I might be misunderstanding and way off base here but…
I’ve quite a few scores that emulate ‘legato’ by extending the previous note a little past the start of the next note.
For some synths/plugins this can cause a problem if the same note is repeated using this legato technique. It’s no problem as long as the ‘next legato note’ is NOT the same as the one just before it, or if the instrument is NOT a polyphonic mode instrument. So…if you’re using that particular legato-emulation technique…
First try increasing the number of voices your plugin allows for itself (more on that later), and if that doesn’t work:
Make sure that a note doesn’t overlap itself. I.E. two C2 on events coming before any note off event: C2 on, then comes another C2 on, then a few milliseconds later a C2 off comes and kills BOTH instances of C2 playing…
As for ‘sound dropping out and what seems to be CPU spikes’, consider this:
When using this legato technique, even if you don’t have a note following itself…it can quickly ‘double’ the amount of voices or polyphony the track needs to play back without ‘notes or events getting stolen’. Effectively, that ‘overlap’ period used to create the legato effect require double the number of voices as the same track with no notes overlapping. So…if your plugin allows it, double the number of voices/tones it can play back.
Just as an example…Halion 5 comes set to 128 voices out of the box. A typical classical score (I.E. Beethoven’s 3rd Symphony) requires around 300 voices when I use this ‘legato effect’ on the string and wind parts in combination with the HSO soundset. In contrast, if I use a logic editor to get rid of all the ‘emulated legato’ (overlapping)…the voice count is less than 200, but STILL more than the default 128. So, right out of the shoot, I know I need to bump up the number of voices H5 will allow itself in the H5 ‘Options Tab’. If I neglect to do this…it SOUNDS BAD…as in drop outs…and possibly even ‘stuck notes’. Once I bump up the polyphony limits for the plugin…it sounds great and even on my ancient AMD Phenom with only 8gig DDR2 memory…it’s good to go for many more tracks of VSTi voices and VST processing plugins before I need to consider rendering some tracks down to audio.
I’ve some plugins that just do NOT like overlapping legato at all. In most cases I only need to watch out for back to back occurrences of the same note (make sure those do NOT overlap)…and otherwise it’s OK. Also…it really does not matter if Cubase is the host, or something else (like Sibelius)…some plugins/synths/samplers just do NOT like overlapping occurrences of the same note…and any of them will need ‘double the number of voices’ available if you use the overlapping legato technique.
For me, as a general rule of thumb…if I’m working with passages that need a lot of ‘legato’…I’ll go in and make a patch that adjusts the attack and release times with a pedal or key switch rather than using the old ‘overlapping notes’ approach. It’ll still need MORE voices than a non legato/connected phrase. Of course, if I’m strapped for time and must deal with a misbehaving track that uses the ole fake legato tactic…I’ll just work up a quick logic editor that makes sure back to back occurrences of the ‘same note’ do NOT overlap.