I am starting to be comfortable with the mechanics of creating expression maps, setting up endpoints, and creating and applying playback templates. Now the question is how to organize things for both maintainability and performance. My goal is to build a big band playback template that for starters primarily uses Garritan Jazz and Big Band 3 (JABB) for sounds. Over time, I may want to add sounds from other libraries. GPO or their world percussion library would be prime candidates, but it could be any SFZ. Libraries that are not Aria compatible require their own endpoints so my questions mostly relate to how to organize the Aria endpoints.
Endpoint granularity is a design choice with both usability and performance considerations. For maintainability, having separate endpoints for trumpets, trombones, saxes, rhythm, drumset gives me the ability to organize sounds by ‘family’. Or I could get started with the minimal number of endpoints needed to host all of the big band instruments. Is there a performance hit for using sparsely populated endpoints? If you use the minimal number of endpoints, I believe you need to add any new instruments to the end, so that leads to having the instruments for each big band section potentially spread across multiple endpoints. Not ideal from a usability perspective.
JABB contains complete samples for several instruments of each type. E.g. 5 trumpets, 4 alto saxes, etc. If I set up an endpoint that uses one slot for each of the 5 trumpets, will Dorico assign the lead part to #1, 2nd part to #2, etc.? Or is it better to do what I am currently doing: define one trumpet, and then after Dorico assigns the same trumpet to all of the players, change the sample being used on a player by player basis.
Dorico doesn’t currently have automatic support for ordinals, so you can’t explicitly specify that e.g. Trumpet 1 should connect to one patch and Trumpet 2 should connect to another one, but what you should find is that if you set up five trumpets or whatever in your playback template, Dorico will assign each trumpet in your project consecutively to the available trumpets in the playback template, then add another instance when it runs out.
Thanks Daniel! That will be very convenient.
What is your take on endpoint granularity? Does my organizing by family make sense? Are my concerns about performance justified?
There is always going to be a performance penalty for loading plugins and samples that you don’t use. However, it will depend on the plugins and your available RAM as to whether there is a noticeable penalty. Loading time and RAM usage would probably where you notice it the most. If you have multiple plugins loading the same instruments then it will usually add little to RAM usage because most plugins are able to use the same samples
In general, I tend to suggest creating endpoint configurations for a section at a time, eg String quartet, orchestral strings, orchestral winds, etc, so that if you have a string section with divisi then you can enable separate voice playback and it will load another set of plugins just for the string section. I’m not sure what to suggest for your JABB configurations - I think it depends really on the ensembles you write for. If you always write for the same basic ensemble size then perhaps create a single whole band template. If every piece uses a different size band then maybe break down into trumpets, saxes, etc, so that if you need eg 8 sax parts then it’ll just load a second instance of the 4 saxes.