Hi folks – I’m working on a statistical analysis of callers that the boys identify as having the “little girl voice” and would love ideas from anyone. Presently, I’m wondering about the callers’ rate of early trauma (as reported in the call), and if the boys’ predictions about age at trauma are accurate.
Obviously, there are tons of things that aren’t being controlled in this kind of study, but it might be of some interest since there is no literature I find on the topic (as of Dec 15, 2017). I’m using CMUsphinx for phrase recognition on a 28 call burn sample for now, but I’ll analyze the full set once the details of the analysis are in place.
Specific things to measure would be really helpful!
Wow, would love to see if you can come up with scientific results here. Anyone who has listened to the show a lot KNOWS that Dr. Drew is onto a very specific, very real pattern here.
Sounds like what you need is a list of all the calls tagged with “little girl voice” on the site… Haven’t had a chance to add topic-specific call searching to the site yet, but did a little legwork and found 24 calls tagged with “little girl voice” and a few other shows where they mention it. Here are the shows:
Thanks spinfly! That’s very helpful – I made my test/burn sample based on an archive search which missed a lot of these shows.
The speech recognition tuning is going well, and these shows will help. Currently ~95% sensitivity and acceptable false positives. Based on the percentage of tags to show time, I think there are well over 100 calls to be found.
I should also ask spinfly the other admins about bandwidth limitations. I don’t want to start downloading too many shows in too short a span and cost you guys money. My email is email@example.com .
Again, suggestions are always appreciated! (I’m not a crazy person; I have plenty of publications, this is the most soft sciencey analysis yet. A labor of loveline.)
Once you’re logged in, you can manually download whichever shows you like. I’d only request that you avoid running some sort of automated script that causes bandwidth to spike unreasonably. That gets me in trouble with my host.