How many vel levels needed to accurately emulate a Wurli?

[dsetto]

How many velocity levels does it take to accurately emulate a Wurli?

 

I got curious as I was prepping commercial Wurli samples for my keyboard yesterday, and noticed very little difference between a lot of the layers, especially above the low register. Sureli the musical physics behind the reeds explains that.

 

[AnotherScott]

You can probably get away with fewer at the top than at the bottom.

 

What's "accurate enough" is a subjective measurement. There are 2 and 3 layer Wurli samples in keyboards that people are perfectly happy with. OTOH, the nice Acousic Samples wurlie employs 10 velocity layers for the sustain, and 20 for the release! I imagine they went that far because it was more accurate than useing fewer.

 

Arguably, the most accurate dynamic emulation would come with modeling, which I guess you could call either one level or 127 levels depending on how you look at it.

[dsetto]

Thanks. From what I saw, I can see that 2-3 is just fine. Maybe a few more for the low end. ... Yeah, it's subjective. Just wondering what folks' opinions are based on what they've seen being used relative to what they felt. Which I suppose includes familiarity with the real thing; which I do not have.

 

And for clarity, to my question I'll add "without the use of any synth techniques such as velocity dependent filter & amplitude. Instead, simple playback of discrete velocity-dependent multisample levels."

 

That's very interesting about the comparatively high release. Thanks for the response.

 

[Josh Paxton]

My first thought when reading this was "more than the Sk1." It uses four (I think), but they sound so different from one another that the result sounds noticeably jumpy. Of course it could just be that it wasn't sampled especially well. I think the Wurli in the Korg module is only four layers, and it plays much more smoothly. So I guess my takeaway is that it's not just the number of layers that's important, but also how well balanced they are.

[Josh Paxton]

Also, release samples are especially important for Wurli emulations. I realized this when I compared the stock Wurli presets on the Forte ("Meh. I guess they'll get the job done.") to the Purgatory Creek samples ("Holy crap! I AM Donnie Hathaway!").

[brenner13]

The Almighty Answer to the Meaning of Life, the Universe, and Everything: 42.

[dsetto]

Ah. And yeah. ... That Purgatory Creek samples guy knows the mechanics of character. __

It definitely matters how the "playback engine" fine tunes it to make the limits of the library at hand & the tech at hand shine.

__

I'm exploring reducing a Wurlitzer library I'm trying to fit in a limited space. Not sure if I'll bother. But, I've decided two things. 3 things. 4.

 

1. For A0 & B0, I could hear 8 & 9 sufficiently distinct timbral differences based on the commercial set I am reducing. There were 3-4 levels that were too close, and I'm setting those aside.

2. If you don't want to mess with assembling your own round-robin, velocity-proximal layers of timbrally similar samples may be a decent, quicker alternative.

3. I accept it's 42. I'll keep searching. And then I'll report it was 42.

4. Prepping multisamples is hard, when you are not using Kontakt, a DAW, any keyboard. And just using a barebones keyboard-specific software editor. Dumb, even.

 

Ok, 5. Anything that leads me posting over doing is a sign - something needs reassessment.

 

 

[dsetto]

Ok, I've decided having a nice Wurlitzer is more important now than another piano & another Rhodes in my board. And, I'm not gonna spend the time reducing the large library I have. So, I'm just gonna use the library basically as it comes.

--

 

As follow-up to my original question:

In my particular commercial library, for attack samples I found:

- C3 (middle C) to have about 6 timbrally distinct samples. (9 perhaps, if detail-oriented.)

- at E4: I found 3 timbrally distinct samples. (5 perhaps, if detailed-oriented.)

 

 

[poserp]

Also, release samples are especially important for Wurli emulations. I realized this when I compared the stock Wurli presets on the Forte ("Meh. I guess they'll get the job done.") to the Purgatory Creek samples ("Holy crap! I AM Donnie Hathaway!").

 

The stock PC3 wurlis have release samples, in fact I stole them for my synthesized wurli:

 

 

IIRC the volume for the Layer that has the release sample is on a slider, so you can have more or less of it if you want. I'd be pretty surprised if the same is not so on the Forte.

[jeffinpghpa]

Yes, yes, yes to the Purgatory Creek in Forte. The 200a and 140b have 7 sustain velocity layers and 3 release velocity layers. Plus you can actually control the nuance of those layers because it doesn't have a turd sandwich action like the Fatar TP/100 keybeds.

[dsetto]

Does a Wurlitzer pitch go a little sharp if you hit it harder?

[dsetto]

I feel like I'm on planet Dumbdor not having a keyboard with Purgatory Creek format samples. ...

So I can file it away, which board has the turd sandwich action you're referencing?

 

--

Amazing command of synthesis, poserp. Must be fun to play.

[poserp]

Amazing command of synthesis, poserp. Must be fun to play.

 

Thanks! It sprang from my unfinished work on synthesizing a Rhodes. Somewhere in that mess I heard something that made me realize I was kinda close to a Wurlitzer, so I spent another six months seeing if that was actually the case and got lucky. I'm doing the same with clavinet, which IMO is probably harder than a Rhodes or a Wurlitzer, but so far I've only managed to create a thing that works as a sort of pseudo-guitar for keyboard players (film at 11, also I'll make the .pc3 file available for those who want it and are curious about the strange sorts of synthesis one can conjure up in VAST). Anyways, onward and upward towards synthesizing a piano...

[dsetto]

Some challenge! ... Today I heard that a very soft Wurlitzer reminds me of a very soft Rhodes.

--

In your clavinet - I hear a lot of resonance & dynamic filter attention. (Is that in the ball park?)

[dsetto]

To me, with my primitive setup, working with samples is like composition. And working with synthesis more like improvisation. Each with its grace.

[burningbusch]

Thanks Danny/Josh/Jeff for the kind words.

 

I'd say you can accurately capture a Wurly with fewer velocity layers than a Rhodes as the tone is more consistent throughout the range. I've sampled two, one at 10 and the other eight with roughly the same number of release samples.

 

You're right that more layers gives the player an experience similar to having round-robins. If you only have a few velocity layers, you will quickly hear it as artificially static. With the Rhodes I've done, they've all been 16 to 20 velocity layers. When I need to shrink them to ~eight, while they're still smooth, there's a playability with the larger vel layers that can't be matched. SO, if the engine is capable of more velocity layers, there are numerous advantages.

 

Also, with regards to release samples, there are release samples that are intended to be extensions of the sustained note and then there are release noises (the key/hammer returning, etc.). I've incorporated the note extension release samples. With Whurlies and Rhodes, the release is clearly audible and, I feel, a big part of the sound. The damper causes a change in pitch as well as the deadening of the sound. I've tried to emulate it, but really release samples are the way to go. They are actually very tricky to implement properly.

 

Busch.

[poserp]

In your clavinet - I hear a lot of resonance & dynamic filter attention. (Is that in the ball park?)

 

Off the top of my head, the VAST chain is something like:

 

shaped saw --> lowpass --> highpass --> [something I can't remember] --> 2 param shaper --> saw+shaper

 

so it ends with "FM", and does some fun stuff in between. The last layer that does the FM is doubled; that helps the clav strings have that sort of "hollow" sound you get with a string (I believe there's some comb filtering at play). However, it's still too "wooden" with not enough metallic "twang". Because of that, it makes an interesting guitar-like instrument (with some minor modifications) that can be voiced with different "pickups" and such. However, I still haven't found the secret sauce for getting more "metal" in my strings in the right way (I've tried lots and lots of wrong ways). I'd hoped with all the FM going on that it would magically appear, but the magic isn't quite there yet.

[dsetto]

Thanks for reminding me of the value of pitched release samples as extensions of sustained notes. For a while, the only keyboard release sounds I've manipulated have been on the stock ones in my motif or settings in my CP4. And there're none of that. ... The Prophet X announcement inspired me to take my Motif to the next level by putting 3rd party sounds into it.

[dsetto]

poserp-- neat. way, way over my head. Looking forward to hearing your results. ... I imagine the sounds are very playable. -- And wonderfully a thing onto themselves.

[dsetto]

I'm filling a limited, remaining space with Wurlitzer. For the purposes of authentic recorded results in a rhythm section-vocal setting, I'd suspect the release noises are least important and can be ignored, if there's no room (storage or motif 8-element).

 

My guess is that an extra velocity levels supersedes the noises. But, one layer of pitched release sounds may very well be critical, and supersedes an extra velocity level.

 

In concrete terms, for the aforementioned scenario:

My guess is it's better to have 7 attack velocity levels & 1 pitched release sound than 8 attack velocity levels and 0 pitched release sounds.

 

--

I'm constructing this based on guessing, as I don't have setup a way to play these things and experiment. Once I fill up my flash and get around to playing it, I'll get a better sense.