Sampling vs Synthesis - do i need synths?!

[DevonB]

compromise is what a sampler is all about

Or, infinite possibilities are what a sampler is all about, and "good enough to fool pretty much everyone" synth emulations.

If the sampler represents compromise in emulating a synth, then a subtractive synth represents complete failure in representing a sampler...unless the sound is formed by basic waveforms and ADSR envelopes.

I say your arguement is flawed. A sampler *IS* substractive synthesis. I can't think of a strict sampler that does additive, wavetable scanning, physical modelling or Frequency Modulation, for example. So in that repsect, you compromise by only using substractive synthesis methods to process the raw sounds.

Everything has its place. Use the right tool for the right job.

Devon

[rounser]

That's not a fair comparison, we could also say that a sampler represents complete failure in doing the dishes...

No less fair than asking a sampler to perfectly emulate a synth to any degree more than another synth can emulate a synth.

[rounser]

A sampler *IS* substractive synthesis.

Nevermind that a sampler doesn't engage in synthesis, I know what you're trying to say, and here's my response: a subtractive synth is limited to a handful of waveforms that it can synthesize, whereas a sampler can play back any waveform at all. That's why a subtractive synth is limited compared to a sampler. It's laughable to call a sampler a compromise when compared to a synth if you keep this in mind, or to consider a synth the easy route to get to sounds when the alternative to sampling a very complex sound (such as someone yodelling and then getting hit by an avalanche) is to attempt to program it using FM synthesis.

(We can talk about FM and wavetable synths and whatnot, but for the most part people use subtractive synths, therefore assume "subtractive synth" when I say "synth". That's by far the most widely used variety of synth, and usually what people are suggesting as an alternative to the sampler in this thread.)

[quincy]

You guys must be REALLY f**king bored. And yes i can take the piss, my thread

[Cyprus]

I say, who cares? Ultra Focus is my favorite VST and I didn't buy it because it was a sampler, I bought it because it sounded f**king spectacular. So whether it's a sampler or synthesis doesn't matter to me when I'm making a purchase, I just want it to sound good.

[aciddose]

i'd say you definitely need a good subtractive, if not to use directly, atleast to sample from. some of the best sampled music in my opinion combines natural sounds with samples of various synthetic sources. a sampler in my opinion is more of a very flexable tool rather than an instrument. i think what makes an instrument is the limitations and distinctiveness of the instrument. a great sampler should be able to fool anyone into beliving it is any type of instrument. while a great sampler may lose its own distinctiveness, it makes up for it in flexability. a good combination of instruments and tools makes music making easy. i think it is knid of like cooking, you know beef and onions go togeather, and everyone has probably had some. the basic recipie can be taken and adjusted though to create new and unique forms.

if you mix icecream, beef, tomatos, pineapple and cheese, i'm not sure if the results are going to be very great. likewise with musical tools. find a basic recipie that you like and make it your own. i think a sampler is kind of like salt, it can go with pretty much anything in moderation. though if used to much or just by itself, it will make almost anyone barf.

[olepro]

The synth with the BEST filters is still Vaz 2010.

vaz 2010 is a synth but it allso loads samples and wavetables.
It can make oscillator sync and ringmodulation allso with samples and wavetables or between the analog waveforms,wavetables and samples.
Besides that it has the best sounding filters and VERY fast envelopes for sharp snappy attacks.
A really warm and analog sounding synth that allso plays samples !

I tried 'Ultra Analog' but i still prefer vaz 2010 any time over it.

Take look at it...
http://www.kvraudio.com/get/105.html

.

[René]

I think that it all depends on how you aim to use the synth sounds.

For instance, a high-end sampler (i.e. Gigastudio, Kontakt or HALion), will allow you to play a one-cycle waveform, and then apply a filter on it. This is indeed subtractive synthesis.

However, most subtractive synthesizers go way beyond that. If you load a single-cycle, looped waveform in any sampler, soon after you start transposing it, loads of aliasing appear. Modern samplers are more and more prepared to load huge sample collections, and not single-samples-across-the-whole-keyboard. OTOH, most wavetable or VA synthesizers have optimized routines which minimize all artifacts on single-cycle waves.

Therefore, dozens of freeware synthesizers will return subtractive sounds with much higher quality than any of the above samplers, loading a single-cycle. It is possible to load a different single-cycle in each key, and that'll result in better sounds in a sampler. However, if you set the pitch-bend wheel at 12 semitones, or apply a big EG sweep to pitch, the artifacts will appear again in the sampler. Most synths allow those sweeps in a much higher range.

Again, if your music doesn't need those sweeps, or if you don't care about the distortion (some top artists have made a 'specialty' of heavily aliased sounds), then the sampler trick might do it for you.

I personally feel that synthesizers can be matched with samples in many music styles. Other styles won't allow for it. I would say that the main sampler will do great for many uses. You might extending it as much as you can, and then looking for a single freeware synth for those highly expressive lines.


-René

[tony tony chopper]

I was about to reply the same. A synth will generate a single-cycle waveform more CPU & quality efficiently than a sampler, because it's dedicated to that.

If you don't use a soundbank (takes time to create, and this for just a single shape), you'll be asking the sampler to resample the preferably low/with lots of harmonics shape, to crazy factors that it's probably not made to handle (and that'd take it a lot of CPU if it was).

Then there's a critical feature the sampler may lack: free running oscillators. But a random start point feature can work ok for that.

And finally a synth will always have more tweakability than a sampler. Not just for patch creation, but for modulation through the song.

[stale bread]

loving this discussion

[Bobro Cameron]

IMO what a sound is is more important than what a sound represents. The real question is, are you making a sound that is itself or one that is theatrically representing another sound?

IOW, does the sound work on it's own or does it require you to know what other sound it is acting out in order to succede? That's how to decide which instrument is best. IMO.

-Bobro

[aaastronomer]

the question (as always) is: how much work do you want to do?

with samplers, it all on you. (kind of ironic to buy a sampler for the presets? but, kontakt 2 is loaded with 'em!)

on the other hand, if you're a lazy s.o.b (and i know, at least, one) you can let some fab sound designer do the work for you. rob papen comes to mind here. albino 2 and blue 1.5 can take you a long way (with some minor tweaks...just to say you did do something...)

[smart]

The problem with synth sounds in samplers is that they'll never really sound quite as alive as a synth will.

If you want to go light, I'd suggest something like 2 synths - one nice VA like Fabfilter Twin, Minimonsta, or Albino for your standard synth sounds, and then something else a little more crazy like Absynth or Blue for more complex crazy atmospheric stuff.

Samplers will cover your real instrument sounds when you don't have the particular instrument handy. (strings, piano, flutes whatever)

[rounser]

i'd say you definitely need a good subtractive, if not to use directly, atleast to sample from.

Agreed.

However, most subtractive synthesizers go way beyond that. If you load a single-cycle, looped waveform in any sampler, soon after you start transposing it, loads of aliasing appear.
Therefore, dozens of freeware synthesizers will return subtractive sounds with much higher quality than any of the above samplers, loading a single-cycle.

Multisampling fixes this.

However, if you set the pitch-bend wheel at 12 semitones, or apply a big EG sweep to pitch, the artifacts will appear again in the sampler. Most synths allow those sweeps in a much higher range.

From memory, 12 semitones is fine if you're using a 44100 sample. Double the samplerate and you can go up another octave (i.e. 24 semitones) without aliasing.

Then there's a critical feature the sampler may lack: free running oscillators. But a random start point feature can work ok for that.

I'd actually consider that a disadvantage. There are threads on this board recommending using a sampler for offbeat bass because without oscillator reset (something which a lot of synths don't have), the synth lacks the punch of a sampler because it's oscillators keep flowing.

The problem with synth sounds in samplers is that they'll never really sound quite as alive as a synth will.

I think that's just a personal perception issue. Subtractive synths are quite "dead" as well, because apart from the modulation a sampler can also apply, they're just spitting out a constant oscillator tone. As with synths, the trick to injecting life into sounds is modulation, such as envelopes and LFOs....and if you're using a sampler rather than a synth, resampling.

Samplers will cover your real instrument sounds when you don't have the particular instrument handy. (strings, piano, flutes whatever)

That's a very narrow view to take of samplers, but a typical one, and IMO completely wrong. Samplers can play back any sound, have rendered drum machines basically obsolete, are your best friend for effects, and can create thicker and punchier sounds than synths (and I can prove it too - any sound a synth can make can be sampled and improved further).

It's interesting to compare KVR's synth bias to somewhere like DOA where samplers are the main tool, and it makes me wonder why things are this way here.

[René]

Multisampling fixes this.

It doesn't. Unless you don't change the pitch, or the samplerate. And still, multisampling will give you a few snapshots of the sound evolution, discontinuous.

From memory, 12 semitones is fine if you're using a 44100 sample. Double the samplerate and you can go up another octave (i.e. 24 semitones) without aliasing.

If you're playing a full-bandwidth sample recorded at 44100, at 44100, then the room you have to pitch it up is exactly.... zero. Any pitcshifting up will result in aliasing.

Doubling the samplerate to get another octave is cool, as long as you accept to live with two times the CPU load. But then, in some music genres sweeps of 4, 5 and 6 octaves are fairly common, so that's only a partial and expensive workaround. A synth is the way to go.


-René