Hi and welcome!
Thank you for sharing your work, Nyarlathotep is surely pleased too
In the recent update there are another two variations of Clouds (#342, #343) but I think the first one is a bit broken, it stretches the sound too much (some of its controls don't react, I guess). Still creates interesting textures after a while so I decided to leave it there.
You may also like the new polyphonic reverb, #322, which creates effect closer to a vocoder (as opposed to the monophonic Clouds). There is no true vocoder in Gecho - maybe there will be some day, however its requirements on the CPU are presumably higher, so we'll see. The basic soundscapes are playable by midi too, and if you turn down the background sample (i.e. water or forest) then only the incoming signal will be filtered by resonant filters, which creates something remotely similar to vocoder too. I see you tried granular already, perhaps it's harder to use that way, will probably not work well with speech but with monotonous singing, it should be possible to turin the voice into chords as well.
About the noise - can we rule out a ground loop? not sure if you meant "different powering" as trying various sources, or that every machine has separate and isolated power. From my experience, Volcas may have faint inherent buzzing, it depends on the power too. Maybe the auto-gain function in Gecho is amplifying it too much when nothing else plays (sounds like the buzzing is buried under the signal later?) Try to disable it, here is a searchable guide for all those options:
http://gechologic.com/settings_leaflet_explained (check the "input levels" paragraph that explains how to switch it off).
Hope this helps!
