anyone see this yet ...drum machine !!!

[pdxindy]

This will be as much a game changer, if not more, as pianoteq was to virtual keys.

Hahaha, that would be cool, but let's keep our feet on the ground

Just to notice - pianoteq is amazing if you've never tried it.

Piano is also relatively easy to model in comparison to various other instruments. Percussion is harder for example, due to more variables.

[david.beholder]

Piano is also relatively easy to model in comparison to various other instruments. Percussion is harder for example, due to more variables.

I think this statement is totally wrong
I think piano is way harder to model than percussion because it's simple more complex i.e. contains more sounding components.

[NerdMcBoon]

I'm really curious how this will turn out.

If it really "just" tries to emulate acoustic drums as accurately as possible, it will be really hard to pull of, as it would need to sound as least as realistic as good samples, which is really, really hard to do. Cymbals are especially hard to get right.

All the cool little things that are possible with modeling like this, like sympathetic resonance, bleed and stuff like that, don't really mean anything if the sounds don't sound really, really realistic to begin with, at least not if its really just aimed at people who want ultimate realism.

I'd be much more interested in "uncanny valley" stuff. Expressive sounds that have physical properties, but don't deny their synthetic origin.

If anyone can pull it of, it's the U-He guys, so I'm definitely very interested!

[Beamboom]

I'm really curious how this will turn out.

If it really "just" tries to emulate acoustic drums as accurately as possible, it will be really hard to pull of, as it would need to sound as least as realistic as good samples, which is really, really hard to do. Cymbals are especially hard to get right.

I believe Urs himself stated in the big drum thread that they will indeed also cover synthesised/digital drums. Anything else would be madness really, coming from our favourite synth mastermind dude

So I am pretty confident that it will not be just aimed at people who want ultimate realism.

Because...

I'd be much more interested in "uncanny valley" stuff. Expressive sounds that have physical properties, but don't deny their synthetic origin.

... that's totally where I'm at too. Not that I mind accoustic drums, not at all (can't wait to create some really organic sounding jazzy rhythms) - but I want both.

This is, hands down, the upcoming U-He product I'm most eager to get my hands on. I'm so dead tired of static samples in my drum patterns, and as a Linux user there's not many alternatives yet.

[NerdMcBoon]

Indeed. I'm quite certain this will be another great product!

If I'm not in the mood for samples, I like to use Bazille for synthetic sounding drums.

Here is an example of Bazille doing some xox-ish drums.

Zebra is already quite good at acoustic sounding drums that don't really sound like any real drums, but it would certainly be very interesting to have a dedicated instrument for that kind of thing

[david.beholder]

Video

[Krakatau]

To add the same comment from instrument's subforum :

this is very, very interesting

I'm wondering at this degree of modeling's accuracy, if it make sense to emulated drum workstations, similar to BFD2, Superior Drummer or Addictive Drums, but only trough physical modeling ?



Could it be in a future accurate enough to emulate one kind of drums or another, as Pianoteq does with different models of piano (Bluether, Yamaha) ?



...or does it currently stays pure utopias ?

i'm confident that we'll reach such accuracies one day or another, the question is :

- how far are we from such achievements, currently ?


____

P.S. as far as i could hear, rattlings (and perhaps leakages) are here already emulated !

[sascha]

I believe that acoustic drums are way harder to model than a piano. Sure, in the piano world you have a user base being very serious about sound, expression, depth and playability. It's probably the instrument most people on the planet are somewhat accustomed to, and it's easy for most people to spot deficiencies right away, even when you're not a player yourself. Drummers can also be very picky, but there's so much stuff out there making noise of any kind that there's no such thing as 'the drum sound'. Unless you play in a classical orchestra, it mostly comes down to taste anyway, or what players or audience are used to. This can be very diverse. For instance, a jazz bass drum (typically closed, reonating and having a rather muffled attack) is pretty much the contrary to a metal bass drum that almost sounds 'electronic'.

Technically, emulating piano is modeling 1D, 2D and 3D objects, where the strings are 1D, basically. Ideally, strings are also 3D (a core and a round-wound layer, thus not only straight transversal and longitudinal movement), but you can get very far with a 1D approach, soundwise. With drums, there's no 1D at all. Except for snare wires, possibly. And you have an insane amount of interaction and very fast resonance buildup. Coupling components is where the system becomes very complex and hard to handle; problem is meeting the stability criteria. (Anyone involved in systems theory can nod his head now...). Additionally, such a complex thing demands heavy calculation. That is the time for putting some 'art' into the equation: find out what's needed and what probably not, cut corners, reduce component interaction, make things easier to calculate but still sound good and provide ways to immerse fully as a player. This is where most academic work I've come across just stops since authors often admitted that it is unrealistic to do in realtime. I see a big part of our work on this thing in finding practical and empirical solutions. Like code, play, listen, trash, try other approaches, code again, play again, listen again...
And the fact that we're modeling an entire kit doesn't make it any easier...

Maybe we end up with something that reminds people of pianotec v1; far from perfect but already having plenty of potential for future stuff.

[glokraw]

I'm so dead tired of static samples in my drum patterns, and as a Linux user there's not many alternatives yet.

Do you have a midi keyboard or drumpads with quality velocity sensitivity?

Hydrogen user sets velocity as user enter notes into a pattern,
and hydrogen allows editing velocity on the pattern notes as desired.
If a given Hydrogen kit does not use all 32 sample slots, you can modify
the most important samples, add them to fill the kit, and integrate them in patterns.
For an 8 drum kit, each piece could have four variations.
I like timemachine for recording samples in a string in one file,
import that into audacity, duplicate the timemachine file
in other tracks, create the variations, then working last sample to first,
export-name the sample to a folder, (then trim it off, repeat
for each sample) naming for clarity like
snareDLY snareLP snareHP snareCOMP yada yada,
then use them to replace samples in a kit loaded in Hydrogen,
(a one-sample-at-a-time process) export the modified kit
using a new name.
And kits don't have to be all (or any) drums.
Cheers

[Arrested Developer]

I believe that acoustic drums are way harder to model than a piano.

The problem with the piano is that it's an extremely complex sound per se.
Change the sound a tiny bit and it's no longer realistic at all.
Maybe there will be some better physical modeling pianos in ten years, however, I'm not sure.
However, good luck with modeling the drums!
As written previously, it would be interesting to have the opportunity to "build" or to morph different aspects of the sound...

[Beamboom]

Any news on this project? Rumours? Heresay? Speculations? Anything?

[Urs]

Any news on this project? Rumours? Heresay? Speculations? Anything?

Existing physical modelling tech from this is supposed to go into Zebra3. As for a drum modelling, we're currently seeking funding to research and implement a novel approach.

[EvilDragon]

I believe that acoustic drums are way harder to model than a piano.

It took Pianoteq about 8+ years of active development to reach the point where it can pass as a believable piano (Pianoteq v5). So no, I don't think drums are way harder to model than a piano. Everything about piano is 3D, including strings, harp, soundboard... (that's the only proper way to model strings, in fact, 1D model of a string will not sound realistic enough (see early Pianoteq versions) and will also impact how other elements of piano sound - like sympathetic resonance, how it responds to different hammer hardness, and so on). You have over 200 strings in a piano that interact all the time. How is that NOT complex? It's also a percussive instrument, let's rememeber - attack of piano is extremely complex and not easy to model (in fact it only started sounding right with later versions of Pianoteq 5 and Grotrian and Steinway B models which were released in 2016, so 10 years after the first Pianoteq version!).

I can understand cymbals being hard to model since they're basically semi-organized chaos most of the time, though...

[david.beholder]

So no, I don't think drums are way harder to model than a piano. Everything about piano is 3D, including strings, harp, soundboard... (that's the only proper way to model strings, in fact, 1D model of a string will not sound realistic enough (see early Pianoteq versions) and will also impact how other elements of piano sound - like sympathetic resonance, how it responds to different hammer hardness, and so on).

And now let's think about modeling of snare drum you have 2 membranes (they harder to model than strings) plus things like towel, sticky tape, tambourine; then rim; locks; tighteners; body; may be strings under, may be pillow in it and many other techniques to get certain snare sound - and it's all resonating. Then your snare is influencing every piece of your kit and it's being captured by various different mics: direct, overhead, room. IMO proper overhead sound that's what is distinguishing bad and good drum recording.

I'm yet to heard any drum synth sounding close to real drum set.

Sample libraries are emulating real drums as good as real piano.

[Beamboom]

As for a drum modelling, we're currently seeking funding to research and implement a novel approach.

Thanks for the reply! Ok so Zebra 3 will be a beast, I get it.
But a unique approach to a drum machine? Oh man colour me excited!