I've lately realized that, while I follow a lot of other jewelry-makers' blogs for techniques, inspiration and tips...that stuff is only interesting to me because I am also a jewelry-maker. And even then, the more interesting blogs have other stuff thrown in, too. Someone's scruffy husband being a dear and fixing the car on the dramatic road trip to the bead show. Someone's apple-cheeked kid spilling ice cream on himself. I'm a sucker for that stuff, and it makes them seem more like real people as opposed to just really-awesome-jewelry machines. I love jewelry, but I will never be a machine. Someone I can relate to is much more interesting.
And I, like every other jewelry-maker and -seller, also have to realize that my target audience is not, in fact, other jewelry-makers; it’s jewelry-buyers, or people who will tell other people about my jewelry. And I can only assume that a string of neat-looking stuff will not hold their attention as much as a real person’s life, because the same is true of me even though jewelry is one of my favorite things. Bottom line, it wouldn’t kill me to show some personality on this thing. Hence my “cooking for absolute idiots” posts showcasing what an absolute idiot I am about cooking. If anyone’s listening, stir-fry is NOT THAT HARD, I SWEAR.
Well, I don’t have a scruffy husband or an apple-cheeked kid...and dear, camera-shy Boyfriend of Mine would probably object to me putting his picture online...but I do have a life besides jewelry. In fact, it’s been taking up all of my time lately. When I said “It would take me 3 pieces a week to catch up” I didn’t mean that I would actually manage that. I knew it wouldn’t happen when I said it, at least not soon, and just recently I started my analyses in earnest, so it definitely won’t. But it is interesting nonetheless. (THOUGH I DID MAKE TWO PIECES TODAY, YES TWO. PICTURES UP SOON.)
So, prepare for a long, laughing, overdramatized account of my adventures in...(dun dun dun) LABELING SOUND FILES!
For anyone who doesn’t know (which is most of the internet) I’m a linguist by major, and my research for this summer has been studying intonation patterns in a production study. Basically I get people to talk into a microphone and then figure out how getting them to say certain things rather than others affects their pitch. It’s the kind of stuff that’s fascinating if you’re a linguist, and if you’re not, you don’t have any opinion about it, so how ‘bout that weather?
But I’m learning a lot of stuff that bears thinking about even if you’re not into intonation, phonetics or any sort of linguistics at all. General research stuff, like: How do you decide when something is an error and when it’s a data point? But back to that in a moment.
My analysis work so far has mostly been prepping things for computers to pull data out of them. I go through sound files and label certain parts, like high points in the pitch and where the vowel starts. Sounds easy, right? Until you realize that I went through a three-day stretch where all I saw was this:
Which by the way, zooms out to this:
Which is fine except it zooms out to this:
And a couple more times, to this:
Which is not quite as large as the whole thing:
Those are 1) the level I work at, 2) zoomed out twice from there, 3) zoomed out twice again 4) zoomed out 4 times from THAT point, and then 5) zoomed out a whole three more times because that’s all of it. Yeah. Eleven zoom-outs total. Notice the way the pitch track disappeared after the first one? That’s because if I have it calculate all of that even if I’m zoomed in to like 2 seconds, it slows down my computer. I have to zoom into ONE SECOND to actually see what I'm analyzing...and the file is ten minutes long. Long story short, I spent 22 hours straight staring at a screen that looks very much like 1) up there, got 8 hours of sleep and then went back to it for another 16. And I only got to halfway through subject 5 before I burned out. There are 6 subjects...I’m taking a break and doing some data-pulling, thanks.
Doing stuff like that really starts to mess with your head after a while. Somewhere in the middle of figuring out where the d ends and the m begins in this squiggly line:
(Hint: The cursor’s on it. Couldn’t tell? That’s okay, it means you’re sane!)
I realized that in trying to label the end of the consonant, I could not remember what consonant “consonant” started with. At one point I freaked out wondering, “Where’s the consonant at the end of ‘be’?? I can’t find it anywhere!” Only to remember that it is, and always will be, in the hammer space.
After several hours of each, I started to get thoroughly sick of nearly everyone’s voice (Okay, just the nasally ones...subject 2 was actually rather adorable). And of course, each and every subject came up with a special individual problem that I had to figure out how to deal with...one subject inexplicably changed all his Ds to Ns – don’t ask me, he doesn’t do it in person – another had a mysterious extra pitch peak that made her samples all look like little hearts, which was cute and all, but I have to label one of those peaks, which one? WHICH ONE? - another subject had weird lilts at the beginnings of the parts that I was labeling that made it impossible sometimes to make an accurate label...and even when it was possible, it was nerve-wracking because every time, I could be wrong. Here I was labeling important parts of the intonation contour, and suddenly this bizarre thing happens and I have no idea what to do...and then each subject proceeded to make his own individual weird thing happen like EVERY OTHER TIME. And since my mentor was on a two-week vacation and is absolutely terrible at answering emails even at the best of times, let alone while hiking with his kids...Yeah. Bad news. Some freaking out happened. When I got too crazy to label anymore, I ended up labeling a lot of them “m” for “help Me please” even when in my normal state they seemed totally fine. And that was for the ones that are totally fine.
Which brings me, by the way, to my previously-mentioned dilemma. How do you tell when something is an error, some outlier to be not counted with the rest...or a data point, a piece of useful information? How do you know when it’s important? For instance:
This is what the contour I was working with looks like:
So what do I do when all of a sudden it looks like this instead?
And there were a lot that were less obviously software problems. Do I assume “That can’t be right, it doesn’t look like the rest of them” or do I have to say, "Well...that’s the way he made it, I have to take that into account.”? Most of the time, it’s the first one...the software I’m using is notoriously unreliable when it comes to tracking pitch and is known to change its mind on what’s what pitch depending on things like how far you’re zoomed in, how far left you are of the pitch in question, and what settings you’ve chosen to display the pitch at. It’s usually reasonable, if something looks bizarre, to assume “That’s an error.” Especially the cases where you can see a normal contour under that, looking all groovy, and there’s just a few artifacts screwing it up. Or, if not by virtue of the software, it’s entirely plausibly throw-out-able just because the person got tired and said it wrong. A lot of them switched into listing mode, for instance, when we wanted a pitch contour that sounded like a standalone statement.
But then there are the brain-stirring errors that completely change everything and you have no choice but to just go with it against your will even though you know the sentence doesn’t sound like it looks like that and why is that bump there, but it’s too big to just be an artifact, what’s going on? MOM?!
My “m” labels go through tides of desperation with my sleep cycle...near the end of my “functional” periods there are lots of notes like, “That has to be an error, why would it look like that?”, “Should I be worried about this?” and “This one's high in a really awkward place, much like a stoner at a funeral.” Part of my way to keep myself sane was to think up a new emoticon denoting devastation whenever someone said it in a way that we’d have to throw out, like using the wrong contour, or coughing in the middle. If my mentor ever actually reads those (unlikely), I think he’ll get a new perspective on the art of iconography. They included upside-down, upside-right, backwards and to either side.
Halfway through Subject 4, I got the idea that I’d blog about it afterward...and that kept me going through another half or two of data set, stopping every once in a while to take screenshots. Every annoyance or delay was like “BOO-YAH. ANOTHER INTERESTING SENTENCE. I CAN TOTALLY HYPERBOLIZE THAT.”...including that one.
I still burned out.
Not to complain, though...it was just a strange time not many of anyone who’s going to read this will ever experience...and it’s probably the easiest job I could have that would still be intellectually stimulating, and the schedule is lax enough for me to still be awake at 5:00 am (now)...and if I had left myself more time to do it before it was necessary to get crackin’, then I wouldn’t have had such crazy days, but actually, it was rather fun. I’ll try not to do it again, but if I could go back and have the chance to do it once, I’d still do it the first time. I feel like that’s an experience I ought to have had, doing almost-grad work the summer after my sophomore year. And I’m learning a ton! Pretty soon everyone (who works in intonation theory) will start hearing about Tonal Center of Gravity – and if anyone asks, that’s my mentor’s idea, oh yeah. I know that guy. I could get you an autograph...if you’re nice.
Oh and also, Boyfriend and I had a second anniversary. That was fun. More rain than expected. Less being outside than expected. Therefore, more cozy than expected, so it's all good.
No comments:
Post a Comment
Thanks for letting me know what you think!