The Light of Day

The sound of somebody not actually singing something
KeithHandy posted in Producing on April 26th, 2008

This is a short snippet of a song that was excluded from the 1998 CD of the rock opera, and is being re-included on the restoration:

That is Kim’s voice… what’s particularly neat about it, though, is that she never actually sang that. Not even some rough version. She never sang that bit at all. Ever. Not back in 1998, not just prior to me posting this, and not at any time in between. But that’s her voice.

You think I’m playing mind-fuck games with you and trying to frustrate you, don’t you? I’m not. That bit was constructed syllable by syllable, by raiding the other five songs she sang on for closest matches (I called it “playing Syllable Bingo”), using Praat to manipulate pitches and durations, and relying on a shitload of trial and error to get the pieces to fit together and sound continuous. Now that you know it’s cobbled together from a series of manipulated samples, you can probably hear that it doesn’t quite sound 100% natural… but, all things considered, I think I got it pretty damn close.

The “Syllable Bingo” step was madness in its own right, even before all the tweaking and molding. I mentally scanned the lyrics on paper while repeatedly listening to existing recordings to find and mark possible matches, and built a crude mock-up without worrying about all the pitches yet. Eventually it came down to a few nasty hard-to-find sounds, which forced me to think hard about how we say and hear certain vowel sounds in certain contexts. For example, in “be afraid”, “be a” has to be a continuous sound, and I believe that came from the word “realize”. The word “memory” contains parts of three words: “remember”, “prisoner“, and “free“.

One thing that did not work (and believe me, I tried), no matter what, was to try to be clever and turn syllables backwards as a last resort. A backwards syllable sounds like a backwards syllable, no matter how short it is. It’s amazing that our brains can call shenanigans on this so quickly.

After gathering, sorting, and whittling down the final sounds to be used, I had to tune and stretch them… and, in some cases, flatten the pitch of two sounds so that I could crossfade them without making a flange-like sound… and then re-pitch and re-stretch, and so on.

What motivated me to do it this way, when most reasonable people would have tracked down the singer or sought a voice double? Well, what motivates you to not do this sort of thing? This is the kind of challenge I like to pose to myself. Sometimes I enjoy approaching art as if I were solving a puzzle. The results and/or sense of accomplishment must feel rewarding enough to me, otherwise I wouldn’t keep starting things that I know are going to be so difficult. And it’s not like I spend hours and hours feeling nothing but frustration until it’s done — each small thing that I get right feels good to me.

More pragmatically (in case I need to answer to the funnyfarm-mobile), using previously existing tracks as raw material helps to keep the continuity, being that it’s the same person, at the same age, at the same microphone and on the same magnetic tape. As a bonus, the whole process gave me a great idea for how to convey that section in the film script. So I’d say it was a weekend well-spent.

Yes, “a whole weekend”, if you choose to word it that way — though I prefer to say, “just a weekend”.