Farewell (case study)
After a long break from making music due to my thesis, here I am again! This time I’ve decided to share my workflow on a recent project. You can listen to the final result below:
The song is part of the Vision of Escalowne‘s original soundtrack (third CD album). I’ve seen this anime more than a decade ago, made cassette copies of my sister’s CD, and listened to it regularly back in highschool. For a cartoon, it has a soundtrack that goes beyond what you would normally expect. It’s one of the reasons I got hooked into orchestra music (as opposed to the serious and profound classics that is less accessible to my less mature mind back then). It’s composed by Yoko Kanno and Hajime Mizoguchi and (probably) performed by Warsaw Philharmonic Orchestra. The score was based on this transcription by ThePochaccos and all thanks to him/her for doing a great job.
Unfortunately it is not a public domain score, so I won’t be able to share it that easily. But basically, I transcribed the video using MuseScore. It is not that difficult, there are parts in the piano that need “voices”, but the whole piece is mostly strings (for me., that’s actually easier and more fun than emailing the youtube user). (I hope to share something public domain next time).
I’ve been itching to get Reaper for a long time already, but so far, my projects are not too complex, so this setup is still fine.
Sonatina Symphonic Orchestra
For this song, I decided to do a proper demo of SSO. I’ve always used multiple orchestra samples and layer the different results in Audacity. Using SSO by itself has also revealed some of its shortcomings. Sometimes the release does not sound well, giving an unnatural sound at the end of the note (i.e. in long contrabass notes or in violas). I work around this by shortening the note until the odd sounding part is no longer audible. The slow attacks also made it less favorable for fast short notes, making them sound mushy, and making the melody less defined. I try to remedy this by increasing the note velocities or decreasing the note velocities of the background instruments. I wonder whether using a VST SFZ player, instead of Syntfont’s native sfz support, might give better results, but I have not really explored this option.
Update: As it turns out, this manual looping workaround was unnecessary ans I should apologize for the misinformation. My mistake was to directly load the sfz file to Synthfont (an older version back then) instead of using an sfz playing VST. SSO strings loop nicely with Plogue sforzando. The text is maintained for historical purposes and as it may still be helpful for other libraries that do not have looping in them or, for another application, to hide the distinct repetitive sound of looped samples.
Another shortcoming of SSO is its lack of looped samples for some instruments, which has forced me come up with a tricky work around. The first note of the first violins is 9 bars long (31 seconds at 70 BPM!). At first I thought of editing the SFZ samples, but that seemed overkill for a single note. So next, I imgained how a real orchestra would actually play a half minute note. If it were a single violin, the maximum amount of time you could slide the bow over a string would be limited by the bow’s length and the minimum bow velocity needed to produce an acceptable sound, maybe five or 10 seconds (my imagination’s approximation). Restarting the bow slide would have made a new note. But an ensemble of more than 10 violins doesn’t have to simultaneously restart their individual bowing. So while one violinist restarts there are about 10 others who are still bowing midway, hiding the restarting guy and creating an illusion of continuity. That’s my guess.
Back to the MIDI editor, I implemented “manual looping” by making an extra first violins track (not to be confused with the second violins in the score). I broke the whole 9 bar note into shorter segments that can be played by SSO. The extra violin track continues the note when it’s about to end in the original violin track. Then the original violin track continues the note when the extra violin track’s note is about to end. Hence, by alternating and overlapping these two violin tracks for the same note, I get a manually looped violin note. To mask the attacks of this repeating violin, I align them with the attacks (note start) of the other instruments in the score. Of course, these alternating violin tracks must have the same volume and panning and go through to the same effects chain.
It may also be worth noting that the SSO updated sustain violins worked better for this trick.
Guitar = Guitar Pro
Since I can’t compete with a real orchestra, I generally avoid making inferior copies of something that it is already great (except for personal studies or demos). Who would listen to that? At the very least, I would change an instrument to give a different feel that is worth listening to. Hence, I changed the piano part into guitar. Being a more common and accessible instrument, and being a long time guitar player myself, I could relate more with the sad sound of a guitar.
I’m also known among my friends for advocating Guitar Pro (GP) as a virtual guitar addition to their DAWs. Even though I play guitar well (used to?), and own many guitars (too many to remember), recording guitars with my limited laptop studio setup has never given me satisfying results. GP actually started as a tablature study program (coincidentally at the age when I was crazy about studying guitar tabs). When it started out, there was a free alternative that can do as much, Tux Guitar. But since Guitar Pro’s introduction of RSE (realistic sound engine), it has, in my opinion, left Tux Guitar far behind. GP would not integrate with a DAW like a VST or soundfont, but it’s notation based interface, optimized for guitar articulations, makes it far more intuitive than any VST I know. GP can simulate vibrato, hammer on/pull offs, ringing, chord “brushing”, harmonics and many more with a few mouse clicks and without having to tweak MIDI parameters. And the demos sound realistic enough for me (listen or download here). It’s probably the guitar equivalent of Finale + Garritan combo, but at a price below a hundred dollars. Software that unify sophisticated music notation and virtual sound production is really something we should be thankful for (although I would also hope for piano roll integration).
There’s nothing really new here, but I would be happy if a newcomer in digital music production/midi orchestration would learn something from this. Note that the only tool that costs money is Guitar Pro, although I’ve also donated a small amount to Synthfont as it is very useful to me and it was the first thing to exactly match what I was looking for before, a simple tool that applies soundfonts and VSTi’s to an existing MIDI file that is not as overkill as a full blown DAW. With diligence, passion and knowledge of what great tools are available out there, making quality music, one that you can mix in to your iPod or MP3 player, is no longer a thing that can only be done with professional thousand dollar studios.
Happy music making!