Video as knowledge documents

I wonder to create a GPTs where the training materials would be the KNOWLEDGE. So my question is:
is it possible to upload a video (mp4) directly in the GPTs?

You can generate captions for your video and submit them as knowledge to your GPT

Thank you. So we mean to have a script of wording and after that to upload the text file into GPT. Correct?

Also take periodic screen captures, have GPT-4-Vision describe what is in the frame, and use that in your knowledge files / embeddings.


Yes, that should be working. You can upload captions as knowledge

Curious if linking or uploading video content is anywhere on the roadmap. My video content is HIGHLY visual and screencapping and describing every frame of a hundred hours of video sounds…like a ton of work :sweat_smile:

Considering waiting it out for incorporating video content…if that’s even in the cards. Which considering how fast things are going, seems like this might be a feature any moment.

For anyone looking to do this today, google thepipe, works for MP4 and YouTube

Thank you. I visite google thepipe’s site. I have not clear if it can produce a script from a mp4 speech.

1 Like

Hi @cinciripini , it can indeed produce a script from MP4.
As stated in the Docs: “Video: .mp4 , .avi , .mov , .wmv: Extracts frames from video files and transcripts using OpenAI Whisper; Returns 1 chunk per minute”

You can send me a message if you still need help :slight_smile: