On Score Recognition and CPU Load During Orchestration

1. Score Recognition
I’m currently taking remote lessons from my teacher on Elgar’s Cello Concerto, but in the meantime I’ve begun programming Schumann’s Cello Concerto in A minor, Op.129 little by little.
Together with the concertos by Dvořák and Haydn (No. 2), this piece is considered one of the three great cello concertos. Naturally, it requires an extremely high level of technique—far beyond what I can play myself.

So, I turn to sequencing.
However, for some reason, I could not find any usable MIDI files for this concerto anywhere online. That meant going back to something I haven’t done in a while: creating a MIDI file directly from the score.

Fortunately, these days many scores are available as PDFs, so there is no need to start by scanning paper scores.
However, the accuracy of PDF-to-MIDI recognition varies widely depending on the software.
I use KAWAI ScoreMaker, and its recognition accuracy is, in my view, quite excellent. Once the recognition is done, you can simply export it as MIDI.
It’s best, though, to correct any recognition errors before exporting, as it makes the later work much easier.

2. CPU Load When Handling Orchestration
Elgar’s concerto required 24 tracks, whereas Schumann’s needed 18. Up until now, all my projects were under 10 tracks, so I never had any problems. But loading 24 instrument instances at once finally pushed my PC close to its limits.

So, a bit late perhaps, but I decided to introduce Vienna ensemble pro 7, which I had been considering for a long time.

This means you can connect a second PC via LAN, turn it into a dedicated “instrument server,” install your libraries there, and sequence without worrying about CPU limitations on your main machine.

With this setup, all my issues disappeared, and I was able to put an unused PC back to good use.