I made (or attempted to make) an infographic of things that a viewer is likely to hear Lowco say during a cast. However, I'm not a designer, so the resulting quality is less than high. So, the data and the graphic are more for novelty than anything else. Anyway, here's the long version of what happened: I was bored at work, so I was watching Lowco VODs and testing some proprietary voice transcription software at the same time. The software can learn to recognize a specific voice and generate statistics on word usage and tonality, like a more complex Dragon Naturally Speaking. These are the data used in the infographic in a spreadsheet on Google Docs. The data is taken from ten, non-overlapping half-hour segments of VODs -- two segments each from five separate days. It's kind of basic. It's just ratios and means for different variables. I didn't determine deviation, variance, or distribution because reasons. If you're wondering how the key phrases are selected, the software picks certain phrases based on frequency, regular interval, and emphasis by speaker. Speaking of emphasis, the category of reading aloud from sources is determined by tonality. The software recognizes conversational acknowledgement (reading chat) versus document recitation (reading from the game). Also, I picked some categories based on what sounded like frequent usage. Here is a transcript of all ten segments. Any typos in the transcript are me trying to make the document human-readable. The software generates one long string with marking codes, weird capitalization, and questionable punctuation. There isn't quite a front-end GUI for this software. It makes it less than useful in most circumstances. Lastly, I dumped the transcript into one of those online word cloud generators. The online generator isn't the best as it didn't combine identical words because of punctuation ("umm" versus "umm,"). Still, you can generally see patterns in the word cloud: -What game Lowco was playing (Fallout: New Vegas) -Gregle is the most mentioned person in chat (20 instances over 10 samples -- an average of two mentions per half hour). -Because Lowco reads chat out loud so much, "Lowco" is a common things that Lowco says, referring to herself in the third person.
