Thanks a lot guys! It is wonderful to hear from you on this.

on what Plowman and clarkcontrol mentioned, I admit I was very concerned about the repeated rhythms. So I only copied large sections in deliberately non-matching layers, never just one measure at a time, and played nearly all in real time. Except for the diabolical 16ths which Holst scored for ALL strings and woodwinds, in a totally off-the-wall combination of individually right but together wrong scales going in every direction during those hysterical sections before the slow middle part and at the very end.

Another thing I did that was a bit weird - on the col legno, which is scored for ALL STRINGS SIMULTANEOUSLY and which (understandably, being rarely used articulations) do not have a huge number of varied repetition samples, I had to use some cheating, since the number of repetitions was perfectly matched to the number of notes in the ostinato (arrghh!!!) --- I decided to do a randomly controlled crossfade between the slow and fast col legnos on all the instruments, as well as some unwritten dynamic changes to vary the attacks, which, if simply programmed at one velocity level, sounded like the dreaded machine gun times five (the number of separate col legnos).

also, on the section slop, I am trying to push it to an extreme. I remember once doing a slowed down re-recording of the New York Philharmonic under Bernstein doing a great performance of "Batuque" - a very rhythmic, spectacular showpiece. I could not believe how inaccurate it sounded, and yet when heard in context, it was a great performance by the (arguably, besides Solti-Chicago, and von Karajan-Vienna) greatest orchestra-conductor combination in recorded history. So I strongly feel MIDI performers have to "re-think" the entire concept of "accuracy" vs. "inaccuracy" in timings, not to mention tunings. Since we are seeing everything in 'slow motion" in a sense, by doing the performances from the ground up.

@William said:

...
also, on the section slop, I am trying to push the envelope there definitely. I remember once doing a slowed down re-recording of the New York Philharmonic under Bernstein doing a great performance of "Batuque" - a very rhythmic, spectacular showpiece. I could not believe how inaccurate it sounded, and yet when heard in context, it was a great performance by the (arguably, besides Solti-Chicago, and von Karajan-Vienna) greatest orchestra-conductor combination in recorded history. So I strongly feel we MIDI people have to "re-think" the entire concept of "accuracy" vs. "inaccuracy" in timings, not to mention tunings. Since we are seeing everything in 'slow motion" in a sense, by doing the performances from the ground up.

I'm so glad you mentioned the Bernstein example. In another thread I tried to explain that the better we understand "reality" the more effective the results will be when creating tracks in virtual reality.

This brings up another thought: wouldn't a col legno round robin patch come in handy?

I recently performed a MIDI test by recording the first minute or so of the first movement of Brahms Fourth using only, in succession, VSL, EWQLSO and the SI libraries. It was quite interesting to observe how wide a fluctuation of timing was required to achieve a similar sound, due to the attacks used, primarily in the strings, by the three- up to 50 MIDI ticks (480= quarter note).

"Since we are seeing everything in 'slow motion" in a sense, by doing the performances from the ground up."

That's it. That's it exactly. Your desciption of col legno reminds me of Serat working with his points.

Recently I was attempting a very simple violin accompaniment, a common alternating pattern of eighth note triplets at about 112 BPM. It was the kind of "barely hear it on a record" accompaniment that often speaks too loudly with samples, and it stranded me somewhere between 0'3s and 0'5s.

We've all dealt with this idiom -- that supporting material so commonly given to strings that needs to BE there and NOT be there at the same time, and one may argue that it is the most challenging string writing to be done with a sampled orchestra. We've got the broad, soaring stuff pretty well covered by now.

Anyway, I copied the notes, assigned the new notes to a whole note trill, and brought their volume way down beneath the short notes. It gave me the "smear" I needed. By itself, the whole note trills (besides being the wrong notes sometimes) would have sounded like a mish-mash. But with the clarifying pitches also in place, and at the right tempo (context being everything), the illusion was made.

Slow motion indeed.

Any ideas what to do next, William? Personal composition, or does another rendering call like a siren? Be careful. I know how absorbing these tasks can become.

Well this has really turned into such an interesting discussion, and kudos to the work done on the piece by William clearly this has touched a nerve in some people and brought about some discussion, and it's nice to see that discussion has reached out to broader areas of music.

I've often thought I was being too picky by thinking it would be great to have pp and ppp recordings of some instruments, (well all of them actually). Does any one else think that also?

The idea of getting into slow motion with parts and really rethinking each individual part not as a performance in technical perfection but as a performance has always been key to samples coming alive and there are many demo's here on the site that are a testament to what a great performance and even just re-thinking through the parts can achieve. It's true that the most subtle and seemingly transparent nuance in a supporting part can completely colour the entire piece in a way that no forward melody or instrument change may be able to do, and then we come back to the fact that it's all about the parts AND it's all about the whole - both at the same time and it's when it all comes together beautifully as a syncronised event that we get a glimpse or even a door into the magic of music and why it is so important to all humanity. Of course then there is the audience experience which also varies!!! [edit but I should say it's true that when something really works it tends to resonance strongly with a lot of people]

lol

Miklos.

Here's another vote for the public flogging of quantization.


Miklos, your mix work is excellent, and reminds me of my own woeful skills. (in a good way. I like a challenge!) I've learnt as much in this regard from Bill's wonderful performance of this work, as i have the deft touch you've applied to that performance to bring it alive still further. (And that's without mentioning the other work you've done with Guy, and your own offerings.)

A good ongoing discussion, and worthy of a thread on its own!

Leaving out the Quantize, and applying a 'performer's' instinct to this piece, proves emphatically that it can be done, and the results are good. Like many here, i did my time playing in orchestras, and when that experience is applied to performing with 'Solo' instruments like the VSL Symphony, it counts for a lot. On top of that, as Bill has proved, when one has an instinct for WHEN and IF, as well as YES/NO, when selecting articulations, dynamics, tempo variations, etc.., it raises the bar still further, away from a stilted or rigid non lifelike performance.
It's not only refreshing, but exhilarating to hear performance potential taken further along the road of discovery in our niche craft, and provides an extra incentive to push the boundaries still further.


My continued respects to you both,

Alex.

This is going off topic, but I just saw the DVZ video and wonder - is the automation of all these things going to cause more problems than it is worth? It seems that all sample libraries are automating to a greater or lesser extent. So with DVZ on 4 perfectly ethernet synchronized highest powered computers available, you HAVE to use the way it splits things, switches articulations, etc. Is this good? I wonder. A lot of people will love it of course, since it is "Instant Orchestration." Anyway, editing control on new libraries is decreasing rather than increasing. In the past everything was changeable from the sample up, and that control is disappearing with increasing automation.

(edit - tried to make this ranting a bit more coherent)

BTW, I reserve the right to make very belligerent posts on this thead, since this is a mission from god...

that is, the god MARS [6] [6] [6] [6] [6]

(please note the number of twisted evil icons)

[[:|]]

This is where orchestra libraries have reached a fork in the road. We've talked about making convincing tracks with VSL-- the pros and cons of quantization, the tricks of blending trills with long notes, etc., etc. Such things interfere with creating an accurate score so easily.

Those who want to create mockups for the sake of scoring only will likely make the best use of DVZ, as William pointed out. DVZ seems to address issues where GPO/Finale/Sibelius might have fallen short. But with VSL already in house, the expense of additional hardware and software are difficult to justify.

Granted, the slowest part of my process is getting VSL programmed. I feel that I need to improve my fluency with what I have rather than to cave in to the virtual instrument money pit... unless my employers are prepared to subsidize such a requirement.