you can use different inlets, funnel, or mc. ... how many voices?

What I can contribute is a few javascript methods from my multitimbral javascript synth. Sadly js buffer peek and poke methods do not compile to standalone, so I dropped it. But it could provide a starting point. This is an LRU voice allocator for multiple voices on different channels with age-based voice indexing for best performance. This is part of 3,000 lines of JS code and I haven't got a simpler version at hand, apologies.

function storenote(pon, velon, chan){ //add note to buffer var a = anum[chan]; if(a < vmax) anum[chan] = a +1; apitch[chan][a] = pon; aveloc[chan][a] = velon; } function deletenote(ageoff, chan){ //remove note from buffer var a = anum[chan]; var b; var c; var d = 0; for(var i = ageoff; i < numv; i++){ c = aveloc[chan][i]; if(c){ b = d + 1; apitch[chan][d] = apitch[chan][b]; aveloc[chan][d] = aveloc[chan][b]; d++; } } apitch[chan][a] = 0; aveloc[chan][a] = 0; anum[chan] = a -1; } if(vel>0){ storenote(p, vel, chan); numv++; }else{ for(var a = numv; a >0; a--){ v = age.indexOf(a); if(p == pitch[v] && chan == channel[v]){ pitch[v] = 0; channel[v] = 0; deletenote(a, chan); numv--; return; } } }

Thanks for your replies, I managed to figure out the issue!

After much head scratching

As always when granular synthesis is involved, I'm a bit curious... What solution did you find out ?