@jc5 said:
I am curious to see just how 'intelligent' an 'intelligent' instrument can be made at this point in time.. while I would value this as much as everyone else, I'm not sure that I'm willing to give up control of my articulation choices unless the system really works beyond what I've been imagining...
I don't know if VSL is doing this, but the need for instruments or tools (whatever you want to call them) to be able to have some awareness of musical knowledge structures has been pressing for some time now. When they finally come, such tools will seem like a great advance to us. Unless we remember that software tools that deal with text have had capabilities such as word search-and-replace for over 20 years, while we still don't have such a thing for, say, a musical motif across all tracks, etc. It's sad that we have to get 'excited' about software development in music-tools moving from the iron age to the bronze age.
I doubt very seriously that much will change in the performance tool area with whatever is coming on Nov. 26. I'll be very, very happy to be wrong, of course. It's just that I think commercial demands for tools to work equally well in real-time as they do in playback are hampering devolpment in this field. Even Eric, the Synful developer, has run into this mentality. If the developers could ignore the absurd requirement that their tools must work during real-time keyboard playing, then they could make huge advances and contributions. For example, if the tool were run "post playing," it would be trivial to analyze data, determine exact note lengths, whether crescendo and diminuendo is happening, etc., etc. and then swap in the most appropriate samples for those musical situations.
Going outside of VSL's focus ... imagine music processing tools that get away from the "signal processing" mindset and focus instead on musical structures. Then imagine being able to select all instances of a theme in an entire MIDI file, alter the second note of the theme, transpose the theme, assign it to a specific instrument, modify its loudness ... or whatever. Woud be cool, no?
Imagine a tool that knows harmonically where the expansion of a phrase reaches it's highest point and then can adjust the loudness layers of the sample instruments (to use EXS terminology) to build energy toward that point, and resolve it away after it. (Or, you could have it render the inverse of that ... increasing energy toward the cadence. You could do whatever you like, once the tool can actually deal with the data in musical terms.) You could click "render phrasing" with a slider giving you the scale at which to do so. And the underlying "best possible sample choice" engine from VSL would select and employ the best sounds to fit the musical situation. Of course you'll be able to tweak anything "manually" as well, just as you can now.
But, my point is that, when you compare tools such as PhotoShop and AutoCAD (not to mention plain old grep) to what we have to work with now, it's clear that music performance software tools fall far short of where they need to be.
Somehow, though, I think the collective desires expressed already in this thread far surpass what the VSL will be able to deliver this month. Such is the danger of "hyping" a crowd of highly imaginative and passionate people! Nevertheless, if the VSL can pull another rabbit out of a hat and address even some of these tired old issues, I'll certainly step up to the plate and buy it!
- Paul
Very well put...
I too have been arguing recently over at the Halion forum, the need to move away from real-time live playing... I think the current integration of VSL into DAWs is pretty horrible IMHO. I have both the pro-edition and the performance tool. Despite my outlay of cash for the latter, I find the integration of the Performance Tool such a pain in the bum that I don't bother with it. I just run my projects on a single machine and like it kept that way.
I do hope that this announcement is something truly revolutionary, rather than just more of the same.
More samples, added together with more CPU and harder integration would not make me upgrade... especially since I haven't bothered with the performance tool.
What I would like to see is less emphasis on the sampler and real-time playing and instead, a new SINGLE cross-platform VSL VSTi that acts rather like Synful does in that it buffers up the midi, looks ahead at what is being played and selects and plays the appropriate samples from a large database.
To do this, it shouldn't even have to pre-load any of the samples into memory... and should therefore be able to load into the DAW immediately. Instead, it merely acts as an intelligent file-indexing service that maps midi for a given instrument to the relevant samples, handling legato, repetition, stacatto etc. along the way intelligently.
This would be partly what the performance tool does at the moment, but with far easier integration into the DAW (e.g. no separate application, just a VSTi plug-in) and the buggering off (tm) of the performance tool as a stand alone tool... plus all the virtual/real midi routing that goes along with it.
I spoke to someone recently who has been developing such a tool (unofficially) for VSL... a sample selection tool.
Presumably this new VSL VSTi (VSLi???) would then be the foundation block for what would become MIR?
MIR would be the addition of reverb and spacial placing of the instruments based on instrument selection...
I think we need to move away from thinking about "samplers"... and about thinking about providing VSTi instruments that - in effect - behave as a set of real professional musicians would behave when placed in front of you.
They don't need micro-managing... they just need to be told how fast, how loud, what articulation to play... they work out the rest based on their 30 years of playing expertise. You then spend time working out what it is you want them to play. That is the interface between composer and player.
So - this is what I am hoping it is... I don't get at all turned on by yet more gigabytes of samples that I won't end up using as setting up in-memory templates is such a pain... If, however, I can use more of these samples, in-passing, without perhaps even realising it and without needing to load them into memory, that, for me would be a DEFINITE reason to spend a lot more money.
here's hoping....
