My transcribed ramblings can be found here, using Otter.ai

Based off of this unedited recording of what, for all intents and purposes sound like a crazy person yelling about aliens living in the walls, there are several inaccuracies and mistakes on the conventions of how I speak and what is written. It deviates from written English in that sentence structure is just straight up gone; it reads almost like “A Pickle for the Knowing Ones“, a famous (or infamous) piece of literature from the early 1800’s written by an “eccentric” rich man who utilized random capitalization, terribly misspelt words, and my favorite of all, almost no punctuation (a convention that was so bemoaned that he later released a second edition of the book, whose last pages he then filled with an assortment of punctuation for the reader to disperse as needed throughout the book themselves). The parallels between my transcribed recording and Timothy Dexter’s insane ramblings is too hilarious to not be a little bit scary. As to what is “wrong” with the transcription, there are a few things that leap out. Firstly, there is no pacing. I speak with a very deliberate cadence and speed, and that is just completely absent in this transcription. Secondly, there are several spelling mistakes: Dr. Boroditsky is suddenly Dr. Borden Ski and I feel kind of bad about that. That being said, there is the very real possibility that when listening to me speak, an audience may mishear me due to a lisp, me not enunciating properly, or simple miscommunication so I cannot fully fault the program for its spelling. The most common mistake in the transcript is the grammatical construction of what I had said; there should be more punctuation, and many more line breaks as I paused for several seconds throughout to get my thoughts in line before I continued rambling. For what the ai got “right”, however, is the fact that I did indeed say all of these things; there are inaccuracies and misinterpretations of what I said vs what was written down, but by and large it is a faithful transcription of the words that I spoke into my microphone. I do not think scripting what I was saying would have made any difference; the ai obviously struggles with pacing and punctuation, and taking even more deliberate pauses and breaks to hammer home when a period is needed vs a coma seems like it would work sporadically at best. To touch briefly on how oral storytelling differs from written storytelling, with the above experience as a framework, the most significant difference for me is the tone. Inflection, pausing for dramatic effect, all of the conventions of voice beyond simply the raw data that is passed by speaking is lost in written storytelling; there are written conventions that can approximate or generate similar effects (I said YOU go kill him vs I said you go kill HIM) such as emphasis, but there is a marked difference between telling a story and how you emphasize the important bits versus how you writer a story emphasizing the important bits. the same is true for this transcript; you cannot tell just from reading it the parts that I put more emphasis on, and what parts I just breezed through.