Something New from Spectrasonics

[ElmerJFudd]

Omnisphere is a beast too, it all depends on how much you have turned on. Or rather how much stuff you need to turn off to get it to run well in live play (it's another thing all together to simply mix with already recorded MIDI where you can raise your buffer way way up).

 

"One of the most difficult questions to answer is how much CPU power Omnisphere uses. The reason it's so difficult to answer is because STEAM was designed as a very flexible and open system, which allows you to freely add FX and more editing power based on the needs of what you want to create. This means that Omnisphere is capable of using all available CPU power and then some...

 

Here's a summary of example things to try:

Turn down polyphony.

Turn down amplitude envelope release.

Look for effects that can be turned off.

Look for oscillator section features that can be turned off.

In the case of Unison, try turning down the depth.

Metal Pipe and AllPass filter types use more CPU than other filter types.

Use one filter instead of 2.

Glitching in the absence of a CPU overload might indicate a streaming underrun. If that's the case, try turning up the Pre-load memory size..."

 

More on how to turn stuff off to get it to run well on your computer system.

https://support.spectrasonics.net/manual/Omnisphere/concepts/page02.html

[Geoff Grace]

Helpful info, Elmer!

 

Best,

 

Geoff

[MAJUSCULE]

Greg Phillinganes wailing on Keyscape:

 

[video:youtube]

 

Best,

 

Geoff

 

Just bumping this video because Greg Phillinganes is an absolute treasure.

[Bobby Simons]

What got me about that great lengthy Keyscapes rollout was how one guy after the next was absolutely killin it on the Roland A-88 controller - to the point where I was thinking, that would surely work for me. All sorts of controlled, nuanced playing - as much an ad for the board as for the software to my ears.

 

And so Roland promptly drops it from their line.

[Keepitsimple]

I know Eric Persing said that they hand-tweak the velocity for each controller in their LA Headquaters before making a preset for it, but i generally find that making a curved ramp will get the C7 to generate all those nuances in the videos. Seems to work for me but i guess could never know how "close" i got until i buy a controller they support in their presets.

[drawback]

Anyone have a list of the controller velocity presets?

[Keepitsimple]

They support 68 controllers/keyboards at the moment (hopefully there's more coming).

 

I would take a screenshot for you but it would be a very long process to do because the presets are categorized as mouseover folders based on brands.

[drawback]

SL88 Studio, by any chance?

[Keepitsimple]

SL88 Studio, by any chance?

No Studiologic presets.

[Reezekeys]

What got me about that great lengthy Keyscapes rollout was how one guy after the next was absolutely killin it on the Roland A-88 controller - to the point where I was thinking, that would surely work for me. All sorts of controlled, nuanced playing - as much an ad for the board as for the software to my ears.

No chance it's not the board, but the people playing it?

 

"It's the artist, not the brush."

 

Maybe some boards have a more even response to playing dynamics in terms of the velocities they output. That helps, I'm sure. Some keyboards, you play a note with the same force and get different velocities each time. If those variations cross sample layer boundaries things might sound a little odd. I sure haven't researched which controllers are more accurate in this regard they might all suck at this, or they could all be fairly accurate. I know how things feel when I use my controller (having used the same keybed for 13 years), and I'm cool with it. If my playing sounds bad, it's me and only me that's responsible not my instrument!

[Bobby Simons]

No chance it's not the board, but the people playing it?

Obviously. My point was only that the Roland was the chosen instrument. Surely they would use whatever would best show off their remarkable software.

[Reezekeys]

No chance it's not the board, but the people playing it?

Obviously. My point was only that the Roland was the chosen instrument. Surely they would use whatever would best show off their remarkable software.

Sure, but how do we know the Roland wasn't just what they happened to have at their studio? No doubt it's a fine piece, but I suspect those players would have sounded just as good playing on any decent 88-key controller (if there are any left, lol!).

 

If we come to find out that they went through a few 88-key controllers and noticed that this particular Roland model made a significant difference in how Keyscape felt like to play or how it made their plugin "sound" vs any other controller, I would definitely like to know this. I'm just a little skeptical about it, that's all.

[o0Ampy0o]

@Reezekeys, are you by any chance a scientist, in the medical field or an engineer by day?

[Reezekeys]

Given the economic conditions for most full-time musicians of which I am one these days, I can only wish I was in one of those professions! Nope, tinkling the [plastic] ivories is all I do...

[Markay]

Drifting OT but Pianoteq has had a controller learning module for some time. Switch learn on, play loud, soft and inbetween and it then recommends a tailored velocity curve to suit the incoming midi, reflecting the way you play that controller. This can then be saved as a preset for that controller, or modified by the user.

 

MainStage has similar DIY option adjustable patch by patch.

 

[hardware]

Pretty sure EPers gets free Roland gear, no need to have a superior controller like a Physis K4/5.

[Reezekeys]

I've said it a few times here, velocity curves can only go so far - they only adjust the response within the 1 - 127 range. They can't account for the hardware translating your physical playing force to these numbers. That's why I like my Roland controller's "light", "medium" and "heavy" settings that go along with its four velocity curves, for a possible 12 settings.

[Markay]

Not having owned Roland controller I can't comment on how it plays, but it seems to me it is still limited to sending 1 to 127 so it has 3 primary velocity curves which are fine tuned by one of the selected 4 options.

 

 

[Reezekeys]

Not having owned Roland controller I can't comment on how it plays, but it seems to me it is still limited to sending 1 to 127 so it has 3 primary velocity curves which are fine tuned by one of the selected 4 options.

You might have misunderstood me, or I'm misunderstanding you! All midi controllers send 1 to 127 for velocity values - that's the full range of velocities, not a "limitation." My Roland doesn't have three velocity curves, it has three dynamics settings, which they call "keyboard sensitivity": "light", "medium" and "heavy", which biases the velocity calculations by your actual dynamics, i.e., how hard or soft you tend to play. So, for example, if you press a key with what you consider "pianissimo" dynamics, it might send velocity = 40 when set to "light", 30 on "medium" and 20 on "heavy." Combine this with four velocity curves (linear, and three more logarithmic types) and you get your twelve options. Here's the page from the manual that better illustrates this:

 

[Markay]

Not having owned Roland controller I can't comment on how it plays, but it seems to me it is still limited to sending 1 to 127 so it has 3 primary velocity curves which are fine tuned by one of the selected 4 options.

You might have misunderstood me, or I'm misunderstanding you! All midi controllers send 1 to 127 for velocity values - that's the full range of velocities, not a "limitation." My Roland doesn't have three velocity curves, it has three dynamics settings, which they call "keyboard sensitivity": "light", "medium" and "heavy", which biases the velocity calculations by your actual dynamics, i.e., how hard or soft you tend to play. So, for example, if you press a key with what you consider "pianissimo" dynamics, it might send velocity = 40 when set to "light", 30 on "medium" and 20 on "heavy." Combine this with four velocity curves (linear, and three more logarithmic types) and you get your twelve options. Here's the page from the manual that better illustrates this:

 

 

No misunderstanding, we are both on the same page. The difference is that the Pianoteq tool reads "how you play" and suggests a velocity curve setting based on how it read the incoming values when played according to the on screen prompts.

 

It is not limited to the variations available on the controller, in Rolands case, the 12 you mention. The Pianoteq recommended curve could be, by reference the 45 degree line, an infinite number of combinations of for example, concave, convex, concave from 1 to 64 and convex from 65 to 127 or vice versa.

 

You can do the same in MainStage, manually by trial and error. I mentioned this in the first place in the context of Spectrasonics mapping for controllers, which would be really helpful in assigning knobs, sliders etc. but I doubt that velocity mapping, if indeed that is what is included, would be of any real value, being a one size fits all basis, and therefore hit or miss for each individual player.

[Theo Verelst]

It's like this is a general problem, when compared with a full velocity sensitive instrument, with no quantizing of the velocities like a piano, there's a lot that can go wrong with the various steps of quantizing, transforming, and interpreting the key velocity data.

 

When a digital keyboard is played, usually there's a timing measurement from the top key contact first opening to the bottom key contact first closes (there's bouncing, too, like the elaborate "clicks" you can hear on an electronic organ). That timing measurement is usually a software time counting routine, which will for instance measure the key depress time by measuring in microseconds. How many microseconds are being distinguished by the measurements smallest quantity determines how accurate the processor reading keyboard data will know the actual key press velocity.

 

From there, you can say: I want to create a certain sensitivity curve, sort of like in the above manual page, for instance a exponential representation of the velocity. Or more emphasis on the middle of the curve where most piano players will statistically centre around while using the keyboard. For an internal tone engine, you might want to take the whole of the available accuracy, and the most detailed available absolute timing accuracy, to drive your tone generator notes with all the available information. So if you have say a thousand velocity values, and you use a quadric curve you use a number of bits that correspond with a thousand squared range, and feed the tone generator with that data.

 

Going over MIDI, you need always to reduce the accuracy to a quantization of 7 bits = 127 values + note off, unless you use the special high accuracy protocol addition some machines and software can use. Also, there's a timing inaccuracy involved. When you map to a curve before quantizing the velocity values to 127 discrete steps, you might have the useful range of velocities in only a small range of values, like half the time you're playing as loud as between velocity 64 and 84 (20 values of 127 possibilities, which is about 17 % or so), after a mapping, they could end up between midi velocities 64 and 74 (compressed to middle velocities mapping), which is only half as accurate.

 

Mapping the velocities from incoming (very quantized) velocity values to the instrument tone generators velocity sensitivity can again happen in various ways, in a pro isntrument, hopefully with accurate mapping math, so the amount of information present in the midi signal is put to good use. Information about what the velocities and the exact note timing was could be guessed, but is essentially lost, so often an instrument will show a certain character with external Midi controller, based on what it will guess. If the instrument is neutral, it's just harder to control accurately over traditional midi.

 

Optimally, it would be best to map the midi velocity values from the players practical playing such that the important ranges receive a bit of accuracy in the Midi quantized translation, a mapping than should then be undone in the midi module connected, and succeeded by a mapping to the desired velocity sensitivity curve. I might have been the first one to implement a midi velocity mapping curve in the end of the 80s as part of a commercial Atari ST Midi program, but that aside.

 

Using the more accurate midi velocity data protocol addition over Usb midi should give better result, but there can still be timing inaccuracies, and people get used to certain idiosyncrasies of the traditional Midi chain.

 

T.

[Reezekeys]

The point I was trying to make is that midi velocity values originate in the controller, before they're modified by a velocity curve they're determined by the scanning routine that measures the time between two contacts closing. So, if you're a heavy handed player and the minimum velocities you generate are e.g. ~30, and you're hitting the max of 127 before you've really begun to "dig in", a velocity curve is not going to allow you to experience the glory of a sampled instrument with many layers that evolve timbrally through all 127 values not unless you change how you physically approach playing the keyboard. The Roland's keyboard sensitivity settings are a way to better match the transmitted velocities values to the player's actual playing dynamics. BTW I'm not implying that Roland keyboards are special in having this feature, I'm pretty sure I've seen it on other brands too. I would be very hesitant to buy another controller that didn't have this setting.

 

Having said this, of course I believe that velocity curves are very useful, and I'm not surprised to see them implemented in software instruments at all. The Scarbee rhodes has a page with a visual that lights up different velocity ranges as you play (and switch between different velocity curve settings), so you can easily see the relationship between how hard you press the keys and the sample layer triggered. My Bidule hosting software goes completely over the top with a velocity mapping module that allows you to map any individual incoming velocity value to any other.

[Theo Verelst]

Arriving at a playable instrument behavio(u)r can be quite the challenge digitally, not just because of controller accurcy, but because certain waves that strongly influence acoustics and human perception are hard to get right digitally, let alone control accurately!

 

T

 

[Markay]

The point I was trying to make is that midi velocity values originate in the controller, before they're modified by a velocity curve they're determined by the scanning routine that measures the time between two contacts closing. So, if you're a heavy handed player and the minimum velocities you generate are e.g. ~30, and you're hitting the max of 127 before you've really begun to "dig in", a velocity curve is not going to allow you to experience the glory of a sampled instrument with many layers that evolve timbrally through all 127 values not unless you change how you physically approach playing the keyboard. The Roland's keyboard sensitivity settings are a way to better match the transmitted velocities values to the player's actual playing dynamics. BTW I'm not implying that Roland keyboards are special in having this feature, I'm pretty sure I've seen it on other brands too. I would be very hesitant to buy another controller that didn't have this setting.

 

Having said this, of course I believe that velocity curves are very useful, and I'm not surprised to see them implemented in software instruments at all. The Scarbee rhodes has a page with a visual that lights up different velocity ranges as you play (and switch between different velocity curve settings), so you can easily see the relationship between how hard you press the keys and the sample layer triggered. My Bidule hosting software goes completely over the top with a velocity mapping module that allows you to map any individual incoming velocity value to any other.

 

Well I think we will have to agree to disagree, I dont see any difference between the velocity mapping adjustements that can be done by software on the receiving end of midi values between 1 and 127 compared to your proposition that it can only be implemented in the controller by scanning the rate at which the values between 1 to 127 are generated, modifying them, and then sending the modified 1 to 127 values.

 

You get the results you need allowing the controller to adjust velocity to your playing style, I get the results I need using the software on the receiving end to modify velocity.

 

That leaves Spectronics users with the option of selecting one of the 88 mapped controllers with a tailored velocity, if it exists at all given it is happening at the software end, that works for them with Omnisphere or Keyscape.

[PianoMan51]

What Reese is describing is real and quite simple.

 

Heres a question: How hard do I need to hit the key to generate a 127?

This is before any midi curve modification.

1fps? 10?

 

The answer is that its completely up to the the keyboard vendor.

 

Some vendors are hip to this and provide user selectable choices. My VAX 77 and CP-4 both have this.

 

A heavy setting means that it takes a higher key velocity to reach 127. The heaviest setting is usually best for me when I play high end piano VSTs like Ravenscroft. On a 9 grand it takes a lot of weight to get the loudest and brightest tone, which is often shrill. I only want to trigger those samples when applying the same weight to my keyboard controller.

 

I hope this helps. Having those programmable keyboard touch settings is a significant improvement for playability.

[Markay]

This assumption is only true if the time between the "note on" message and "note off" message and the timing of the transition in values between these two events, 1 to 127, are always sent by the controller as a standard flat line, irrespective of the speed at which the key is struck, and cannot be sensed, and therefore varied by the receiving software.

 

I don't believe this to be case.

[Reezekeys]

What Reese is describing is real and quite simple.

 

Heres a question: How hard do I need to hit the key to generate a 127?

This is before any midi curve modification.

1fps? 10?

 

The answer is that its completely up to the the keyboard vendor.

 

Some vendors are hip to this and provide user selectable choices. My VAX 77 and CP-4 both have this.

 

A heavy setting means that it takes a higher key velocity to reach 127. The heaviest setting is usually best for me when I play high end piano VSTs like Ravenscroft. On a 9 grand it takes a lot of weight to get the loudest and brightest tone, which is often shrill. I only want to trigger those samples when applying the same weight to my keyboard controller.

 

I hope this helps. Having those programmable keyboard touch settings is a significant improvement for playability.

Thank you for making clear what I was trying to say, in many fewer words!

[Theo Verelst]

You could program the midi module/software to play note velocity 127 when the input velocity value is 30, provided it has such a very proressive curve built in, but that would make the whole range of velocities that the module responds to come from playing velocity 1 through 30. So the next softer velocity, 29, which is like 29/30=~3% lower would most likely be played at 97% of 127 (full value). So the resulting velocity steps are a lot courser and playing becomes a lot cruder in terms of velocity accuracy.

 

T

[Reezekeys]

Exactly. Thanks Theo. Another reason why a velocity curve is an imperfect fix for a keyboard with mismatched finger-to-velocity values.