Looking for video transcriptions

[Dablooe]

If anyone has free time please think about transcribing an LTT video, I want to put it all together and make an AI. I think it would be funny. Please if you do use the system where you put their name down like [Linus] : "XYZ". Thanks!

[LogicalDrm]

It currently reads like you want to rip off LMG content to create AI for unknown purpose. LMG is fine with using their content for personal uses and educational purposes. But your end goal is not clear here.

 

Additionally, asking fellow members to do work for you is not allowed. You can ask help on tools and methods, but asking others to do something for you while you don't do anything yourself is not cool.

 

Lastly, most recent LMG videos are already transcribed. Either by LMG with submitted script for the video or by Youtube. Example from latest upload:

 

Restored after review. 

Edited June 23 by LogicalDrm

[Dablooe]

I know, to be clear it is for personal use, and I was wondering if anyone could help add names since their transcriptions dont have that.

 

Like

Linus:XYZ

instead of just XYZ

 

To confirm, for personal use in Azure Agent Playground, not to train ai. I can see how my first message can be misleading.

[LogicalDrm]

On 6/21/2025 at 7:00 PM, Dablooe said:

I know, to be clear it is for personal use, and I was wondering if anyone could help add names since their transcriptions dont have that.

 

Like

Linus:XYZ

instead of just XYZ

 

To confirm, for personal use in Azure Agent Playground, not to train ai. I can see how my first message can be misleading.

Fair enough. 

[05032-Mendicant-Bias]

On 6/21/2025 at 5:17 AM, Dablooe said:

If anyone has free time please think about transcribing an LTT video, I want to put it all together and make an AI. I think it would be funny. Please if you do use the system where you put their name down like [Linus] : "XYZ". Thanks!

The easiest way is to get the youtube transcript that already is in the youtube videos, but that is not speaker qualified. I do it all the time to recap videos that get sent to me.

 

The second way I would look into is to extract just the MP3 track, and process through a STT tool.. I use Whisper to transcribe my D&D session, but I still haven't found a tool that can identify speakers. I think they exist but are hard to run.

 

I would avoid scraping and processing the videos. It's unnecessary... I am working on a ComfyUI workflow to use Florence2 to describe the frames and make a richer transcript to describe the scene and have a richer log of our campaign, but it's a bigger scale project. It would be needed if you want the trace to build one those slop youtube channels (don't, it's a waste).

 

But if it's just for fun, just scrape a bunch of transcripts and finetune a llama 3.1

 

Much easier would be to craft a good system prompt and make a persona that way. "You are Sinus Lebastian, a tech youtuber..." that would be quicker, easer, and possibly better anyway.

 

Also, why don't get transcript from videos that just have Linus? You bypass the whole speaker identification issue. Linus is solo host really often.

 

And please, don't misuse your powers. Impersonation is illegal in most places to various degrees. Use GenANI for good, to create more, better content and solve problems. Not to create more problems and issues.

 

EDIT:

https://huggingface.co/pyannote/speaker-diarization-3.1

https://github.com/AIFSH/ComfyUI-WhisperX

WhisperX seems to add speaker recognition, give it a try

 

[KidKid]

On 6/21/2025 at 12:00 PM, Dablooe said:

I know, to be clear it is for personal use, and I was wondering if anyone could help add names since their transcriptions dont have that.

 

Like

Linus:XYZ

instead of just XYZ

 

To confirm, for personal use in Azure Agent Playground, not to train ai. I can see how my first message can be misleading.

I think you might have to do that, unfortunately.