@JWL said:
I don't think the VSL team would withdraw *all* formats of one of their libraries.
I'm not so sure, actually. Their player is cross-platform, and is highly optimized for their libraries. The data itself is encoded differently, and the programming required for each new instrument must be significant. So why would they necessarily continue developing for other platforms? I mean, it would seem sort of extreme to stop supporting other samplers, but with proprietary formats from so many companies, the only reasonable way to keep up is... well... *not* to keep up. Build your own integrated system, combining the samples and the player... And that's exactly what they've done! [;)]
As for "next", well, they have to continue making money. So they've got to keep releasing things they can sell - incremental upgrades to software obviously don't count. So, MIR is one thing. Major revisions of the VI player are another And, well... more sounds! Although it means very little to me, I'd say the Choir is in the cards. And as I've said a million times before - extended techniques! I do think they'll do it, one day. There's just sooooo much that can be done with any single instrument, and a lot of these so-called extended techniques aren't really so "extended" any more. They're quite commonly used for "effects" in film scoring, so I think it's only a matter of time. What a happy day that will be for me!
I also wouldn't entirely rule out a sample-resynthesis technique in the future. I guess it depends on how the whole RAM access puzzle plays out. However, there are realizations of sample resynthesis using SDIF (which is basically an open format - you can get as picky as you want) that sound pretty much identical to the original wav files (IRCAM's AudioSculpt is an example of such high-quality resynthesis). The big benefit here would be to unlock the sample in the time domain, making it possible to freely adjust note-durations. Also, in the event that VSL came up with their own notation-based sequencer (my number one day dream!), SDIF could provide a way of adjusting note durations on "dynamics" samples, for example, to perfectly fit the length of the note written in the score, or "tighten up" staccato notes for certain passages, or marginally stretch portato samples for a softer effect. SDIF could also but used to solve the phasing-on-crossfade problem, while at the same time providing the possibility for virtually continuous dynamics on all instruments and all samples...
Anyway, the big issue is that a highly-accurate, well-defined SDIF file is often *larger* than the original wav, so memory is still an issue... On the other hand, SDIF is frame-based, and thus entirely stream-able. I honestly think a top-notch SDIF implementation would be indistinguishable from the original, and could thus maintain VSL's devotion to the raw sample. Not saying it will happen, but it's a possibility...
J.