Velocity cross-switching in pno and e.pno patches

[guestuserguestuser.com]

Is this important to anyone else? I'm talking about velocity switched layers in rompler piano and epno patches. For me, I find that the typical velocity-switched Rhodes patch, which switches from a soft-strike sample to a hard-strike sample, is not nearly subtle enough. I find the abrupt change in timbre to be un-musical, difficult to control, and generally annoying.

 

And yet, strangely, I don't see other players complaining about it. I often read articles complaining about stair-stepping as it applies to filters, or velocity levels. Velocity cross-switching is the ultimate stair-stepping, there's only two steps!

 

Typical example: manufacturer (Yamaha, Korg, Kurzweil, Roland, whoever) puts out a digital piano of some kind. Included is a Rhodes patch with two velocity-switched layers, soft-strike and hard-strike. The timbres are completely different. At some specific midi velocity, the hard-strike sample sounds instead of the soft-strike.

 

Perhaps this is so commonplace that people accept it without question as being "the way it is"?

 

It's not like it's the only option. For Rhodes patches, I think velocity cross-fading is a better option. I understand that velocity cross-fading has some problems associated with it that make it unsuitable for acoustic piano patches. But I've had success using velocity cross-fading on Rhodes patches.

 

And I'm not against velocity cross-switching on principle. If it's done subtly, I've got nothing against it. For instance, on my Yamaha P80 digital piano, my understanding is that the acoustic piano patches use 4 velocity-switched layers. If this is true, then they have done an excellent job, because I can't hear any switching going on. I'm only complaining about the velocity cross-switching that is audible and obvious.

 

So, I'd be interested in hearing your views on this subject.

 

 

[This message has been edited by guestuser@guestuser.com (edited 11-11-2000).]

[marino]

Velocity crossfade is a wonderful concept in theory, but assuming that the programmer is able to set the right velocity windows (very difficult), you'll always have points where you are listening to two sounds together, in various doses. This, far from sounding like an interpolation between the two sounds, creates big phase problems, especially in the attack portion of the sound.

So you will never have anything like a "linear" type of response with crossfading; it will sound even more "stepped" than switching. I banged my head against this years ago, but now I'm convinced that multiple switching is the most we can achieve, with samples at least. (I would be very glad to be proven wrong!)

 

I believe that the only way to have a continuous touch response (short from playing an acoustic instrument, obviously) is

when you play a digital model, or a resynthesis of the original. The latter should be interesting: you just input a sample analysis of the model at various dynamic levels, and let the resynthesis engine interpolate among the values, to "fill the blanks", so to speak.

 

So.... WHEN we can have a machine like this, please? It's been a LONG time since the first talks.

(Coming soon: a thread about Harmonic Resynthesis)

 

As usual, I hope my English makes sense.

 

marino

 

[This message has been edited by marino (edited 11-11-2000).]

[guestuserguestuser.com]

Originally posted by marino:

So you will never have anything like a "linear" type of response with crossfading; it will sound even more "stepped" than switching.

 

Excuse me for saying so, but the above quote sounds like complete nonsense. Could you please explain how that could possibly make sense?

 

My personal experience, having programmed some velocity cross-faded Rhodes sounds on my old M1, and more recently, on a Triton, is exactly the opposite of what you say. They didn't sound stepped at all, and the phase problems were minor. They sounded pretty good, and even though not perfect, they were far more pleasing to play than abrupt velocity-switched patches.

 

I agree that multiple switching, if it's done well, can be good, and I also agree with you points about resynthesis.

 

But, my real question is, are you as turned off as I am with the sound of velocity switching? Especially on those abruptly-switched Rhodes patches I talked about?

 

 

 

[This message has been edited by guestuser@guestuser.com (edited 11-12-2000).]

[marino]

The simple answer is: yes, of course. It is unmusical and not very controllable. Another close example is the fingered bass that becomes a slap bass at high velocities. How do you *play* a sound like that? Still, most "workstations" (aarghh) I've played proudly show both of these sounds.

 

About crossfading, I'll try to explain with a little more detail.

The electric piano sound is, admittedly, one of the few *imitative* sounds that can benefit from velocity crossfading. That's because the chorused sound that results in the crossfading zones usually enhances the overall Rhodes sound: we are used to associate the Rhodes with chorusing, and there could very well be a chorus processor from the synth to smooth the transition even more.

BTW this works mostly if you only set a couple of velocity zones, or maybe three, but with an emphasis on adding sound in the higher velocity range.

But you have a point here; sorry if I sounded a bit drastic in my post.

 

When you try to deal with a purer kind of sound, or anyway to have a natural/realistic response over the whole velocity range, things change a lot.

What I tried to do a few years ago was to built a library of multisampled solo wind instruments with four crossfade zones (yes I know, I am a little perverse).

What happens, as I explained in my previous post, is that you have zone 1 (very low velocity) with a pure sound, then it starts combining with the sound in zone 2 as it enters its velocity window (medium-low velocity), and the two sounds start beating.

Then, as you leave zone 1, the sound becomes purer again and you only hear the zone 2 sound, then as it enters zone 3 (medium-high) it starts beating again... and so on.

 

So you hear definite steps, with 3 (in my case) phased/chorused "sub-zones" in the transition ranges.

An alternative is crossfading in such a way that you always hear two zones at least, so you have a constantly beating sound, but smoother in the whole range. That could be acceptable in some situations, but not for pure, realistic acoustic instruments imitation. (and yes, I DID fine-tune the samples)

 

Again, please forgive my English

 

marino

[Guest]

Another example:

The rhodes sounds I use are from Wizoo's Magnetica disk and have four velocity zones. Although the patches are very "playable", I once programmed a crossfade program for them (which was a lot of work because the K2000 isn't very prepared for this).

I was very proud when I had finished working on this wonderful program, but now I don't use it. For one reason, there's this phasing effect that marino describes, for the other, the final program has five layers, so whenever I want to make a change, I have to do it five times. The hard switched version has only one layer.

Even when I play rhodes with a lot of chorus added (that would overshadow the phasing), I don't use the subtle crossfade ep. It's just not "handy".

 

Greetings

 

------------------

Marc

[guestuserguestuser.com]

This phasing effect you guys are talking about must be more prevalent on some samples than others. Like I said, I've programmed cross-faded Rhodes patches on both M1 and Triton that sounded quite good, and didn't suffer much from phasing.

 

To me, even if there is some subtle phasing problems, that would be preferable to those abruptly velocity-switched patches that we see so often. I would like to have the choice of one or the other. For me, most velocity-switched patches are wasted because they're un-playable. I would rather have the choice of a cross-faded one, even if it has some flaws like phasing or whatever.

 

What I'm trying to say is that I don't think an un-subtle velocity-crossfaded Rhodes patch should become some kind of defacto "standard" that everybody accepts. If nobody ever complains about this, we'll be stuck with it. At least give us the choice of a velocity cross-faded patch instead of a velocity-switched one, and synth manufacturers should put more support for velocity cross-fading into their synth engines.

 

Like one of the posters mentioned that there wasn't much support for it on his Kurzweil. That's typical. But you find support for velocity cross-switching quite often.

 

 

 

[This message has been edited by guestuser@guestuser.com (edited 11-13-2000).]

[Jim Aikin]

Glad to see that I'm not alone in detesting cross-switching. If the crossfading results in a "chorusy" sound, that may be musically acceptable, but it's likely to happen because the two fundamentals are actually out of tune with one another. Not good. If the fundamentals are in tune, you're more likely to get phase cancellation, resulting in a drastically bottomless sound (intermittently) or one that's comb-filtered.

 

There's a simple answer: If you want a truly velocity-responsive Rhodes sound, pick up a DX7 somewhere! In the hands of a decent programmer, FM synthesis is quite good at responding to velocity in a smooth, realistic way. The classic over-used DX7 bell/Rhodes is not the only EP program in the FM universe, by any means -- and if you've only heard this sound sampled into an electronic piano, you haven't really heard it. The whole point of it is that it can't be sampled, for exactly the reasons alluded to in this thread.

 

--Jim Aikin

[synthetic]

I sure would like true velocity response, too. But since we live in a world popluated mostly by sample-playback synths, I decided to actually answer your question.

 

I really like velocity cross-faded samples. One of my favorite examples is the Bob Clearmountain Drums II disc for my Roland S-760 sampler. It has 5-way cross-faded snares which sound like real drums on projects I've done, even a jazz trio. Roland includes a sampler disc with the unit that has some cross-switched Rhodes sounds, but you can get better ones from their Rhythm Section collection.

 

As far as synths go, I wrote a program for the Alesis Vintage Keyboards Q-Card which (IMO) is a pretty fair crossfade Rhodes. It's called "VeloRoads1" and it's program #8 on the card. It has four samples which it steps through based on velocity (modified by the Tracking Generator, for the propellerheads out ther.) Unfortunately, the piano wasn't sampled this way, so it isn't a true multisampled crosfade, but it works nonetheless. Very playable.

 

I think that some Roland synths and/or expansion cards have velocity cross-faded electric pianos, but I'm not sure which ones.

 

Good luck, -jl

[guestuserguestuser.com]

Synthetic, just to be sure, when you say "velocity cross-faded", you really mean cross-faded, and when you say cross-switched, you really mean cross-switched?

 

It's just that you sort of mixed the terms together quite a bit in your post, so that I thought I should check.