Speaker 1: Okay, this is gonna be a relatively quick kind of in-between video while I'm working on some other stuff but this is subtitle edit if you remember from my subtitle video from Jeez, I don't know how many years ago, but this was a software that I use and it's awesome because it's powerful and it's open-source So I'm gonna drag a video on here real quick and they actually added a few things that are cool One of them was a feature request that I put in and another was and another is actually really powerful So right here I have a video right and here's the waveform and all that this was the last video I uploaded on the channel and instead of what I normally do which is Put it in YouTube and have it auto time it and then put my script in YouTube and have it auto time and then I have To go through and manually watch everything and make sure it's perfect They actually added whisper in here which whispers basically a machine learning model or whatever made by open AI that can transcribe audio into text so this is actually way better than especially YouTube auto captions which don't capitalize letters or anything so if I go to tools and then okay I lied if I go to video and then audio to text whisper click on here and then it's gonna say you want to download it yep it's really fast and so yeah right here generate text from whisper speech recognition so then I'm gonna choose a model which I guess I have to click on here and then there's tiny and all this I'm gonna go actually large I'm just gonna do v1 I don't really know the difference between the versions but you know what why not let's go v2 but large will have a lot more words that it can recognize so especially for my channel where things are a little more niche maybe it won't have it in there so I'm gonna go download here this will take a while okay and once that finishes downloading then you can just click generate and and it's gonna have auto-adjust timings and this merges lines, fixes casing, obviously as it says there so actually let's go to settings now there we go, add periods, probably want that and fix casing, might as well short duration means if it generates a .5 second subtitle it'll merge it with other ones and generate this will also take a while But what's cool is this is running locally on my computer, I'm not relying on sending the entire video to a server, and then waiting on that to transcribe it, and then bringing it back down, and then I'm not paying someone to, like on Rev, which is a whole other thing. Why I'm excited about this is because captioning the video is the part I don't like, because I've already listened to my voice 8,000 times while I'm editing it, and I already know the script because I wrote it myself. I don't use any tools really to help me write scripts or edit, but this is the part that is the most boring and can easily be automated. Okay, so it's done now. I actually had to switch to a virtual machine on my MacBook because for some reason the latest version 4.0.6 wasn't working, but 4.0.5 on my MacBook does, so I might file an issue on that, but if you look at it, it did the entire transcript here. it even added line breaks and stuff which is awesome because it makes my life so much easier. I do notice it didn't extend out the lines all the way but if I just do ctrl a right click extend the line after then after a second it will extend all of them out to the next line. So now look it even added capitalization and punctuation which is so much better than YouTube's out of captions which doesn't do that at all it just is one long continuous string of text. So yes, this did take 30 minutes to do, but compared to me sitting down for 45 minutes and captioning it, or even 30 minutes, I mean it's automated so I can just do this and then go eat lunch or something. It'll really improve the workflow. And about the roll-up captions, the GitHub issue was marked as completed, but I cannot find anything in either version on how to do roll up captions. So if any of you can figure out how that'd be awesome please leave a comment. Roll up captions are the one where when a new caption line comes up the last one gets pushed up and then it slowly comes on the screen. I think it's kind of a cool look and that's how broadcast tv does it so I want to figure that out but I guess I'll just stick with the current method of finding and replacing the file. Anyway so yeah I was looking through the changelog and stuff and man they've been adding a lot of this automatic speech-to-text stuff. Seriously it's such a powerful tool and so I just wanted to make a short video on how awesome and easy it is to do this and they just keep adding stuff so I'm definitely gonna start using this and again I'm still gonna make sure the transcript is accurate I'm gonna watch it through but it looks it looks pretty accurate and they even capitalize stuff like eBay correctly and new to capitalized harmony which is interesting pretty impressive USB is capitalized I know it's a shorter video I just wanted to cover this so thank you to everyone on patreon who supports the channel hope you learned something and thanks for watching