If you're looking for something more comprehensive, AWS Rekognition has a pretty sophisticated service for segmentation.

Wow, I didn't know that. Is it for video segmentation?

Great idea. Is this really AI though? Are you not using the subtitles generated by Youtube and/or provided by the video creator?

Thanks for asking. Transcript inputs are from YouTube but the actual segmentation part is done by a machine learning model. It analyzes the context of a given transcript to separate the transcript into different segments, and then comes up with topics that represent each segment.

This is great. @deepbuilder I’d love to chat more on this, my email is in my profile if you would be willing to.

I am more than happy to chat, Mintone. But I didn't see the email in your profile. Would you mind sending it to my email ? My email: [email protected]

Doesn't work on videos without subtitles. Bummer.

It doesn't. Because it relies on video's subtitles as input at the moment.

I tried it with a video that has auto generated subtitles but it still didn’t work. Is that expected?

HM, I guess auto generated could be harder to segment for the AI due to sometimes sentances not making sense

It should also work for auto generated transcripts! At least it's supposed to give you some output.

Thanks for letting me know. Do you have a link to the video? Btw, it only works for videos with English subtitles