Does traditional sampling have a future?

[Bernard Quatermass]

I think part of the business tactic, in making huge sample libraries, some of which are now being delivered on hard drives, may be an attempt to thwart piracy.

[Mokafix]

We didn't stop painting when photography appeared. We just started painting differently

I think the abstract painting of al kind went unlashed at the moment they where no need of painters for realistic reproduction

Extapolate this to the work of sampling...

Yes, that was the point.

Wow, did I deserve that? My answer was not intended to be provocative ot anything...
First time I get insulted here, maybe not last...
But I'd have expected it to happen on something really offensive from me...
Sorry if you felt like this.
I was just kidding, because anyway, even FXs made out of synths can be easier and to use with a sampler, especially if you need several different FXs for percussions for instance, so really, I was not mocking you in any way.

[Bubbamusic]

If these libraries get any bigger, it will be cheaper to buy the nine foot Steinway than to buy the sample set for it.(and of course the Kontakt player) They should throw in a couple of Neuman mics with the piano.

Peace: bubba

[Bernard Quatermass]

Wow, did I deserve that? My answer was not intended to be provocative ot anything...
First time I get insulted here, maybe not last...
But I'd have expected it to happen on something really offensive from me...
Sorry if you felt like this.

Sorry. I get cranky sometimes. Just ignore me.

[steff3]

Physical modeling is probably going to be the future, but I imagine convolution as well as a limited degree of traditional sampling will be used. For example, as I understand it, TruePianos is mostly physical modeling based on a small amount of real samples. Synful Orchestra is PM based on a small amount of sampled phrases.

No, synful is in no means PM! it is concatenative synthesis and additive synthesis - no PM there!

best

[Krakatau]

concatenative synthesis

- doesn' t this one urgently deserves to be on the wiki ?!

[hollowsun]

Amazing things can be done with samples and some filters alone (see the old hardware synths).. It's however starting to look more and more as a forgotten old-skool skill.

I totally agree on that. And when you look at what, say, Steve from Hollowsun is doing with a rather minimal amount of samples, you just know that a lot of sample library producers are simply too lazy, especially given that most of nowadays samplers (the soft samplers at least) offer an incredible wide variety of tweaking options to make your patches "live up", such as sample start modulations and the likes.

I have to agree with those statements ... of course

And yes - I'm afraid that 'old skool' skills are pretty much dying out and there are (literally) just a handful of people who can create efficient professional sample libraries now.

Unfortunately as well, compact and efficient sample libraries are difficult to produce/sell now because the public perception now is that 'size matters' - i.e. if I were to release a killer piano library that's 200Mb, few people would be interested because it's just not big enough!!! In fact, I had one customer return his CP70 CD to me for exactly that reason! Despite veteran keys player Nick Magnus saying in his review of that CD in SOS "The CP70 must be an elusive sound to capture; the sampled renditions of this instrument that I've encountered so far seem to lack a certain je ne sais quoi. It's notable then that Hollow Sun have rendered this amazingly accurate and instantly recognisable representation, which is a real pleasure to play.", this customer wanted a refund simply because the collection wasn't big enough! And I get enquiries from people asking about my stuff asking "How big is it?" and/or "How many samples does each patch have?". I tell them and get the reply "Ok thanks - it's not big enough for my uses".

And yet one THE best sampled pianos I've ever heard is East West's 'Ultimate Piano Collection' for Akai samplers where none of the pianos are greater than 32MB - expert, seamless looping and programming. Such is the nature of the market now, East West are blowing this out for a mere $20 ( http://www.soundsonline.com/Ultimate-Pi ... W-080.html ) and I recommend it very highly - 20 bucks for two Steinways, a Fazioli and a Bösenforfer? Can't be bad!

I believe that there is a happy medium between the old days (when you had to squeeze an entire grand piano onto a floppy disk!) and the new breed of bloatware where one instrument is shipped on ten install DVDs by combining 'old skool' techniques with a decent quantity of samples that are a decent size but are looped expertly and programmed well! This is the approach I took developing library for Akai's S5/6000 and Z4/8 and more recently, the Alesis Fusion. And in fact, here's an example of how efficient sample-based sounds can be made:

Starting point : My CP70 is 200Mb - it has two full-length velocity switched samples (no looping) per note and sounds pretty good. Now....

1. Remove the low velocity samples - they are rarely played anyway - and use velocity > sample start to govern hammer attack and velocity > filter to affect timbre.

2. Instead of a sample every note, have one every minor third. Transposition is never more than +/- 1 semitone and artefacts are almost inaudible.

3. Loop the sounds, typically letting a lot of the sound to breathe and kicking in with a loop someway into the decay. Discard all data after the loop point.

4. Downward transposition tends to sound ok so examine the very lowest notes (the longer ones) and see if you can lose a few. I found I could.

Step 1 halves the size to 100MB, Step 2 reduces the size to 1/3rd (around 30MB), Step 3 gets the size down to around 15MB or less, Step 4 gets it down to around 10MB and with clever and judicious program editing (mostly velocity > sample start, filter, amplitude but also envelopes and keyboard to decay/release times) makes this greatly reduced sample set equally as playable as the 'bloat' version. In fact, this is exactly what I did when I ported my CP70 across to Fusion and you can barely tell the difference between the full 200MB version and the 'lite' 10MB version in Fusion, even less so when it's in a mix. I did the same with all the Trons I ported across to Fusion as well - a sample every minor third, etc., which got them right down in size. And so on. But ironically, optimising these samples was a lot more work than just sampling each and every note and then mapping them out to every key with little or no program editing.

And therein lies the ultimate irony - there's no way I could sell a 10MB CP70 for a price suitable to recoup the costs of buying a CP70 to sample, having it overhauled and tuned for the sampling session by a specialist and the amount of time taken to loop and program those samples to ultra-efficient optimisation so that it loads in seconds and has no CPU/memory impact ... but people will happily pay good money for something that is huge, is unecessarily sampled at 96/24, has had very little programming work done, takes ages to load, requires shitloads of memory and hits the CPU hard!!!! Errrmmmm.

But also, as someone else pointed out I think, the huge libraries are more difficult to pirate simply due to their sheer size - my 10MB CP70 would be all over the internet like a rash as it is small enough to download even without broadband! That said, I doubt many would now consider it seriously coz it's not big enough

It's a dilemna!

But I'm a stubborn bugger and I take great delight in producing efficient sample library and I was extremely gratified when, in his 5-star+ review of 'Nostalgia' in SOS, John Walden said "As you might have guessed, I think Nostalgia is excellent. Zero G and Steve Howell should be congratulated for putting together such a large number of 'classic' sounds in a single, very affordable library. I had no problems creating complete instrumental pieces using just the sounds within Nostalgia. What's more, unlike some Kompakt-based libraries, this one does make very good use of Kompakt's own processing options to add further colour and movement to many of the sounds. True, some of these sounds can be found with more comprehensive multisampling in other libraries, but what is here sounds consistently good, the samples are well programmed, and, unlike some sampled instruments, the instrument will not put a huge demand on the host computer."


Steve

[Bubbamusic]

I've had that Ultimate Piano Library for quite a while now and I paid a whole lot more than they want for it now. A few tweaks on the ADSR and it's not half bad - especially at the time it was released. I've used it on many tracks and it has always done the job that I asked of it.

Peace: bubba

[Sascha Franck]

Just to add this: The very first Rhodes patch I've been using was made of 3 (yes, THREE) samples. It's also been my first experience on doing a sample patch myself (I've been using the onboard sampler of a Terratec EWS 64 back then, so memory was quite limited), but I tweaked it a lot, using the already mentioned options and I think it really wasn't sounding too shabby. Admittedly, the interpolation quality of the EWS sampler apparently hasn't been too bad, you could even build patches out of a single sample without running into too much artefacts (something only possible with the latest versions of nowadays softsamplers), so that certainly helped - but yet, I was amazed how well such a minimal patch worked, especially in a mix.

In addition, what Steve said is quite true. People are often sort of complaining about the lack of samples. Or, the other way around, they are often like "wow, this comes with 74674 gigs of samples!!!". Which, in the end, doesn't mean all that much.
Sometimes it really strikes me as being crazy. For instance, you can often hear requests for super detailed snares. Ok, fine with me. But in the end, at least in a pop/rock context, a lot of people expect their snare backbeats to sound *exactly* the same on each hit. That's what drummers often are asked for as well - to make each and every snare hit sound identical.
And in case that's not so, snares are compressed and EQed to death, so each and every hit will sound identical indeed. I wonder how that justifies 50+ samples for a single snare.
Yes, if you program a lot of ghost notes, things will look quite different - but even in that case, quite sometimes tweaking your velocity to sample start and/or attack and/or pitch modulations can do miracles.

I can actually see the "need" for things such as detailed piano, strings and probably other orchestral (or big band) instruments. But for quite some others, I often prefer a well-programmed patch with minimal CPU/disk overhead. Especially when it comes to synth-ish sounds, I'd rather have the programmer use a handful of high quality samples and then get the most out of the sampler used.
Fwiw, a brilliant example of the latter would be Emagics "Xtreme Analog" CD for the EXS (now delivered with Logic 7). The source samples are absolutely excellent but very small, all else you hear is some proper EXS programming - and, as their done by Peter Krischker (one of *the* famous synth programmers), it shows how much you can do with just a good source array of samples and some programming knowledge. In the end, it's absolutely amazing how these patches almost don't tax your CPU at all, yet, as the "organic" elements of the patch are done by the sampler, they play like a really nice synth.

[soulata]

Sample sets for Nord Electro are usually under 10mb. I had one and sold it (I actually lent it to my friend and he had an accident with it so there you go...).

Rhodes and wurly sounded amazing (although I prefer Scarbee libs), esp. amazing for the size.

k

[emdot_ambient]

"Traditional" sampling to me is more what Steve described: limited samples, looped and optimized for efficiency. Whether that's in SFX or recreating other instruments.

Right from the start, though--and I've used sampling since the mid-80s--I've always believed using samples to emulate other instruments is kind of the no-brainer use of samplers, and ultimately not their best use. Don't get me wrong, I use samplers like that all the time, they definitely have their place. Even sample libraries of orchestras really have their uses--for game soundtracks, or for pre-production on movie soundtracks (i.e., as reference tracks for what will later be played by a live orchestra). But to think of using samplers only for that purpose is closed-minded, kind of like believing vocoders are only good for doing robot voices (when actually they're even better suited for complex active filtering).

The best sample programming I've ever done has been on an old Ensoniq EPS, which had extremely limited memory and 12-bit sampling. But I've done instruments on that which would be extremely difficult to duplicate with even something like Kontakt (while retaining the same level of playability, that is). These were instruments for creating abstract soundworlds, the original sound sources so transormed, truncated and obfuscated that even listening to the the unprocessed sample it would be difficult to determine from whence they came. Often we'd sample, transform, and resample several times. Or we'd exploit bugs in the software, record and resample. We'd use mod-wheels to alter loop positions, performance switches to combine different layers during live play, LFO's to modulate pitch/filter/sample loop positions . . .

You know, I'd bet that 90% of the sampled sounds we most often use today totally ignore 95% of the features our samplers offer.

I think we're all pretty much a bunch of lazy bastards (myself included).

[emdot_ambient]

Oops . . . OT: Yes, I think "traditional" sampling has a future. We just have to remember what it is and start doing it again.

When you see sample libraries of retro-samplers, you know that "traditional" sampling still has it's place, yeah?

[Roel de Witt]

Hi,

Well one of the points is that when accepting certain imposed or decided on limitations that this usually results in a better product. Limitations can be turned into new options and forces a person/developer to think about the actual product he/she's creating and to some extend this will get this person involved with it...

Having too much options is usually limiting since it makes one lazy. Best for creativity would be making your own engine (easier said than done) which supports all the features you really need/want. Not more, not less, as it helps to understand the product you are creating.

The negative side however is that time is money these days. If a company has the choice to create a virtual instrument in 2-3 days or spend at least several months or longer on it, then it'll usually pick the easy, least expensive method.

In the end, for commercial products, the market will decide on the methods used. The market at this moment in time seems to favor the bigger-is-better idea. Which is something that will take 1-2 more years to turn around on a more global scale I think.

The good thing about this is that the big companies (let's call them the dinosaurs) are stuck into current ways of thinking, which means that they will lag behind on new developments. This opens up the market for innovation and longer development cycles of products which can only be made by smaller companies or individuals, simply because time-is-money doesn't count too strongly for them (yet).

The trend of new companies appearing with products (WIVI, TruePianos, Pianoteq, etc) using these newer techniques has started already and this pace will only accelerate. They have a head-start, and experience which will be harder and harder to catch-up with as time passes.

Again I'm sure that within the next 6-12 months realism of these new range of products will start to surpass the 'dinosaur' products of today on most if not all levels. It's not only the size and acoustic realism that counts, but also the level of natural control/feel these products offer which simply isn't available within the average huge library available today.

---
Roel / 4Front Technologies

[shamann]

Yes that's true but that is still comparatively recent compared to use of recorded samples in a creative way as musical collage and composition.

I suppose, but if we take the appropriation of recorded materials in a new musical context to equate "sampling" (outside of just its digital context), then consider that Harry Chamberlin started work on pre-fab sound instruments as early as 1946 (the model 100 rhythmate drum loop player for organs beginning manufacture in 1948). So on that grounds, I think it safe to say that instrument libraries are intrinsic to the history of sampling as collage works.

I agree with you about copyright problems, though, music deserves a healthy fair use policy on the use of quotations as much as any other medium.

It's funny to consider that users want the big/bigger/best instrument libraries. I find myself thinking "good lord, how would I even be able to keep track of all those files?" Those things aren't for people like me, who like working with sampled materials, rather for folk who simply want a library of sound generators at hand. I can understand how it would be difficult for some sample producers to charge some of their enormous price tags for very sleekly designed instrument patches. Often in the eyes of the market, you have to have heft to warrant a hefty asking price.

[zircon]

Physical modeling is probably going to be the future, but I imagine convolution as well as a limited degree of traditional sampling will be used. For example, as I understand it, TruePianos is mostly physical modeling based on a small amount of real samples. Synful Orchestra is PM based on a small amount of sampled phrases.

No, synful is in no means PM! it is concatenative synthesis and additive synthesis - no PM there!

best

OK, this is getting a little ridiculous. Synful is quite obviously not a synth that relies on a large sample library. Can we all agree on that? In fact, it is extremely small considering the range of sounds it creates, and the level of realism. The system integrates a form of sampling, drawing upon complete phrases (rather than individual samples) along with additive synthesis (as you said) and noise elements. I don't know if there's some academic definition of "physical modeling" but this system is CLEARLY not traditional subractive synthesis, nor is it traditional sampling. Unless you want to make another category for it all together, I think it makes more sense to just call it a PM system.