DEBRIS.COMgood for a laugh, or possibly an aneurysm

(Day 1 is back here)

Tuesday morning I was cleaning up the session files, removing some of the track plaque in preparation for making my final backup and rough mix. I’ll repeat that: deleting files before making a backup. Can you see where this story is going? I’ll cut to the cut: instead of erasing two aborted overdub attempts, I speed-clicked through too many confirmation dialogs and wiped out the kick drum track for two verses, a chorus, and the big drum fill I’d spent hours recording the night before.

And then I panicked. I felt a wave of full-body anxiety that caused me to jump out of my seat and wrench my Etymotic earphones out of my ears so violently that the plastic earbud of one of them stayed behind, wedged in my ear canal like a cork. This was the extreme low point in the recording process, a disaster entirely born of my own haste and mindlessness: not like getting into a minor car wreck, but like getting into a minor car wreck in a friend’s 700-series because you’re trying to beat a new Corvette off the line. Not like a bad haircut, but like a bad haircut on your wedding day from the cousin in cosmetology school. Not like dropping dinner on the floor, but like dropping the baby on the floor. Well, maybe not that bad, but still.

I’d seen it coming, existentially speaking. Andrew had suffered a system crash that wiped out tons of data. Steve had come down with a headcold that threatened his ability to sing. It was simply my turn. Bad things come in threes, they say: I was born, I grew up, and you know the rest.

After remembering to breathe, and having my wife extract the remaining piece of my headphones from my skull, I looked into un-delete software for OS X. There is only one such product, and it failed. I was a bitter, bitter man.

I checked my backups and was much relieved to discover that I did have backups for the verses and chorus. I regained a modicum of perspective: OK, so I’d have to retrack the previous evening’s overdubs. That’s a loss, but not a terrible one.

I restored the deleted tracks and played back the result. “Why does the snare drum on these tracks sound so different?” I thought. I had heard the difference before; all the song sections recorded in the B set of inputs had a different snare sound. Realization hit me like a raw fish to the forehead: I had no overhead mics in the B set of inputs; instead, I’d recorded a duplicate of the stereo toms track. The snare sounded different because 75% of the snare sound comes from the overheads, which I hadn’t recorded because I hadn’t double-checked my input assignments. So much for my backups.

Re-tracking all the affected sections only took an hour. My adrenals were working overtime. And finally I re-recorded the overdubs, achieving not perfection but something close enough for rock and roll.

Ode to Soup (final drums, final bass, scratch dulcimer, no guitar, rough mix with some compression and EQ) (Copyright © 2004 matthew mcglynn)

And then I stayed up half the night tearing down the drumkit, packing away acoustic foam, and stowing cage parts in the garage, assisted throughout by my outrageously patient, incredibly supportive, and hugely pregnant wife. Steve (he’s the guitarist, not the baby) would arrive in 18 hours.

With only four days to go before my bandmates were to show up, I knew I’d have to change my plans somewhat in order to finish the drum track for the final song, Ode to Soup. It’s a longish and somewhat complex song, because I wrote it and that’s the way music is supposed to be, as far as I know anyway.

The previous long/complex song I’d tracked (Groove95) took me eight days. I recorded it within a day, but I needed a week of rehearsals to get to that point. I didn’t have the same luxury here.

I did have a map of the song, though; last Spring I’d composed the drum track and recorded it using an electric drum kit on loan from a friend. I wanted the final recorded version of the song to feature my acoustic drums, because the electric kit hadn’t worked very well for some grooves. Two examples: ghost notes were buried in the mix, and the 16th-note hi-hat pattern during the solo section sounded like machine gun fire.

That electric-kit track was recorded in sections, as I composed them. The version of the song released here previously was assembled from the resulting parts. The assembly process was ultimately successful, but had left a bad impression on me because each edit took 10 or 20 minutes and some frustration when the joins weren’t clean — for example, when the tail of one section and the start of the next both hammered the downbeat but on different cymbals, or worse, when these downbeats were even a sixty-fourth note apart in time. During editing, it seemed it would have been a lot faster to just play the song straight through in one pass, and my takeaway lesson at the time was “it will be a lot faster to just play the song straight through in one pass.” Which is of course why recording Groove95 took me eight days.

But I had no alternatives, so I created two complete sets of input tracks in ProTools with the intention of alternating between the sets: record the first verse in set A, the second in set B, the first chorus in set A, etc. Each set had seven tracks: kick, snare, hats, stereo toms, stereo overheads. I skipped the room mic for this tune because I didn’t want to have to stitch together yet-another stream of audio.

The recording process was quick. Each section is less than 90 seconds long. It just doesn’t take very long to record 90 seconds’ worth of music. Within a couple hours I had relearned, rehearsed, and tracked the entire song, minus two drum fills that I could overdub. That was Sunday afternoon.

Monday night I overdubbed the two fills. I was so close to completing the entire five-song process that I was almost not totally stressing out. Did I mention that my wife is 38 weeks pregnant? I have been fighting a million years’ worth of nesting instinct here, turning the nursery into a recording studio for the last six weeks of my wife’s first pregnancy. I could taste my own stress, and it tasted just like your armpits after a job interview. Yeah, the one where you choked.

So it took two hours and about 30 takes to nail those fills. That sounds ridiculous, but they’re complex and I needed them to be perfect. To that end, I used ProTools to zoom in to view the waveform at about high magnification so I could see the spacing of the notes. I could see when the fill went slightly out of time, or when a velocity anomaly would cause a volume spike. Finally, the fills were perfect, and I was done.

Or so I thought…

As of Saturday morning I had two songs left to record, and not a lot of time to do so — my original schedule called for tearing down the kit no later than Sunday. But my toms needed new heads and the whole kit was ready for a tune-up. I made the mistake of sitting down with my toms and a pile of heads in front of Rush in Rio. This is an old pressure-coping technique of mine, called procrastination. Forty minutes later, Neil Peart hadn’t yet made eye contact with the audience, and I’d changed exactly one head.

I powered through the rest of my head-changing and tuning without the company of aging rock stars. It still took me a half-day.

The next song on the block was called Best In Me. It presented a different sort of challenge… not one of technique or stamina so much, but of memory. The arrangement was unusual, in that bar counts were occasionally odd (e.g. the intro is 7 bars long), the vocals in the verse start on the last eighth note of the measure, and various sections are separated by a 2-count (that is, half-measure) guitar tag.

I tried the approach that worked for Cincinnati Summer, of printing an arrangement chart and singing along in my head. It didn’t work this time; I couldn’t memorize the song structure. I’d play through what I thought was the entire song, only to hear four or six more clicks from the click track — a sure sign that I’d miscounted somewhere.

Normally, I’d ask for a scratch guitar or vocal track, but this wasn’t an option, both because I was very late in the process and because the song’s author is stuck somewhere in 1978, technology-wise. I mean, I could have asked him to cut me an 8-track tape, but I’d have had to go to Goodwill to buy a deck to play it.

The solution was to record my own vocal guide track. Hey now, take your hands down from your ears. I wasn’t singing. My guide vocal was a recitation of the section names and bar counts, a sort of voiceover for the song. It worked well, although double-checking that my guide was accurate to the demo was tedious. Had I made a mistake in the guide track, the other guys in the band would have had to rewrite their parts to fit a skewed drum pattern.

Best In Me (drums only, rough mix, dry) (Copyright © 2004 matthew mcglynn)

My performance would have benefitted from a couple more rehearsals. The final take doesn’t groove quite as well as it should. Solo’d, it works, but I was feeling the time slightly behind the beat, which may not work for this tune. Also there are a couple subtle timing errors that are a testament to my not being in the right space for this particular session. That, or maybe I was constipated, because that’s what the choruses sound like. The groove has a lot of 16th-note interplay between snare, kick, and hi-hat, and the more I listen the more I think it’s not smooth enough.

The question now is, can we “fix it in the mix?” Stay tuned.

Before the bits were dry on my final Groove95 take, I knocked out the next tune, Cincinnati Summer. It’s a simple song, and more to the point it’s short: about two and a half minutes.

Steve had recorded an acoustic-guitar version of the song for a previous solo CD, so we had a good demo to work from. He also sent a fresh demo on cassette, which I thought I might import into Pro-Tools as a guide track, but due to tape deck inconsistencies the click on the cassette demo did not match the desired tempo. I therefore started from scratch by programming a click track, and (as I had for Bleed) recorded the drums with no external guide track.

To help me keep my place in the song, I made an arrangement chart showing the song’s sections, with bar counts and lyrics. For the most part I was able to sing the song from memory, but I know from experience that a moment’s uncertainty can blow a take, so I like to have a written guide.

Getting into and out of the song quickly was the challenge here; there wasn’t enough time for a gradual build or fade. For example, I considered laying out for the first verse, to provide impact by coming in on the downbeat of the first chorus. But I thought that would make the song sound choppy, e.g. if the drums played only for the middle ~90 seconds of the song.

I settled on a couple basic, feel-good grooves that seemed to fit well. The verses are dead simple: two and four with a cross-stick. This gives a clear contrast to the chorus sections, which are built on a ghost-note feel on the snare drum and a bit more interplay between the kick and snare.

Rather than share the bare drum track, I’ve provided an MP3 of the drums and bass together. This is a second-generation recording: I made a rough drum mix, converted to MP3, and put it online for Andrew, who imported it into Cakewalk, recorded his bass to it, then mixed the result back to MP3. This is modern asynchronous multracking at its finest. And it’s a hip bassline too. Check it out:

Cincinnati Summer (drums and bass only, rough mix, dry) (Copyright © 2004 matthew mcglynn & Andrew Thomas)

It was during my frantic tracking of this song that I learned that Andrew had suffered a catastrophic hardware failure, wiping out at least a few recent recordings, or at worst a few years worth. I had been making backups, but even so I did what anyone who hears a data-loss story should do immediately: I made another one. I burned all my audio thus far to CD. I mention this only to set up my own data-loss story, which you’ll read in a few days’ time.

Andrew sent two bass tracks for Groove95 — one bass and one melody (played high on one of his extra-stringed basses). He sent no suggestion for the drum part, even though he’d composed the tune to a specific pattern on his drum machine. He wanted to see what sort of groove I’d come up with without being influenced by the programmed pattern he’d initially used.

The melody was a great, bouncy line, with a loopy pickup that comes in on an offbeat. I struggled to come up with something equally cool… and then anything cool at all. Finally I settled for something that fit, but was basically uncool. I wasn’t pleased with it and told him as much. He offered to send a sample of the drum machine pattern he’d been listening to when he composed the song. I was grateful, because I liked it much better than what I’d composed.

The groove was a half-time shuffle, one of my favorite things. But it was really fast, as fast as the old JAR tune Pandemonium Clockwise). I felt my schedule shifting, because I would need time to work my shuffle up to this tempo again.

I decided to record the song with dowels rather than drum sticks, in order to get a lighter sound than I would get with sticks. I started with ProMark H-Rods, which have a nice bounce, but ultimately switched to Vic Firth’s Rute model, which is more durable although uncomfortably fat in the handle and crippled by a dumb sliding band that moves around during the song. (Note to self: well-designed dowel sticks should be the first “signature series” item, just as soon as that stick endorsement comes through.)

I worked that damn shuffle for a week. Days I thought sure I’d be able to record a final take became rehearsal days, when after a few hours of takes I had taped nothing I could use. I spent a whole Saturday this way.

On Sunday, my first pass was perfect, but some of my acoustic foam came crashing down during the second chorus, blowing my concentration and, two bars later, my flawless take. Still I managed to blast through a couple of strong performances by mid-afternoon. No single pass was good enough, but the first half of one plus the second of another would yield a decent result.

With all my rehearsing, I’d worked out a couple of neat fills that were tough to pull off. The challenge of a complicated song like this, especially one as long as this (6:40), is deciding whether to risk an otherwise good take by blowing a fill late in the song. I don’t care what the magazines might lead you to believe; punching in to fix drum tracks is very tricky. Seamless patching when 8 mics record one sound requires a delicate touch.

My solution in some cases is to leave a gap in the main take, then overdub just the fill. I did that for the two longish fills in this song.

The end result is about 90% of what I wanted it to be (and I was shooting for a Simon Phillips/Jeff Porcaro feel, i.e., far beyond my actual abilities). As usual, my favorite section is the one that I improvised, around 05:10 where the groove opens up but matches the rhythm of the bass (which you can’t hear in this sample, alas). Also I pulled off another three complicated dance moves that were giving me occasional fits — short passages requiring smooth 4-way coordination.

Tracking this song taught me an interesting lesson in concentration. Passages I could play without thinking about I couldn’t play when I was thinking about them. I play better on autopilot.

Anyway, the final take, presented dry and in genuine rough mix fashion, was assembled from two passes plus two tiny overdubs. I’m not happy with the snare sound yet, and there seems to be a sonic disconnect between the top of the kit (e.g. the china cymbal during the chorus) and mid/bottom (snare and kick), but overall I’m happy with it.

Groove95 (drums only, rough mix, dry) (Copyright © 2004 matthew mcglynn)