The New Kid on the Block

[plowman]

VSL has had a very open forum policy. At one point, Dietz even posted a link to Synful, if I recall correctly. In that spirit, I note the following. NAMM will include a debut of a high-end entry into the sampled orchestra market. The company is Audio Impressions. Search as you will.

Stray thoughts.

1. Here's a paradigm shift. From a world where users crave a single computer solution, now we have a high end approach that boasts not only the likelihood, but the need for multiple computers. One computer alone will be the control center, another for the sequencer, and multiples for sample generation.

2. What irony. Ai stresses the ease of their system, and yet it demands sprawling hardware. VSL may have a more complicated method to work, but the computer(s) it needs are modest.

3. Modeling and sampling continue to converge. It allows the library to have a smaller footprint, but can it be done? Sampling issues don't just disappear. How do they handle the legato? What of phasing with mod cross-fading? Ai is a bit coy about how they do it. While promising, the only demo I've heard (and seen) is in an echoic convention hall.

4. Mac types face another nudge. Some, but not all, components require a PC.

Much more waits to be disclosed. Many here were aware of this developing library. But the official roll-out has started. It all gets curiouser and curiouser.

[JBacal]

The thinking behind their technology is intriguing and promising. But will it deliver the results everyone craves? Time will tell. Interesting times.

Best,
Jay

[DG]

I think that there will be a lot of disappointed potential users regarding AI. Even if the sounds are good (no evidence to support even this ATM) there will still be a need for keyswitching, whether visible or not. To me there is no difference between selecting "pizz" on a control unit to hitting the keyswitch that does pizz. I quite like the divisi aspect, but again unless the sound is good, the marginal ease of use over VI is not really worth it. The one big fault in VSL VI is the phasing issue on Velocity xFade. This is the area where I waste most time whilst programming, so if that could be fixed (Herb says not until faster PCs exist) then the next real technological leap has to be much further than anyone has so far suggested.

The other thing that has been mentioned is the ease of use transferring to notation programs. This potentially is a big one or me, but as I now use so few MIDI tracks it already is fairly easy.

I like the idea of a "random" button which causes the violin section to be slightly ensemble challenged (like the JB Ensemble maker) but I already find it annoying that I can't choose repetitions in VI, so I think that losing control of ensemble would drive me mad. The only good part about the proposed solution in AI is that with this version I wouldn't have to shift notes, which sometimes messes up the transfer to Sibelius.

DG

[nliberg]

I wonder how they will handle the divisi playing. The notes in a chord never arrive exactly at the same time and since the a program cannot look into the future (at least not until virtual instrument-sequencer integration improves) to see what notes are to come it has to wait a certain time. This usually implies a latency, I'd guess somewhere between 10 and 20 ms, to allow for sloppy playing. That's a possible downside to automatic divisi handling.

[DG]

I guess if there is already a random timing function built in there will be no reason not to play the notes at the same time.

DG

[nliberg]

@DG said:

I guess if there is already a random timing function built in there will be no reason not to play the notes at the same time.

DG

Yes, I agree, but I was referring to live playing in which case some degree of sloppiness is unavoidable. One has two choices as I see it, either one doesn't care about the sloppiness case and let the division among voices be wrong in some cases (can later be corrected by aligning notes in a sequencer), or one introduces a latency corresponding to the maximum allowed sloppiness.

/Nils

[Rob Elliott]

I think this technology is promising. My biggest interest and concern is - will it ultimately SOUND BETTER - free of artifacts of any kind. Daryl mentioned the one 'artifact' of xfading for solo instruments in VI that is really annoying and in some cases requires 'writing workarounds'. BUT, this is really the only one that is a showstopper for me in VI.

I don't like dinking around programming to get the 'sound' I want (mostly working in strings), but am willing to do it if the client hears something better. Of course having both would be nirvana. [[;)]]

As Jay says - interesting times lie ahead.


Rob

[Jack Weaver]

It is with some trepidation that I enter into talk about another manufacturer's product on a manufacturer-specific web forum out of respect to the people at VSL.

However Brother Plowman has bravely entered us into waters where many have heretofore refrained from going. I'm sure that Herb et al are aware that the customers that purchase VSL products also use other manufacturers' products. I don't know anyone that uses one suppliers instruments exclusively. So integration is going to be an ongoing issue. Maybe it would be more proper to limit our conversations with respect to other manufacturers to intergration - hardware & workflow.

No one knows what Ai really is - there are no demos yet - or what they will become. One thing should probably be clear to us though: VSL is very watchful of the situation and we expect Herb's continuing innovation.

Twelve months ago we were all amazed by Vienna Instuments and there is still nothing like it. What will the next twelve months bring? NAMM 2008 should be interesting.

Ladies and Gentlemen, prepare your wallets - they are going to be used.

[William]

I haven't heard any demos other than an inconsequential little blurb for tenor sax which sounded o.k.

However there is one complete fallacy in the DVZ (DIVISI) hype. That bad sample performances sound "like a pipe organ" because of too much doubling. That is NOT the reason, and is in fact a relatively minor factor, especially considering you can already do divisi with solo and ensembles of varying size as in VI.

The real reason for the "pipe organ" sound is lack of expression. So a "note-on, note-off, note-on" effect occurs. You can have all the divisi in the world, and that won't change a damned thing. It will always be a challenge to create real expression in a musical performance.

[plowman]

"It is with some trepidation that I enter into talk...."

I felt conflicted when I posted too. I held off last year -- and many here probably held off longer. I waited for one week before NAMM, and when their website (finally) updated, it was time. At that point, it becomes the elephant in the studio that no one is talking about.

More importantly (at our best) this forum is above idiotic side-choosing and biased deprecation of any product. The maturity of the above posts shows it.

What fascinates me is, we have this core issue of taking finite, essentially dead samples and making them sound alive. And every company has their own take on how to do it with minimal hassle to the user.

The element of Ai that most resonates is the single desk recording. We'll see how effective that is. I can't deny that samples of a violin section recorded en masse, whether 14 or 20 players, will create a sound of limited variety, and where there is less variety, there is less life. I've heard sampling companies take on the issue of compelling strings for twenty years. The per desk approach has promise.

And yet, VSL has the means to do the exact same thing, and surely the brains and the experience.

So, will Ai's approach result in strings that are significantly better than our own admixture of solos, chambers, and ensembles? And is the improvement worth two more computers and some five figure amount? When puzzling over my VSL purchases, the question was never if VSL was better. The question was, "Is it ten thousand dollars better?" And that will be Ai's tallest hurdle.

[JBacal]

@Jack Weaver said:

Maybe it would be more proper to limit our conversations with respect to other manufacturers to intergration - hardware & workflow.

I am certain that Herb and his VSL magicians have many exciting future developments up their sleeves. Unfortunately, we don't yet know what they are.

So hopefully they will not be offended if there is some discussion about features that other developers have announced for their products that we would like to see incorporated in some form into VSL.

I have only tremendous respect for the people and products of VSL and hope that they will remain on the cutting edge of tools for musicians for many years to come.

Discussions like these just might help them do that. My 2 cents.

Best,
Jay

[Rob Elliott]

Good point Jay. I am also certain the the 'next generation' of VI (on the chalkboards and perhaps already in actual development) will be further groundbreaking techology.


No sercret, as mentioned above, that we all use other competing libraries to satisfy our clients, but I made the decision to lean my 'main' ladder up onto the VSL building a few years back and have never once regretted the decision.

[[;)]]


Rob

[musos]

nliberg wrote:

@Another User said:

I wonder how they will handle the divisi playing. The notes in a chord never arrive exactly at the same time and since the a program cannot look into the future (at least not until virtual instrument-sequencer integration improves) to see what notes are to come it has to wait a certain time. This usually implies a latency, I'd guess somewhere between 10 and 20 ms, to allow for sloppy playing. That's a possible downside to automatic divisi handling.

Well, if you listen to their demo movie, there is such a delay that it sounds like a delay plugin has been applied. I know one can't judge from a low quality movie, but it sounds pretty sloppy to me....

[tomhartman]

@William said:

I haven't heard any demos other than an inconsequential little blurb for tenor sax which sounded o.k.

However there is one complete fallacy in the DVZ (DIVISI) hype. That bad sample performances sound "like a pipe organ" because of too much doubling. That is NOT the reason, and is in fact a relatively minor factor, especially considering you can already do divisi with solo and ensembles of varying size as in VI.

The real reason for the "pipe organ" sound is lack of expression. So a "note-on, note-off, note-on" effect occurs. You can have all the divisi in the world, and that won't change a damned thing. It will always be a challenge to create real expression in a musical performance.

That certainly enters into it, but there is no doubt that playing trombone chords, for instance, with way too many bones on each note, doesn't help.

The other reason things are often described as "organlike" in sound is just plain poor orchestration, i.e. having close voicings where the shouldn't be, etc.

I don't want to knock any product before I hear it, this new library may be amazing. That said, until someone knocks it off the top of the heap, VSL is still the one to beat, divisi or no divisi. JMHO.

TH

[drew buchan]

I fall into the category of very keen amatuer (little talent). It feels like I have spent decades drooloing over owning a VSL orchestra set ... and late last year I managed to afford the complete cube.

Even if AI is better (by whatever measure), and nobody has even heard a demo yet .... then I am not going to spend another 5 years drooling over a new toolset. Life's too short, and I've all the tools I ever dreamed of to make great music.

I guess very, very, very few people will have the money, determination, need, and obsessiveness to start out all-over again on a new architectuve, new set of tools, new headaches ....


.... and anyway, all this specualtion before anyone has heard anything?

[William]

Tom,

You're right on that. It is a factor, I just meant it is not the main reason. Also that's a good point - bad orchestration is a huge factor. In fact I'd say the three main reasons for bad sample performance, assuming a decent sample library, are in order of importance (sorry, I seem to like to make lists) : 1) lack of expressive nuances, 2) bad orchestration, 3) bad mixing and 4) too much doubling. So I'd put it about fourth in importance.

If VSL had alternate solo instruments and a couple two-player desks in the strings, DVZ does not have a chance at unseating them. I think they will probably be good, but I predict the musical quality of VSL will still kick their asses.

I don't believe this total automation stuff. For example, they say everything is right there, instantly. The trills, the tremolos... etc. O.K. But you still have to select them. that's the whole problem - selecting articulations and expressions. So you select them a different way than tracks. That is not necessarily easier. It could be harder depending on if the interface is clumsy.

[anton]

....however i am scared from the number of computers necessary and in
any case who guarantees the perfect operation...?

[JWL]

@drew buchan said:

...
Even if AI is better (by whatever measure), and nobody has even heard a demo yet .... then I am not going to spend another 5 years drooling over a new toolset. Life's too short, and I've all the tools I ever dreamed of to make great music.

I guess very, very, very few people will have the money, determination, need, and obsessiveness to start out all-over again on a new architectuve, new set of tools, new headaches ....


.... and anyway, all this specualtion before anyone has heard anything?

This sort of crossed my mind, too. Somehow, I don't think anyone would be talking about the "next thing" if there weren't the sense that something else were needed- or that something is currently missing (regardless of manufacturer).

From what I can tell, users of VSL VI love it-- and judging from the insane mp3's, there's no question that VSL VI is more than capable of truly awesome results.

I invested in VSL first, so adding something as costly as AI has little appeal for me at this point. To read a preview report that includes the blurb ".. if you need to ask about the price, you probably cant afford it. ..." implies something at or beyond VSL's MSRP. (For those who haven't seen it-- http://www.sonicstate.com/news/shownews.cfm?newsid=3264).

Even if money were no object, we're still bound by the limitations of hardware; the same article also cites: "...You need massive computing power and storage to get the best out of these libraries..."

We can chase after all the software in the world, but we're still back to square 'one' with hardware limitations. If this is a non-issue, then AI could clearly emerge as a powerful innovation and addition to the ongoing timeline of VI developmental history.

But I'd rather have users' first-hand reports than developers' pre-release ad claims.

Did someone say "time will tell"?

[plowman]

I persist in the thought that sampled strings can be improved. I hope Ai succeeds.

But, beyond the string sounds they may achieve, my interest in other Ai ventures drops precipitously. "Space" seems to me to be the fifth iteration on IR's. And maybe technically it has nothing to do with IR's. But however they do it, do we really need another way to put a sample in a room? Altiverb, Gigapulse, Space Designer, MIR. I'm just saying it's hard to get excited about another one, and equally hard to imagine that it could be so amazingly better. I'll try to keep an open mind.

"Orchestrator" interests me the least. It carries three issues for me. The first is, we have an extremely high-end product here. By definition, it targets music professionals. But these are the very people who know how to orchestrate, who (I'd hope) jealously guard what note goes to whom. As a rule high end uses don't want these choices taken out of their hands. How can a software program orchestrate a woodwind chord? (EDIT: I can't tell from the site if "Orchestrator" is limited to DVZ strings, or which is what.)

Secondly, "Orchestrator" stresses the ease of dumping to a score. But we've been doing this so long, we have MIDI methods to extract and re-group and print divisi. Yeah, it's annoying, and no, ensembles don't sound like divisi, but these are not Herculean problems to overcome.

Lastly on "Orchestrator," Apple: take heed. If you'd free your other arm from the incessant back-patting, perhaps you could more quickly write us an Environment object that senses the number of notes we're playing and lets us allocate them to different VI's set to different patches. As good as Ai? I doubt it. But a giant step forward for us.

I'd encourage Ai to tone down the "all you have to do is play" angle. As someone mentioned already, we know better. There are a finite number of ways to hit a key, a mod wheel, and a pedal. They are not sufficient to distinguish between staccato, pizz, and jete. Eventually, you'll have to tell Ai *something*, and that's a keyswitch or a CC#, no matter what term they decide to use.

But any active sensing improvements to minimize controllers are welcome.

And one final irony: Ai claims to be a high-end sports car, but that demographic is probably the least likely to futz with the coordination of seven computers, networking, etc. The mid-rangers tolerate the tangle of farms and compatibility, like the gurus here. They do it because their resources are limited. But the Porsche crowd has a low tolerance for it.

I think strings will make or break the deal for me.

[JWL]

@Plowman said:

And one final irony: Ai claims to be a high-end sports car, but that demographic is probably the least likely to futz with the coordination of seven computers, networking, etc. The mid-rangers tolerate the tangle of farms and compatibility, like the gurus here. They do it because their resources are limited. But the Porsche crowd has a low tolerance for it.

I think strings will make or break the deal for me.

Great point, Plowman. "If money were no object..." where dwells the common sense, indeed!?

It's got to be more than simply expensive-- and I dare say it needs to go further in its presumed price echelon than to simply sound good.

I, too, have grown a little weary of hearing how easy something is to use rather than hearing about how easy or difficult it is put to good use. Regardless of cost, it's the call of the consumer to determine the value.