Wednesday, January 2, 2013

HaikuStorm

This post has been lurking in the drafts pile for quite some time. You may as well read it: Hey, I'm actually making progress on my thesis. I think I'll be able to finish it. Someday...

Well, I'm back again with another song (one you will actually get to hear). This time for the Open Source Musician Podcast Tunestorm #7. A tunestorm is like a brainstorm but with music.  Everyone comes up with their own song based on some criteria. There are usually 4-8 songs submitted by different listeners with a decent breadth of genres that come out of it. Themes for the past tunestorms have included freedom, 15 second songs, using household items etc. This time the rule was lyrics in Haiku.


I was excited, since I'd never contributed to one. I heard the announcement podcast while visiting my in-laws for a weekend.  I was away from the studio, but I wanted to get started right away. I whipped out some Haiku lyrics (5 syllables 7 syllables 5 syllables, 3 verses) based on how I'd been feeling (discouraged about my thesis). I'd been listening to a bit of Pink Floyd recently and decided I wanted a slower prog-rock, sort of based on "Time."

I opened up VPMK and Yoshimi and started playing around. I wanted a fairly noisy pad sound and so I opened up the pad synth editor and started adding harmonics randomly and made a pretty slow attack and decay with some modulation stuff and I don't even remember exactly. Using VPMK I started trying some chords. I wanted it minor, not a traditional progression, but since I was using VPMK I wanted to try to have as few sharps or flats as possible. I settled on a D dorian scale with progression DmEmAmBb (If it's not actually D dorian I apologize. I know just enough theory to be dangerous).

I programmed in some fairly broad voicings of these chords into seq24 and just had it loop. I decided it was a little weak on bass so I opened AmSynth and set it to a simple sine wave for a whole note bass line.

Once I had my chords going I selected a soloing weapon: Sineshaper. I'd never really used this synth, but since I was already using most of my old trusties already I decided to give it a whirl. I was pleasantly surprised by the presets. It had a lot of good voices. I wanted to create all my own patches though so I tried playing with the controls a bit, to figure them out. I think I ended up just tweaking one of the presets a tiny bit for a mellow sound.

I needed some expression for making a truly epic solo so I started up my own TouchMIDI for some pitch and mod action. All I had was my laptop.

So then I pulled up Ardour and started jamming. At first I had Ardour set as the time master, which interacted strangely with seq24 and so only the first "note on" message was sent to the pads and bass and they became a Dm drone. I wasn't expecting it, but rolled with it and it worked pretty well just to have my little cadenza over Dm. I did that for a  while and then stopped and debugged why seq wasn't rolling. I got it going and soloed a bunch more and ended up with about 8 minutes of solo of varying quality.

When I got home and some more time I chopped the solo up for 3 minutes of the best parts. I wanted a long instrumental introduction before the main part of the song. I programmed in some drums in Hydrogen and finalized the form. I think I used the big mono kit and a snare from some other set. Then I moved on to keys.

My Casio actually has a really nice sample e-piano. Its a lot like a cp80 but a little more mellow and rhodesey. I'm not sure. Either way I really liked it, but since there's no line out I never use it for recording (except as a midi controller of course). So I set about on my  search for the perfect e-piano sample. After sludging through many unfantastic samples, I found a few real contenders. I tried Bristol's Rhodes emulation, the JRhodes3 from learjeff, and the cp80 samples from Greg Sullivan. I've had stability issues with Bristol for whatever reason, and I just couldn't quite get enough edge for the sound I wanted. The cp80 was a strong contender, but I decided it was just a bit too "clavi." Learjeff's samples are very very good so they did the trick perfectly. I added a rhythmic background and bass part over the drums using the rhodes.

Then I did vocals. I went into my "isolation booth" (closet) again when my wife wasn't around to tease me about it (I think she did come home in the middle of my session and only cocked an eyebrow). I hadn't really decided a melody yet so I had to play around on the piano and actually recorded a track to give me the notes as I practiced it. It was somewhat strange because the vocal comes in on the 2nd measure of the form (Em) right after hearing a Dm, so I kept coming in a step too low, till I got used to it. I did two takes and decided it was pretty good.

Then the only thing left was the guitar solo. I opened up Rakarrack and tried to imagine what it felt like to be David Gilmour. I had a fuzzbox patch in Rakarrack that I liked and thought sounded good through a deep reverb. I jammed over the background parts I'd recorded until I had worked out enough licks to attempt putting it on HDD. As novice as my keyboard solo skills are, my guitar chops are worse so I pretty much compose my solos in advance. I recorded the solo dry and kept most of the first take with a few punches and other edits.

Now it was mix time. I threw a HPF on every track & tried to set appropriately to leave headroom for everything. I realized I had a lot of early entrances on the guitar solo, so I cleaned that up (I'll have to work on my soloing groove), and a few places where I didn't like my keyboard work.

I compressed my vocals a fair bit just for the sound and tried to add some distortion in a send to brighten them up. It didn't quite get them to where I wanted, but I thought it did improve them. I put a short reverb to try to get better blend in the tracks, and a long one on the guitar solo. After that the levels all set pretty well. I thought there was some masking between the pads and vocals in the end. After spending a fair bit of time on EQing it, I finally just multed the pads for that section and turned them down. It worked. Maybe next time I'll try keyed compression.

I fiddled with JAMin to try to master it myself. In the end I decided I was just guessing at settings. Referencing it to "Time" went pretty well so I declared it done. Hear it in Episode 60! HaikuStorm (or: Thesis).

dot dot dot...

Well it's been a year and the podcast has gone on haitus. They're too busy and I think I might have been the only entrant, so here you go, the results of Tunestorm #7: HaikuStorm (or: Thesis). I did go through and remix it but its exactly the same arrangement and all.

No comments: