Luckily, quite a few of our customers don't share your opinion! [H]

/Dietz

Sorry, I don't mean to alienate Metrobot - that came out harsher than I meant to write. (I guess because I hadn't posted anything in a while since I was considering whether to use razor blades, barbiturates or carbon monoxide. Which do you favor Evan?)

However I would go so far to say that Metrobot is correct on one thing - it is very difficult to get a sample solo violin to sound natural, perhaps more difficult than SOME (not all) of the other instruments, and the reason is that the violin naturally makes so many instantaneous tiny changes in the entire basis of the tone. For example, to a large extent, a really great clarinet player will strive to obtain a beautifully clear, constant, even tone much of the time. Also a horn. I know that for sure on horn, because most of the time I cannot do it. [:O]ops: Whereas (and I am only generalizing so violinists please don't start frothing rabidly) a violinist usually strives to do the opposite - to shade the dynamics, the vibrato, the bowing on a moment by moment basis exactly relative to the phrasing of the music.

This is incredibly hard to do with samples. I only tried it a little on my Andante for Violin and Piano on the user demos. I don't know how musically valid that performance is. But the ultimate approach to using the solo violin and other solo strings is potentially very expressive in an actual solo performance, due to all the dynamic and performance legato/detache/spiccato samples. If one really did it right one could actually program all the bowings and every single nuance of dynamics without compromise. This might take weeks or months to fully use in a performance the expressive potential of these samples on even a simple piece.

@Dietz said:

Luckily, quite a few of our customers don't share your opinion! [H]

/Dietz

As is their perogative..

DG

@William said:

For example, to a large extent, a really great clarinet player will strive to obtain a beautifully clear, constant, even tone much of the time. Also a horn. I know that for sure on horn, because most of the time I cannot do it. [:O]ops: Whereas (and I am only generalizing so violinists please don't start frothing rabidly) a violinist usually strives to do the opposite - to shade the dynamics, the vibrato, the bowing on a moment by moment basis exactly relative to the phrasing of the music.

I think that you've hit the nail on the head here; this is what playing a string instrument is all about. Sure, one should be able to achieve a constant sound, but that is not where the greatest characteristics of the instrument lie. The answer? I don't know, but if I figure it out I'll be sure to patent it and make a fortune... [:D]

DG

You see, I think we agree more than not, William! [:)] Please take my frankenstein comment as a bit of hyperbole. I suppose if you really love something (VLS), it seems more fair-game to point out things that work less well.

But William's comment about consistency of tone are exactly on, I would think. Most other instruments have more easily defined "events" that make sampling them easier, whereas violin, sonically, is probably best divided up in bowstrokes, not notes. But the fact that the fingers are completely independent of the bowing makes the possible sampling scenarios almost infinitely higher in number. (and this is why i disagree that artistry helps much with this problem.)

Another thing that has always baffled me is, why does sampling work for string sections so much better? I suppose the modulation effect blends the individual nuance away. And then a section does what a soloist doesn't - strives much more for tone consistency as a section. But it's still strange...

Thanks Metrobot for not getting mad at my own hyperbole. (BTW I like that name. You're actually an android that lives in the city, right? As opposed to a Ruralbot.)

One thing you said contradicts your argument though -

"But the fact that the fingers are completely independent of the bowing makes the possible sampling scenarios almost infinitely higher in number. and this is why i disagree that artistry helps much with this problem."

You're right about the nearly infinite numbers, however the fact that the instrument's expressiveness in this case can, to an extent, be reduced TO NUMBERS indicates how with sufficient sampling power it can be done. To a degree this has already be done with the VSL solo violin (and the other strings) because if you look at the number of permutations and possibilities of combinations with the samples that already exist - it is huge. As I said the bowings can essentially be duplicated, but you have to work at it (to put it mildly).

I've also thought about this question of section vs. solo sounds being more easily represented, and have come to the tentative conclusion it is because of the fact that string players in a section are more "locked down" and restricted to doing something uniform - whether with steadier dynamics, unified (and therefore less obvious) bowing, etc - and this corresponds to what samples can do much more easily than a solo instrument, which is free to play to the limits of expressiveness. AFter all, why do string players enjoy much more chamber or quartet playing, and often consider symphonic playing a "chore"? Because they are grunts in the symphonic infantry, doing as they are told in unison, whereas they can be generals in a solo or chamber performance.

you got my name right!

I think the solo violin solution you suggest would probably only be usable with an intelligent, realtime articulation search engine for the sample database. (Maybe that'd be Performance Tool Mark II?) Otherwise it would be unwieldy. But the results would certainly be interesting. VSL's sample rate (ha ha!) for the other instruments seems to work just fine, but not so well with the violin. Which implies additional violin samples are still necessary. I'm only guessing, but suspect the creation of such a new sample-intense library might possibly collapse under it's own weight and drive the developers insane. Which makes other kind of algorithmic solutions interesting (forgive the heresy.)

If you are William Kersten, your user demos rock! (figuratively, of course...) What great string section performances. I don't think any other library can approach that level of expression and realism. And the allegro (violin & piano) is a lovely piece of music. [:P]

I am guilty as charged of being William Kersten. Thanks a lot, Metrobot.

(Sorry, didn't mean that to be poetic.)

You're right on the number of samples that would be required - maybe 1.5 million for solo violin ALONE (Herb is feeling queasy right now). Maybe it would drive people insane (as well as bankrupt) but it would be a great database. This brings up the often discussed question of the ultimate end of sampling, which obviously will include algorhythmic or somehow automated ways of selecting and using samples. If sampling continues as it is now going, without converting to a "processed" sound, a hugely musically intelligent program will be needed. Since this CAN be done, it probably WILL be done. Though right now, people are doing all this manually. We are still in the Stone Age of Sampling compared to what could be done. Though Herb is the leader of our tribe since he has the biggest club.

@William said:

[...] Since this CAN be done, it probably WILL be done. [...]

You bet! [H]

/Dietz

How would *you* know that, Dietz?

[6]

... uh ... just a wild guess ... ;-]

/Dietz

@William said:

For some reason, if this technology develops and is successful, I feel I may not want to do orchestral recordings with it. ....
So this seems to be a fundamentally different approach that could be considered an artistic divide.

That technology could be interesting if used for creating "real" synth sounds with being close to human expressiveness.
I imagine the expressiveness of a solo violin to be combined with all my solo synth sounds. [[:|]]
Well, that's not for recreating classical music, that's obvious...

tele