Sample players vs algorithm type vstis

Hi all,
After working for a several years with sample based vsti (Ivory, Kontakt etc), I recently read where there is a school that prefer not use them and instead use “algo” (algorithm) based instruments Live. How is this approach superior to the Sample player approach or is it at all? I guess there is an argument for a few points (memory, no disk streaming etc…) but I was wondering if there are other problems I don’t know of. I have had good results with sample based vstis playing live for what it’s worth. Thanks

1 Like

If by “algo based instruments” you mean things like FM or Additive synthesis, there are many new VSTi synths that have improved the GUI and front-end experience for the user making both of those far more approachable than they were in the past to get really incredible sounds. They CAN at times be CPU hungry, but that depends on what you are trying to do, typically.

FM now goes way beyond the DX7 or TX81Z kinds of sounds (and have been showing off what the TX81Z with its non-sine operator selections might have been capable of had not everyone run away scared with their tails between their legs!)

I have a TX81Z and recent exposure to the recent versions of FM have opened my eyes to what I myself had missed as being possible with that box.

Additive also has been combined with wave synthesis and FM, AM, PM and Ring modulation to make it easy to change waveshapes to one’s heart’s desire.

I also have a Kawai K5000 additive synth, and believe me, the current crop of additive synth VSTi interfaces makes things a WHOLE lot easier!

There are other pure synthesis methods that are spin-offs of those two. Iris by iZotope uses actual graphical information to build its waveshapes! But it really is just a really cool interface for doing additive, from what I can tell.

I love that sort of stuff myself! :smile:


1 Like

Thanks for your excellent post Terry! Yes, i was talking about these things and also pianoteq, Arturia and other type modeled instruments. :grinning:

Edit: I changed the name of the topic to be more clear Thx

1 Like

Hi Dave,

there are essentially two key advantages to modeled instruments: switching/loading times and RAM requirements!

With a sample-based instrument, to change the sound/patch, you need to load samples from disk to RAM (either all of them, or in case of streaming samplers, at least the initial phase), which takes time. Whereas with a modeling instrument, you just load a different set of modeling parameters, achieving near instant sound switching.

Second, sampled instruments usually require significant RAM to hold the active sample set - not an issue for modeled instruments.

OTOH, some modelers put significant strain on the processor(s) - synths like U-he Diva are used as processor benchmarks for a reason…

Another argument made by the modeling faction is the expressiveness of modeled instruments: since parameters can be influenced directly, modeled instruments can react very directly to players’ input (key number, key velocity, aftertouch, modulation, various controllers) in their sound generation. For sample-based instruments, this requires a ton of zones and layers, plus some post-processing via filters and effects - and never gets quite to the same level of expressiveness. This is especially what divides the piano players amongst us - Pianoteq is certainly not quite at the same level of fidelity sound-wise as sample-based instruments like Ivory, but it is superbly playable in terms of responsiveness!

For my live set, I tend to run as much as possible with modelers, using only two sample-based instruments (Addictive Keys for grand piano and TX16Wx for sample effects) - and these only with one sample set (loaded at the beginning of the set). Everything else is modeled:

  • Lounge Lizard and Pianoteq for electric and additional acoustic piano sounds
  • VB3 for organs (Hammond, Vox, Farfisa)
  • AAS UltraAnalog 2 and FabFilter Twin for analog-type synth
  • Dexed for FM synthesis
  • Korg M1 for all bread&butter workstation sounds (OK, this one is very sample-based, but no loading times)
  • Scuffham S-Gear as guitar amp

This gives me practically instant song switching times with pre-loaded setlists in Cantabile 3 and a very reasonable RAM footprint.

Hope this helps!




Thank you, Torsten, for a superb article on this question! The replies so far have been very informative! :smiley:

Oh and by the way my friend uses Pianoteq and says you can tweak it pretty close to some sample sets, I must try it.

Yes, absolutely - Pianoteq is a “must-try” for Pianists. Its responsiveness to dynamic playing is great, and the ability to customize it is unparalleled. Want to play with slightly harder hammers? You got it! More resonance? Got it! A piano that is 30 feet long? OK!

My personal favorite: the “Condition” slider; range from “mint” to “worn”. None of my pianos has the slider all the way to “mint” - it’s far more interesting to add a bit of character by aging the piano somewhat.

My dissatisfaction with Pianoteq is its mid-range and sparkle: I like my pianos very crisp, having muscle and punch so that they hold their own in a rock band mix ; Pianoteq is a bit too mellow and has a somewhat “honky” mid-range for my taste; EQing does not quite cut it. But the Pianoteq models are great for classical type and more full-bodied piano playing.

And don’t forget their special add-ons: very decent Rhodes/Wurlitzer models, a great Hohner collection of Pianets and Clavinets, and also a very nice set of chromatic percussion (Steel drums, celesta, vibraphone, xylophone, marimba, …).

Once you get all of the expansions, things do get somewhat expensive, but it really is a superb one-stop shop for a broad range of sounds.



P.S.: no, I don’t get paid by Pianoteq for this :wink:

I’ve switched from Native Instruments Komplete 10 (The Maverick) to Pianoteq as my go-to Grand Piano. It is difficult for me to choose one over the other for all scenarios, but like Torsten, I like the customization options that Pianoteq offers. That, plus the smaller memory footprint and near-instant load times, sealed the deal for me for most situations.

Basically the same reasons I like VB3 over NI’s Vintage Organs. But then again, I’m an old-school synth kind of guy, which is why I also love the Arturia products (although not for a Grand Piano simulation).

1 Like

When i was younger I attempted to get Grand Pianos out of every subtractive synth i had, (mini moog, cs80) but didn’t even get close (I did however find a plethora of unexpected and truly wonderful sounds from the mini and the cs80) until the FM based DX7 came out. For the time it was the closest affordable thing that would even stab at it. Things have come a long way with the advent of “Romplers” like the Kurzweil and other companie’s offerings. Thanks for the post Roland, it got me remembering…