I thought the point of notation is to directly tell player how the composer wishes the music to be played. I’m not sure I understand the claim of “fundamentally decoupled”…isn’t the point of notation to make the coupling between the composer and the player? Sure, there is wiggle-room for interpretation, but the system of notation handed down to us is surely to maximize synchronicity between the composer and player. Any playback dictionary or expression map should ultimately make that coupling between the score (composer) and the VSTi (player).
I think the fundamental problem here is that makers of sample libraries have different interpretations of how to arrange the available articulations and the control channels that modulate those articulations. Spitfire has, for example, tried to give some degree of universality to articulation selection with UACC, VSL has approached it differently with their custom sample player, and again East-West with their player. In fact the arrangement of control signals has become so diverse for VSTi’s the Native Intruments, Alesis and others need to offer “control frameworks” of how buttons, sliders, and keys are to interplay for consistent playback.
So then, the next natural questions are:
- who will be responsible for the “bridge” between the score in electronic form and the VSTi players available
- how will this bridge be managed and customized
Taking the first question at face value, the three obvious vectors of solution are 1) the scoring program owns the interpretation 2) The VSTi instrumentals own the interpretation or 3) neither do the job.
Let’s take those in reverse order:
For a long time we all lived with 3) neither do the job. Hence, orchestral mockups were the domain of those with time and inclination to work the MIDI and available samplers with a traditional DAW. Other composers, lucky enough to have orchestras of live players available on a regular basis, simply waited until they could present the score the orchestra. There are a lucky few that fall into the camp of having an orchestra of live performers readily available. There are many that have to settle for mockups and then months (if ever) to hear there piece performed.
Most of the VSTi vendors have subscribed to having a DAW, with a human driver as the primary source of delivering the signals for playback. Only a few have given serious consideration to the idea that a scoring program can serve as the main driver; Noteperformer, Garritan (after their merger with mMakeMusic) and VSL come to mind as putting some work into it. Spitfire and Eastwest are largely engineered around the idea that a DAW or similar MIDI engine will be the driver, with the tools to capture performance subtleties, usually as MIDI control channels, from the composer or other performer working the mockup, usually from a piano-like keyboard, mod-wheel, expression pedal, etc.
This brings us to the current, point in question, what if the scoring program owned the interpretation between the score as it would be printed and the VSTi performers. This has been a tantalizing “holy grail” in the scoring industry for a while. It isn’t a simple task. Here I will confine my remarks to what I perceive as the “Big Three” in professional scoring programs (though some of my colleagues will berate me to call it the big two-and-a-half for different and conflicting reasons), and which I have some degree of personal proficiency, which are (in alphabetical order) Dorico, Finale, and (boo-hiss-Avid-is-evil-etc-etc) Sibelius.
Finale and Sibelius have taken this on through the use of playback dictionaries which can be customized by the end user, sometimes supplemented with ability to integrate simulations of human playback or recorded human playback. They have also favored some specific libraries of curated sounds to make this happen: Finale with Garritan, Sibelius with it’s own library and Noteperformer (mostly through the brilliant engineering work of Arne Wallander. Both offer open access to more VSTis through a VST2 interface, which can be harnessed with varying degrees of success through the playback dictionary mechanisms. Many enthusiasts have, over the years, tried various ways of making VSL work, with which (IMHO) mixed results compared with VSL in a DAW mockup.
Dorico, has replaced the the playback dictionary approach with their own unique attempt at solving this problem. The main function of the playback dictionary has been replaced by Steinberg’s expression maps technology and they have included the beginnings of a piano roll-type editor to ultimately support more refined modifications to the playback that can’t always be captured standardized notations or whimsical directives of the composer (“play as if riding on a merry-go-round”.) IMHO opinion, these are a good start, but are handicapped by two serious issues. First, only expression maps have been though out for Halion Symphonic Orchestra, and these are annoying incomplete and somewhat unbalanced in playback (again IMHO.) Second, the piano roll lends the impression (despite repeated claims to the contrary) that they have DAW-like capabilities. This second factor is compounded by their marketing which really over plays their hand in this department (again IMHO.) I think time will have to tell if these nascent features will live up to the hype in some future iteration of the product. For now, they don’t offer the same level of interpretive capability out of the box as the others (although I am including Noteperformer as an out-of-the box capability although it is a priced add-on from a third party.)
Of the two main approaches, Dorico has the upside of much greater interpretive capability with its expression maps and piano roll than the others, but as delivered, they have not been fully built out as far as they need to be for most who care about the quality of audio performance from the scoring program.
My personal desire would be for Dorico to get the playback system up to the level of the others in so far as it Noteperformer (or similar) level of accuracy. As much as I still need a DAW for mockups, merely synchronizing with Cubase would be more than enough for me. An underlying issue for me is that despite some wording to the effect that “playback is important”, after a year I’m seeing the course of investments aren’t really going into this area. I can live with a scoring program that is missing a lot of helps (exploding staves, proofreading and note ranges, or a full plugin interface)–I’m having a hard time with composing on a program that doesn’t sound so good orchestrally (again IMHO.) I think the team has to get this far to be fully competitive in the broader scoring market.
My apologies to LSalguerio for using his comment as the springboard for this essay. He’s probably saying to himself, “Jesus wept; how long is this guy going to go on about automated score interpretation? Doesn’t he have anything better to do?” I was bored and wanted to at least capture my thoughts. 
I am still excited for the future of Dorico, but I do wish they would clearly spell out the direction of playback. If it’s just going to be a clunky proof-listener tool, that’s OK. If they’re really interested in solid playback capability, I think some users are going to be wondering as to when, especially when the price for upgrades starts falling due.