vsti like Ensoniq Fizmo ?

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion
Post Reply New Topic
RELATED
PRODUCTS

Post

maybe. actually, i think SFZ can handle crossfade layers... i may try to put some of these samples into an SFZ and load it into alchemy player or sforzando later, to see if they work better there...

Post

zerocrossing wrote: One might instead use a granular synth to achieve the kind of "wave scanning" we're after. No?

Nope!

The transwaves are actually very basic things architecturally, and that is their strength as well as their weakness. Not really much weakness though. They are what they are.

There is not any VST that does this, nor does it look like any VST ever will. Buy a used Fizmo, after checking it works. Sample it to heaven and back. Enjoy. Move on.


I've gone into this quite deep and it is just not possible. My SD-1 was a generation or two before the Fizmo. It can not be emulated in any way by anything. You need to set up complex chains and even then it defeats the whole point coz it takes ages where as the original SD-1 takes seconds. Less than that! The SQ8L is the closest any one has come. Couple that with some nice FX chains and then again, well, maybe...

The Fizmo had a very discrete set of components that provided a certain kind of synergy. Without everything in exactly the same place, it is always going to be close but no Fizmo!

I'd love to be proved wrong.

I'd love a TS-10. My SD-1 is on loan and it took me a long time to track down and find one in practically mint condition. I still owe the money for it too. But I have hundreds of songs recorded on its sequencer. All on floppy disk which mostly work.

I would say, if you want a Fizmo and can't get one - get a TS-10, if you can't get a TS-10, then buy an SD-1 or VFX - it sounds as if you are going as much for the sound as the sequencer.

I'm sorry if I got it wrong and you are a connoisseur.

How many people have compared the SD-1 and TS-10 and Fizmo together? I haven't. But I would listen intently to someone who had.

This shit is worth the hassle coz you can't get that sound any other way. Nothing to do with analog/digital bollox. It is just a very well built up synergy, like I said. It's on tap at the flick of a switch. Hardware is still good for some things.

Post

Okay, so my only experience with a Fizmo was when they were trying to get rid of them at Guitar Center, back in the day. I looked at the cheap silkscreened graphics, blech. I played a bunch of presets. They all sounded thin and harsh and brittle. In fact, "Fizmo" was a perfect description for what I thought it sounded like. Fizzzzzz....mo. Okay, so I suppose it could be a flavor, but not one which I would want to use often enough to keep a whole synth around just for that.

Years have gone by, and occasionally these conversations flare up, raving about how unique and cool sounding it was. I figured my judgement and tastes might be a bit different now. So I just went and listened to a whole bunch of YouTube demos.

Nope! Still sounds thin and harsh and brittle. Fizzzzzz...nope.
Incomplete list of my gear: 1/8" audio input jack.

Post

Sampling Fizmo would be quite a futile enterprise as this would not capture the liveliness of Fizmo's sound. One of the strong points of Fizmo was that you could jump from one sound universe to another only by turning a couple of knobs.

Also I tend to agree that Fizmo has a very digital sound but that's what I like. It's digital in a similar way to Waldorf Microwave XT II and that's a huge positive in my book.

Old but good article about it here: http://tapeop.com/articles/bonus/fizmo- ... r-was-and/
No signature here!

Post

Dune2 with four voices with their own wavetable should do the trick !
Image

Post

Veering off topic (as is THE WAY OF FORUMS) a google search turned up some cool, though not really Fizmoesque, plug ins I had not heard of. I was aware of Scanned Synth Pro, but not of Enzyme. Scanned Synth Pro always had a sort of "dead" sound to me, but Enzyme looks super cool.

http://www.humanoidsoundsystems.com/
Zerocrossing Media

4th Law of Robotics: When turning evil, display a red indicator light. ~[ ●_● ]~

Post

robotmonkey wrote:Sampling Fizmo would be quite a futile enterprise as this would not capture the liveliness of Fizmo's sound. One of the strong points of Fizmo was that you could jump from one sound universe to another only by turning a couple of knobs.
True that, but a good MIDI controller should be able to get you in the ballpark.
robotmonkey wrote:Also I tend to agree that Fizmo has a very digital sound but that's what I like. It's digital in a similar way to Waldorf Microwave XT II and that's a huge positive in my book.
I've poured over dozens of Fizmo and Microwave XT (was there an XT2?) demos and IMO, the Fizmo sounds so much more "sweet" and musical even when it's sounding ugly. I know there are legions of XT fans, but I've never liked it. I totally remember messing with it back in the day and thinking, "Oh, I like my TS-10 way better."
robotmonkey wrote:Old but good article about it here: http://tapeop.com/articles/bonus/fizmo- ... r-was-and/
Nice article! I love hearing about the history and drama of the synth world. I once read a book... I forget it's name.. it was a history of early commercial analog developers (Moog, ARP, etc.) and it was fascinating. The take-a-way was, "Just because you're brilliant enough to design a great synth, you may suck at running a business."

I think it might have been this book:

http://smile.amazon.com/Vintage-Synthes ... entries*=0
Zerocrossing Media

4th Law of Robotics: When turning evil, display a red indicator light. ~[ ●_● ]~

Post

carrieres wrote:Dune2 with four voices with their own wavetable should do the trick !
Not even in the ballpark. First of all, like we've mentioned, a giant part of the Fizmo's sound is due to the quality and nature of their "Transwave" wavetables which Dune 2 does not have and will not let you import. Dune's wavetables almost seem like an afterthought. There aren't really a lot of them (I'm using the demo) and the highest amount of waves per table seems to be 60 where the Transwaves consist of 128 per wt. Secondly, the Fizmo has 4 individual "parts" that consist of 2 oscs each. So, you could have 8 oscs running at one time doing different wavetables.

So... if you want a big trancy pad or lead with lots of unison voices, Dune seems excellent. As a "Fizmoesque" instrument... not even close as far as I can tell.
Zerocrossing Media

4th Law of Robotics: When turning evil, display a red indicator light. ~[ ●_● ]~

Post

well, tbh, it's pretty rare for wavetables to actually use 64, let alone 128, individual waves, so that might not be a large limitation - most of the time, the 'large' one is interpolated from a smaller set (i.e. there might only be 4 or 5, so that 0,32,64,96,127 are actually unique, and the others are just crossfades of the two closest ones).

it's not too different from sampling. if you, say, sample only every 3rd note, it pitch-shifts the nearest sample down/up to fill the empty places - you still get roughly a full range of pitches. sampling every note is a little more accurate though.

if you imagine, say, a filter sweep... in old midi synths, for instance, it would have 128 values. so, if you left the osc static (say, a straight sawtooth), and swept the filter, you would get a sweep of frequency responses - with the filter down all the way, it would have almost no harmonics, and with it up all the way, very bright.

all a wavetable does is, in lieu of an actual filter, sample each of those 128 steps. so now, instead of playing sawtooth and sweeping a hardware filter, you play the sample for 'sawtooth at filter 53'. to get a filter sweep, you change the samples you are playing so at the start you are, e.g. playing 'sawtooth at filter 87' and at the end, 'sawtooth at filter 19'.

now, if you're just trying to replicate a filter, then this is boring. but you can now create any arbitrary spectral shift into that sequence.

so the power of any wavetable synth really isn't the wavetable synthesis itself - that's cool, but if it only gives you sampled filter sweeps, not going to be that interesting, just like if you had kontakt, but only had samples of basic waveforms to put into it, you would say it's worse than any analog synth. but put some cello samples into it, and now it does something new. the spectrums that the included wavetables give you is what makes one implementation unique. and creating a transwave/wavetable to do something unique that's actually useful is not easy.

the asr-x is what I always wanted.... if you have a SW editor, you can create your own transwaves and drop them on there in their full 128 wave glory (can also do this on the MR/ZR series, but only using the rarer-than-hen's-teeth flash ram expansion)....

Post

One thing about the Fizmo is that it has a bunch of "hidden" parameters that can neither be seen nor edited from the panel. The only way to get at them is through a version of Sounddiver that was hacked for the purpose. I have a copy, but the Windows XP machine that it ran on died, and I have not managed to get it to work on Windows 7/8 (it just barely worked under XP). At one time I was trying to reverse-engineer the patch dump sysex format, but it was insane. The OS code, at least that part of it, must have been a real hack job.

Post

zerocrossing wrote:
carrieres wrote:Dune2 with four voices with their own wavetable should do the trick !
Not even in the ballpark. First of all, like we've mentioned, a giant part of the Fizmo's sound is due to the quality and nature of their "Transwave" wavetables which Dune 2 does not have and will not let you import.
Dune2 let you import wavetables, you can create them with audio-term
zerocrossing wrote: Dune's wavetables almost seem like an afterthought. There aren't really a lot of them (I'm using the demo) and the highest amount of waves per table seems to be 60 where the Transwaves consist of 128 per wt. Secondly, the Fizmo has 4 individual "parts" that consist of 2 oscs each. So, you could have 8 oscs running at one time doing different wavetables.
Dune2 is exactly a 8 part fizmo with 2 oscs each. So, you could have 16 oscs running at one time doing different wavetables.
so, i will call Dune2 a fizmo on steroid, don't forget the zero delay feedback filter, two awesome FX bus, the arp with a step sequencer or midi import and 4 mseg !
i can't believe you can't do fizmoesque sound with Dune2
Image

Post

Actually Transwaves are quite different than the wavetables used in Waldorf and all the other synths. In PPG wavetable there are 61 slots + the usual common waves (saw, square). Moreover every wave in a slot is actually divided in two so that the first half is mirrored to make up the second half. All the waves in a wavetable move from one to other more or less smoothly but do not have to.

In transwave there are 128 steps that are complete waveforms (not mirrored). Also the transition from one step to another is always smooth. But not only that, transwaves are calculated in such a way that transition from any step to any other step is always smooth! So you can modulate the loop points however you want and still get completely smooth transitions between waves. That's why Ensoniq synths do sound more silky and liquid than Waldorf's. I'm too lazy at the moment put and image of the raw transwave here but it looks completely different than you would expect from usual wavetable.
No signature here!

Post

robotmonkey wrote:Actually Transwaves are quite different than the wavetables used in Waldorf and all the other synths. In PPG wavetable there are 61 slots + the usual common waves (saw, square). Moreover every wave in a slot is actually divided in two so that the first half is mirrored to make up the second half. All the waves in a wavetable move from one to other more or less smoothly but do not have to.

In transwave there are 128 steps that are complete waveforms (not mirrored). Also the transition from one step to another is always smooth. But not only that, transwaves are calculated in such a way that transition from any step to any other step is always smooth! So you can modulate the loop points however you want and still get completely smooth transitions between waves. That's why Ensoniq synths do sound more silky and liquid than Waldorf's. I'm too lazy at the moment put and image of the raw transwave here but it looks completely different than you would expect from usual wavetable.
The slots of Waldorf MW II / XT (and also Dune2, Surge, Blofeld, VAZ Modular...) can also hold arbitrary (= not mirrored) waveforms. Further there's no significant difference between wavetables and transwaves besides that the latter are limited to 128 slots (Surge: 512 * 1024, VAZ Modular even has unlimted number of slots and slot size).

And in transwaves the interpolation steps must be precomputed and stored - which leads to that most of them have a "sweeplike" content to keep them smooth. The more different the content of the slots, the more the stepping becomes audible.

There's no realtime crossfading for transwaves. But indeed crossfading is not the only type of "morphing" from one slot to another: Audio-Term - like most of the synths with realtime interpolation - uses crossfading. Tranzilon uses a vector approach, which requires manual setting of points and shaping of the lines between.

But you could load a precomputed "vector - transwave" into the wavetable osc of VAZ Modular (or into Surge, this only requires another file header). Turn crossfading off (in Surge you have to load them into the "window osc mode" an load the rectangular window) and - at least the raw osc output - will sound very close to the original.

Post

I had an ASR-10 and can confirm all of that robotmonkey sad above -- I can add that the reason for the smoothness when moving a loop along the wavetable seemed to be due to the way the algorithm made the loop points jump from 0 cross point to the next zero cross point of the next cycle (and not simply from one sample byte to the next). This technique almost totally eliminated clicks and glitches while moving the loop points in parallel across the wave table. I came to this conclusion by making experiments in which the wave table was a sampled sequence of 16th notes and the loop start and end points were separated by exactly one 16th. Then, using the mod.wheel I would move the loop and it would happilly go plink-plonk jumping from one 16th note to the next without any glitches. If the sequence had a filter sweep from my MS20, for example, lo and behold -- a filterless sampler sounding exactly like a mod.wheel controlled MS20 filter-swept in real time!!! :D
I miss transwaves just because of this trick -- it really was a marvellous and outstanding technique!
www.youtube.com/Synthillator
er... keep on rocking (despite all obstacles :shrug: ) :band2:

Post

A transwave data set holds fixed loop point data and if these data don't fit equal sample voltage (amplitude), the transwaves will glitch. This is more audible on short single cyles than on longer (like 16th) wave snippets.

As far as my experience as former ASR10 user goes there's no "mystic" realtime smoothing algorithm in the ensoniqs. How would this algorithm distinguish between a zero crossing at the end of a wave cycle and a zero crossing within a wave cycle? For example when using a complex wave and then doubling the frequency per slot. In this circumstance the transwaves made with Audio-term also wouldn't work (they would become too short because of the 128 step limit), but they do.

On the other hand, the first experiments with Audio-Term and ASR10 produced very glitchy results, because the loop range data where slightly shorter than the loop in the wave data. A "realtime smoothing algorithm" would have corrected such mistakes.

The ensoniqs don't have the computational ressources for doing such complex decisions and corrections in realtime. They barely manage to "synthesize loop" some seconds of audio within five minutes, and this task is easy in comparison to detect zeros, decide which of them are important, change loop points, smoothing artifacts and adjust playback to keep pitch in realtime. Indeed the zere crossings are usefull when looking for initial loop points, but this is offline processing.

By the way, none of the wavetable synths I've "reverse engineered" for format support purposes shifts the scan loop simply from one sample byte to the next. They all shift by slot length (when the slot is done) and/or crossfade the current and the next required slot.

Post Reply

Return to “Instruments”