The fluidsynth implementation doesn’t fit our needs, as we are forced to spawn FXs processes for every channel (16 reverbs + 16 choruses = 32 processes), wasting resources . OK, that is not totally true, as we only spawn 8 reverbs + 8 choruses, so some channels share reverb and chorus (1 & 9, 2 & 10, etc.), what reduces the CPU load, but it adds complexity and it’s difficult to explain and remember. In other words, it’s a mess and it’s still wasting resources!!!
Fluidsynth internal FXs are not very good. We have much better reverbs and choruses available on the list.
The other soundfont engines (sfizz and linuxsampler) have no internal FXs and it doesn’t seem an issue. We like the idea of having the same behaviour for all soundfont engines because it simplifies implementing a “unified soundfont wrapper”
Anyway, we don’t want to make this movement without asking the community, specially those who are actively using fluidsynth. So, please, what do you think about it?
FYI, this discussion comes from here: