Oberheim MC3000 a different programmable velocity curve can be applied to the soft pedal, which helps when trying to reach the lowest velocity layers.

Thanks For Sharing Your Bluthner DMO With Us.

[electone2007]

Hi, Jimmy!

Sorry you had to go to great lengths to explain sostenuto, una corda, and proportional pedaling. I appreciate it.

I am a pianist, and I know how these work ON A REAL PIANO. I just want the developers' definitions on how they did it ON THEIR SOFTWARE PIANO.

Case in point: If you read Ernest's definition of sostenuto in his post above, it differs from what I think sostenuto is ON A REAL PIANO, and that is:

sustaining only the notes which are depressed at the time the middle pedal is depressed.

He seemed to be talking more of sympathetic resonance rather than sostenuto.

Also, regarding proportional pedaling and variable sustain, I want Ernest to confirm whether or not BDMO has this feature, and HOW IT IS SUPPOSED TO WORK in BDMO.

I am very interested to know because I just recently bought a software piano for the reason that it was advertised to feature sostenuto and proportional sustain pedaling. Sadly, I discovered that both features do not work. So, I'm still looking for one that works. Maybe BDMO?

[Ernest]

electrone2007

I was speaking about sostenuto - it like sustain has sympathetic resonances between the string where the middle pedal is user selective whereas the regular sustain pedal is global.

The BDMO does have variable sustain or proportional pedaling. There are 116 different values for this feature. If you want you can email me a midi file with proportional pedaling in the midi file and I will render it for you.

Ernest

[scope4live]

Please share the rendering.

Meanwhile I will research the way that Kawai, Yamaha, and Kurzweil achieve their sostenuto pedal. Surely they don't have Dual 6 Core Xeons. Probably Linux based custom O.S. and a couple of ADI 400MHz Black Fins.

Let's pray for the future of software, as I refuse to buy another keyboard to lug around. Even though I don't use it on many renditions, it is sorely missed and irreplacable when it's needed.

Thanks To All, Nice Thread.

[electone2007]

Thank you for response, Ernest. I can't make the MIDI file at the moment, but it's really good to know that BDMO has proportional pedaling.

[electone2007]

Thank you for esponse, Ernest. I can't make the MIDI file at the moment, but it's good to know that BDMO indeed has proportional pedaling.

[kensuguro]

just checked the pianissimo demo.. why does it sound like the bdmo part is continuosly pedal down? Something wrong with the MIDI file?

[Ernest]

kensuguro

My mistake - the variable pedal in the BDMO does not trigger until a value of 10 in CC#64 is reached. This way if a user lightly rest their foot on the pedal it will not trigger the sustain. The sequence did not have all the pedal off values below 10 so the sustain was always active. Thanks for pointing this out. I changed it and updated the demo.

Ernest

[Raindog]

kensuguro

My mistake - the variable pedal in the BDMO does not trigger until a value of 10 in CC#64 is reached. This way if a user lightly rest their foot on the pedal it will not trigger the sustain. The sequence did not have all the pedal off values below 10 so the sustain was always active. Thanks for pointing this out. I changed it and updated the demo.

Ernest

Hi Ernest,
actually that was my point when I was mailing to you that the Bluethner seemed to have too much sustain pedal level. I didn´t realise though that the mistake was in a false trigger. Glad it could be sorted out s the Bluethner can even sound much better-
best regards
Raindog