ChatGPT struggles to parse my 80 video transcript dataset

I have an Excel table with ID numbers in column A and the transcripts of each video in column B.

I’m trying to give this Excel file to ChatGPT so it can read the entire dataset and generate new mix-and-match possibilities from the video segments, for example:

NEW VIDEO = ID_003 (IN: 00:05:31 OUT: 00:08:12) + ID_042 (IN: 00:05:31 OUT: 00:08:12) + …

but it keeps getting confused and returning unrelated segments. Sometimes it even sends me IN and OUT timecodes in the middle of half-finished sentences, so the extracted clips don’t make sense, even though I think I gave it a solid prompt.

The transcripts were exported automatically from Premiere Pro and placed inside exel file with IDs, and I even tried converting everything into a JSON file, but the same thing happens. I also tried creating a custom GPT and uploading the file into the knowledge section, but then it can’t even read some parts of the transcript, as if it can’t access that information.

I know the transcripts are long and there are around 80 videos, but I don’t know what other option I have.

Leave a Reply