On Tue, Dec 9, 2014 at 9:06 PM, Brent Busby <brent@keycorner.org> wrote:

On Tue, 9 Dec 2014, Rui Nuno Capela wrote:

there aren't plans to have separate outputs for each element, here called like so as each map to a sample triggered per midi note key and only.

why not use several drumkv1 instances? each one loaded with its own set of sample elements (if more than just a single one (you might keep an exclusive mute group under the same instance eg. open and close hi-hat)?

i know it might seem a ridiculous solution, but hey you're not wasting nothing much in resource terms, because each instance only loads the sample(s) it is supposed to run anyhow ;)

That would certainly work, since really the main thing is that they should all be run by the same sequencer. And I suppose on modern computers, you'd have to run a lot of Jack clients to add up to a serious load doing almost anything other than DSP processing.

For the record, I am also much more in favor of the approach outlined by Rui too. It has one disadvantage, which is not inconsequential: lack of integration at the track level for different "channels" of the MIDI data. Each sampler instance is in a different track, being fed data that is conceptually distinct from the others. There are ways to work around this, but for the user, it isn't always the right way to work. If the user really wants to be able to mix the relative levels of the snare vs kick *within the host*, but also edit the same drum MIDI data in the same track, then separate outputs from the same sampler instance is really the one sane way to do this, but alas, even this is not so sane.

In ardour, the right way to approach this would be to use a cloned MIDI region or shared playlist in each track. Edits to 1 track will affect them all. Then filter by channel, or simply don't load samples for irrelevant note IDs into the sampler for each track.