Dear community members and members of the VSL company, I am a registered user of VSL’s highly praised orchestral library. I’ve studied music at the Amsterdam conservatory from 1985 till 1990. Ever since, (Atari computers), I am following and using the computer techniques into the world of music. The development of playing real instruments with a computer seemed to have ended with the magic of sampling. Hughes amounts of sampling data, the bigger the better. When I heard the stunning results of a only 17 mb program from Modartt; the Pianoteq 3 which is only based on physical modeling I wondered what future there is for only sampled based companies. My questions and thoughts are these; How does VSL sees the future on this matter and are they going to have part into the physical modeling of instruments. What are the pro’s and con’s of physical modeling sounds. I’m not a professor in physics nor math but I can understand that these new techniques of constructing a sound are going to play a major part in creating sound libraries. Friendly greetings; Vincent Vaneker mail@vincentvaneker.nl www.vincentvaneker.nl Explanations of physical modeling can be found here; http://www.harmony-central.com/Synth/Articles/Physical_Modeling/ http://digitalmusics.dartmouth.edu/~book/MATCpages/tableofcontents.html

Hi Vincent,

 

I have Pianoteq and find it to be very impressive. However, this physical modeling system works well and is ideally suited for an instrument like a piano which is really tuned percussion. I have also experienced some of the modeling from Yamaha in the old EX5R - not bad at all. 

 

However, when you consider instruments like violin where there are so many nuances of sound production, tone, attack, portamento, dynamics on sustained notes etc etc, I feel it will be a very long time before such an instrument can become a modeling instrument and be really playable and expressive. 

 

I may of course be wrong and I'll keep an open mind on this. In the meantime, the Vienna Instruments are still my first choice.

 

Regards - Colin

I'd say instruments like strings and horns are all the more suited to sample modeling, especially things like legato/portamento, and crescendos.  Check out things like Sample Modeling's Mr. Sax and The Trumpet, those blow away any sampled versions of those instruments I've heard.  I think the future of most instruments will be a hybrid of sample material and modeling/synthesis/interpolation.

You might want to keep in mind that the Samplemodeling instruments:

 

1- use tremendous amounts of processing power.

2- to date are only solo instruments (as was the case with the Stradivarius and Cello that they produced for Garritan). Perhaps this technology is not as good for ensemble samples, i.e., violin sections, etc.

3- because of the above two items it makes it a bit difficult to create, say, a horn section of 4 individual trumpets, 3 or 4 trombones and 5 saxophones.

4- with the specific instrumentalists that Samplemodeling have sampled so far it is difficult to use them in an orchestral setting. Check out their orchestral demos - they are pretty tortured sounding.

 

And yes, they do sound good and work great in pop and jazz productions. Hats off to them for what they've been able to accomplish.

 

I wouldn't doubt that sometime in the near future VSL will adopt some aspects of this technology. I would expect that there will be some sort of update of the Vienna Instruments sample player that accomodates this. Also, because VSL has become such an advanced software company we should expect them to figure out what is needed to make ensembles work with this technology and make the sample player not require huge amounts of computer resources.

Vincent Vaneker:

Dear community members and members of the VSL company, I am a registered user of VSL’s highly praised orchestral library. [...]

 

Welcome & thanks for your friendly words, Vincent!

 

Personally I've always been fascinated by the idea of physical modelling. I owned a Yamaha VL-70m when it was new (mid-90ies), and I used its big brother, the VL-1, in the studio of a fellow producer. The colours you get from these instrument are simply unparalleled, especially when you don't try to emulate existing instruments, but go for unexplored territory.

 

The main problem I see with physical modelling in day-to-day use is a very basic one: The better the "model" gets (read: the more it is able to mimic a real instrument), the more you have to be able to _master_ this very instrument with all its specific idiosyncrasies. You have to be able to really play a trumpet as well as a flute, as well as a tuba, as well as a violin, a timpani, a harp, and so on. The better the model is, the more you inabilities to play that _certain_ instrument will become obvious. And we didn't even start to talk about ensembles, eg. 20 violins!

 

That's the good thing about samples: They will always sound as masterful as they were played by the recording artist(s). They have "built-in" musicality, so to say. This will be a constriction for those instruments you actually know to play, but it will come just right for all the others.

 

In other words - I wasn't able to create the articulations with the VL-70m back than like I did with samples (... even the old-school ones we had back then 8-) ...). This may change in the future when user interfaces progress from where they are now, using intelligent macros and other amenities --- but then again this means that we depart from the puristic "as real as life"-approach.

 

In any case - interesting times ahead!

 

Kind regards,

 

Dietz:

The colours you get from these instrument are simply unparalleled [...] 

The better the "model" gets (read: the more it is able to mimic a real instrument), the more you have to be able to _master_ this very instrument with all its specific idiosyncrasies.  

 So for a trained pianist the Roland V-piano sound is better than Vienna Imperial?

Sergino Futurino:

 

Dietz:

The colours you get from these instrument are simply unparalleled [...]

The better the "model" gets (read: the more it is able to mimic a real instrument), the more you have to be able to _master_ this very instrument with all its specific idiosyncrasies.  

 So for a trained pianist the Roland V-piano sound is better than Vienna Imperial?

 

:-D ... good one, Sergino!

 

Hi,

I am not sure, but I thing that about every 2 years a new computer generation with nearly doubled performance cpus are on the market.

So I think physical modelling will become more and more important.

I am quite impressed by the trumpet, played with an Breath-controller, which I got a live presentation by my brother. (Only wished I could play such a controller as my brother is able to)

 

As a "fan" of VSL I also hope they will use the best features of both technologies in the  future in order to present the best available product-option to their customers.

 

regards

 Christian

aribaldi61:

[...] (Only wished I could play such a controller as my brother is able to)[...]

That was exactly my point ;-)

 

Dietz:

....The better the "model" gets (read: the more it is able to mimic a real instrument), the more you have to be able to _master_ this very instrument with all its specific idiosyncrasies. 

Well said Dietz - this just about sums it up. The challenge to play such a virtual instrument becomes as great as the challenge of playing the instrument itself - so you may as well learn trumpet, oboe, contrabassoon, cello etc etc. That's a lot more labor-intensive than using the Vienna Instruments!

Jack, those are all valid points.  But I don't see why any of those issues are things that can't be addressed as the modeling and the computers themselves continue to improve.  You're right, it's not totally there yet, but in the future it seems inevitable that eventually it will make libraries with a million individual samples taking gigs of space obsolete.  It will be interesting to see if companies like Vienna and some of the others that rely on exhaustive recordings will add this approach or not.

 

Dietz, I don't know that I agree that as the more modeling oriented instruments get better, they'll get harder to play.  Generally, I'd say those instruments I've tried have been much easier to get a good performance out of than purely sample based ones.  The one major change is just having to add breath control or another controller (or trigger via wind controller, which often works fantastic for real time performance on those instruments).

 

Even without changes in performance, it seems that the quality of transitions between dynamic layers could be greatly improved and the samples required cut way down by using interpolation instead of having so many layers as is often required with pure samples.

musos:

Well said Dietz - this just about sums it up. The challenge to play such a virtual instrument becomes as great as the challenge of playing the instrument itself - so you may as well learn trumpet, oboe, contrabassoon, cello etc etc. That's a lot more labor-intensive than using the Vienna Instruments!

 

Having tried a few of these modeled instruments, I don't agree at all.  Sure, there's some additional technique involved to take advantage of things like breath control, but once you master it, it is just as useful for every brass and woodwind instrument.  If you get to the point where you can get a convincing performance on a modeled oboe, it's easy to get performances that are just as good on other modeled instruments with little or no adjustment.

 

Frankly, in my experience I find it easier to play the modeled instruments - sampled instruments tend to require more dealing with keyswitches and controllers, which I find more of a challenge to learn than getting down the intricacies of a modeled instrument.  One example - is it easier to dig around for the right keyswitch for the right length of prerecorded crescendo...or to just play a crescendo with a breath controller and have it just respond properly?  Seems like a no brainer to me.

 

If you haven't tried any of these instruments, I'd definitely recommend checking some out, some even have downloadable demos.  It probably seems daunting if you haven't used them, but it was shockingly easy to get a ton of expression out of the ones I tried.

I am not so sure I agree it is more labor intensive.  I use both types of instruments and in the first initial passes I can easily get a more realistic experience with a model than with samples.  Because the samples require you to go back and edit them (in the VI), smooth them out between transitions, lower dynamic levels so they sound like the same instrument, choosing the best sample etc.  There are tons and tons of things you have to do with samples to get them right and to sound as good as they can.  With a model sampler, once you get it down pat, you can just play in your parts and it is almost completely done on the first pass (minus any wrong notes you hit).

 

There are arguments to both sides but I've worked with both and it is definitely much faster to get a finished recording with a model than it is with samples (if you are doing any type of advanced techniques especially).  Because I don't have to sift through 10 thousand samples trying to find the one that fits the passage/note.  For instance it may sound absolutely brilliant to use a 4 second crescendo patch for a note that is only .5 seconds long (because it has the speed, atk and growth I am looking for on that note).  As where with a model, I could have played that in perfectly in one pass and not spent 3 days looking for the perfect patch to fit that note.

 

The downside would be, there isn't a model out there that sounds as good as the finished sampled product so it's a catch at this point.  My opinion is your best bet at this point is still to setup extremely elaborate presets with all your patches perfectly smoothed out for one another etc to at least save you as much editing time as possible.  Possibly created your own speed controller presets.  It really depends on the stlye of music and the amount of dynamic changes you take in your passages.  I welcome modeling when it comes.  I know I will move to that when it becomes available and good enough sound quality wise.

 

Maestro2be

I also must say, I disagree with these technologies being more difficult to play. On the contrary: with those sample modeling instruments I get an immediate and very versatile playing experience, that is impossible with normal samples an a lot of key switches. The only drawbacks at the moment are indeed the CPU usage, the lack of strings, generally the lack of available good quality instruments.

Those pure PM instruments, like Wallander for example are in my opinion sonically not up to the point. This becomes obvious, when you wish to use them rather dry ... in that Sample modeling is the first and only competitor, offering great orchestral instruments. There are great saxes, a trumpet (which also works great in orchestral style in my opinion) and and upcoming trombone and solo strings. The company is rather small though, so development is slow - enough time for VSL to catch up! :-)

I really believe that this is the future. Some years in the future a software based pure sample playback won't stand a chance - especially with those rather high prices! We have to remember, when VSL came out it did already break some limitations of sampling: First the interval legato and repetitions, which has become standard nowadays, then the Vienna Instrument software with speed detection etc. and finally MIR. The next step MUST be a new Vienna Instruments software with the next revolution: Transitions between samples (take the AET in Kontakt 4 for example, even if the sample modelling technology is far better!), playable and realistic pitchbend-, vibrato and speed (e.g. tremolo or legato transitions - only sample modelling offers that sounding like the real thing),  Mircro tuning, etc.

I know, that is gonna be a hell of a lot of work on so many samples and many sample in the library (e.g. crescndos) might become almost obslolete. But if VSL wants to keep the pole position, it will be necessary. Finally, all of us want to do music and have a great time doing that. And what playable instruments have is simply so much fun playing! Almost no editing, no constructing ... This is almost the most important factor for me, because i spend a lot of my time doing it and wanna have fun!!!:-)

 

I just heard a demo of the upcoming SM trombone doing a Jabba excerpt from star wars and it sounds absolutely flawless to me.  I'd be shocked if a purely sample based instrument could come even close.  It will be interesting to see if Vienna gets into this sort of technology in the future.

 

www.robertosoggetti.com/jabba.mp3