Another impressive demonstration of playback from Musescore
I just completed my latest score, called L’Arrivée. It’s an orchestral transcription of the first movement of a piano suite I wrote in Montréal some years back. The YouTube score and midi performance are here.
As forum members know, I’m devoting a lot of time to showing what can be accomplished playback-wise with Musescore. My dream is to be able to set up a full midi orchestra that I can use with confidence when running Musescore. I’ve still a ways to go because audio processing under Linux is a tangled mess of partially complete, abandoned, and buggy applications. I wish this weren’t so, but it is, and since I run only Linux, I’m stuck with it. The setup for L’Arrivée was Musescore attached to Linuxsampler with individually loaded soundfonts for each instrument. The output of each Linux sampler channel was connected to various plugins (equalizers, amps, reverbs, etc) which had been loaded in a plugin host (Carla, from kxstudio). Output from Carla was piped to a mastering tool (jamin) and then to a reverberator (zita-rev1). The whole thing was cabled together in Claudia (also from kxstudio).
It took hundreds of hours to create the score and final playback, and again, while I wish it weren’t so, a lot of those hours were the result of problems in Musescore 1.2. Not to put too fine a point on it, if I have to use a screen ruler to eyeball the placement of elements in what is essentially a graphics program, something’s wrong. Aligning dynamics, hairpins, tempo markings, text and extenders (e.g. “dim......“) *by hand* across fifteen staves in a 200-bar score is an insane amount of work, made even worse by the fact that small changes in the score can knock them out of place. I surely do hope that 2.0 fixes dynamics/hairpin/text placement issues, group select, and constrained moves. The only word to describe them in 1.2 is “broken”.
I would very much like the Musescore team to reconsider its attitude with respect to playback. The official line, “Musescore is a primarily a notation program”, ignores the fact that playback is part of Musescore. Limited, to be sure, but limited doesn’t have to mean awkward or counter-intuitive to work with. I nearly went blind right-clicking on thousands of notes in order to change the Note Properties, all because you can’t group select notes and change the offtime offset without wiping out your individually-tweaked velocities of the same notes. It’s a Herculean task. Equally, it’s insane not to include volume along with velocity in the Note Properties dialogue. Imagine this scenario: You’ve individually tweaked the velocities of a 21-note phrase, but later have to correct the balance of that phrase within the orchestra by having it played a little louder. Yes, dear readers, that means right clicking on every one of those 21 notes and increasing their velocities (and praying you get it right, which you most likely won’t, meaning you have to redo the same 21 steps repeatedly until you do get it right). Simply changing the volume would take only a minuscule fraction of the time.
Achieving realistic playback of a score will always be a “silk purse out of a sow’s ear” scenario. I do understand the Musescore developers’ decision not to go hog-wild on midi playback. But they have provided enough playback facilities for anyone, not just people with OCD, to realize their scores decently. In my case, Musescore lets me share with the world music that might otherwise never be heard (who can afford to hire a symphony orchestra for an afternoon?). Am I wrong to think this as valuable a part of Musescore as beautiful notation?
All that said, do have a listen to L’Arrivée. I hate to blow my own horn, but it’s an impressive demonstration of what you can achieve with Musescore (and, I hope, supports the points I've made).