Repurposing Video Content Into Multiple Mediums With AI Tools

Happy 2026. Hey, how's this for a New Year's resolution? How about you Master AI marketing? The AI Business Society gives you live training with experts who are getting real results. No hype, no fluff, just proven workflows that work. And you've got access to an incredible library of training from many of the incredible guests that you've heard on this show. So secure your AI transformation by visiting socialmediaexaminer.com AI.

Welcome to the AI Explored podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.

Hello, hello, hello. Thank you so much for joining me for the AI Explored podcast brought to you by Social Media Examiner. I'm your host, Michael Stelzner, and this is the podcast for marketers, creators and business owners who want to know how to use AI. Today, I'm going to be joined by Matt Wolf, and Matt is the AI tools guy. We're going to talk about all sorts of cool tools that you can use right now to take any of your existing videos and repurpose them into all sorts of things like the written word, things like images, things like audio, and so much more. I think you're going to find today's show absolutely fascinating. By the way, I'm Elsner on Facebook, Ike Stelzner on X Twitter, and if you're new to this podcast, be sure to hit the follow button on your favorite podcasting app so you don't miss any of our future content because we've got some great stuff coming your way. Let's transition over to this week's interview with Matt Wolf.

Helping you simplify your AI journey. Here is this week's expert guide.

Today, I'm very excited to be joined by Matt Wolf. If you don't know who Matt is, you got to know Matt. He is an AI tools expert who helps marketers and content creators embrace the future. His AI focused YouTube channel has more than 580,000 subscribers. His FutureTools IE site curates the latest in AI tools and news. And his podcast is called the Next Wave. Matt, welcome to the show. How you doing today?

Yeah, thanks for having me. I'm doing great. Excited to have more AI conversations with you.

Yeah. Matt and I are going to explore repurposing video into multiple mediums using AI. Now, Matt, some people might be familiar because you've been to my conference or been on my other show, but there's people here that probably don't know who you are. So why don't we start with a little bit of your backstory how in the world did you get into AI? Start wherever you want to start.

Sure, yeah. So I actually started doing digital marketing back in 2009. I actually quit my day job and time to focus fully on content marketing. At the time I had a blog. In 2010, I actually started a podcast. So I've actually been creating content for my living for 15 years now. So this has been all I've done. But only in the last two years did I really, really shift my focus into AI. So in that, let's say, 13 years that I was doing content marketing, most of that was around teaching people how to set up WordPress blogs and the various plugins that were out there and, you know, the cool tools I talked about Click funnels and leadpages and Opt In Monster and, you know, all of the cool marketing tools. So I've always kind of been this, like, tell people about the cool tools that are out there kind of guy. That's always been my MO, really. But in 2022, end of 2021, I really, really started to get into AI. I came across Jasper, which, you know, used GPT3 under the hood, helped people write their marketing copy using AI. That was one of my first sort of avenues that I ventured down for AI. Got really, really into that. Wanted to learn what was running under the hood of Jasper. Started learning more about GPT3. Eventually, I came across a bunch of AI image generators, learned about stable diffusion, learned about Mid Journey. And whenever I came across any of these cool tools or cool new ways of doing things, I just wanted to make content about it. I've always been like that excited kid that finds a cool new tool and then goes, I need to tell everybody I know about this. This is the coolest thing I've ever seen. And that's really like my content philosophy that I've had for the last 15 years is I find cool, fun tools and then just show them off. So I started doing that with Chat GPT, I started doing that with Mid Journey, I started doing that with stable diffusion. And I was just, you know, slightly ahead of the curve. Before the big AI bubble, before everybody started talking about AI, before it was all over mainstream media, I was kind of there making YouTube videos that were a little bit underground, talking about these AI tools. And when it started to bubble up, my channel was there and I kind of got lucky with the timing and the channel just exploded when everybody started talking about AI. The biggest sort of pop that I saw was when ChatGPT came out. I made a video about Chatgpt maybe a few days after it came out and that video got over a million views. And from that point on my channel hasn't slowed down. So that's pretty much the quick sound bite of how I got to where I am today.

Matt, I know a little bit of your story because you've been on my other show. That million views, did that come in really fast on that video?

That video? It did, yeah. So that was the first time I had a video that went viral almost right away. I had some videos earlier on that know I made them about AI. They didn't really get any views. And then, you know, two months after the video was live, when AI started to get popular and you know, more hyped up, those videos all of a sudden sort of jumped and started, you know, getting into the tens of thousands, hundreds of thousands of views. The Chat GPT video was the first video I ever made where I hit record. I published it, went to bed, woke up the next day and had like almost 100,000 views on that video the next day after publishing it. That was the first time that ever happened to me. And then it just kept on growing ever since and to this day still gets quite a bit of views.

So you are a multi year content creator. You've been doing this for 15 years. You have this massive explosion with this AI video. And for those that don't know the rest of the story, Matt decides to go on and create more content and more content and more content. At what point did you decide you were going to go all in on AI and kind of tell us what happened then and then bring us up to what you're doing now?

Yeah, so the first couple of video I made about AI didn't really go anywhere when I first published them. I made a video about Mid journey. I made a video about stable diffusion and how to train your face into the AI model so you can generate images with your face on them. Those videos I published, they maybe got a few hundred views. When I published them, they kind of started to pick up some steam, got into the thousands of views a few weeks after they got published. I wasn't really planning on going all in on AI. I was just going to continue talking about cool tools and things that interested me. But then when I posted that ChatGPT video, that one went viral. And then people started going through my backlog and started looking at my older videos and then some of my older videos started to get tens of thousands, hundreds of thousands of views. And so it was after maybe seven or eight of my videos got into the tens of thousands of views that I went, okay, I think this is where I need to hone in and focus. And at the time I had a podcast, it was called the Hustle and Flowchart podcast. And I decided, you know what, I need to focus fully on this. So I stepped away from that podcast. My co host is still running that podcast to this day, but I stepped away from it and went, I'm going all in on AI. And when I decided I was going in all in on AI, I decided I'm going to make a new video every single day. So I went about, I don't know, 35 days, almost 40 days straight, where I published a new video around an AI tool every single day for 30 or 40 days. And during that window, my YouTube channel, within about a six week period, grew by 200,000 subscribers in like six weeks. It was just, it was crazy.

So bring us up to the present. Where is Matt now with all this AI stuff?

So I'm still creating as much content as I possibly can around AI. The news cycle, the hype cycle around AI has slowed down a little bit. There's not as much buzz around AI at this exact moment. So because of that, I've sort of slowed down the pace of putting out videos. Now I'm publishing about three videos a week instead of a video every single day. But on average, my videos now will get anywhere from 70,000 to 150,000 views every time I publish a video. So I'm still putting out around three videos a day.

Three videos a day or three videos a week?

Yeah, sorry, three videos a week? My brain's going a little too fast right now.

Yeah, yeah, yeah. And real quick, what do you cover now in the videos? Because I think you do have one of those videos is kind of covering the news. News, is that correct?

Yeah. Every single Friday, I make a video that's a breakdown of the AI news for the week. And those are my most popular videos. So every Friday I almost know for certain it's going to be a hundred plus view video. Because every Friday it's like, here's everything that you might have missed from the past week that happened in the AI world. And I'm covering what happened with OpenAI, Microsoft, Google. But I'm also trying to find smaller stories that nobody's talking about. Things like I did one about like autonomous jets that are using AI and there was like AI dog fights that the US military was doing and things like that that none of the other AI channels or AI news outlets are really talking about. So it's kind of a blend of here's the big stuff that's happening in the AI world, but also here's some of the kind of fun, cool, interesting stuff that might be flying under the radar a little bit. So that's an every Friday video and then the other two videos a week are. They're kind of random. It'll either be like an AI tutorial, you know, maybe breaking down some cool AI tools that I came across that nobody's really talked about yet. Today I'm publishing a video about. It's like a prompting guide, how to get better at prompting using cloud and chat GPT. So it's the other videos outside of the news videos are kind of just whatever is interesting me in the moment.

It's an amazing story, Matt, and it's great to see your journey. And you and I both know that you might hit a million. I'm sure you're going to hit a million subs this year. Just been absolutely phenomenal to see your success. You are a creator through and through. And there are creators listening right now and also marketers listening right now, and they're wondering, I want you to explain to them why is AI going to make their job easier. Like I'm not here for you to explain why they should become AI evangelists, but. But what is it about AI that can help marketers and or creators that are creating content to really do this more effectively?

I really think that AI is great at getting you that first 80% of the way there on almost anything you're doing if you're writing blog posts. I think AI is really, really good at creating that first 80%, that rough draft that you can go in and sort of tweak and make it sound a little bit more like your voice. For video editing, I actually use an AI tool to do a lot of the first part of the editing, cutting out the gaps, cutting out the US and ums and mistakes where I maybe spoke twice and had to repeat myself. So I use AI tools to help with that editing process, but then I still have to go in and do a lot of cleanup myself. With audio, AI can help with things like your show notes for podcasts for YouTube. I use AI to help me write my descriptions and help me get the timestamps for the videos and things like that. AI is just amazing at getting you that first 80%, that rough draft phase. And then you come in and just do the last little bit to make it what you really want it and I think that just really gives you an advantage because it speeds you up over everybody who's not using AI, anybody who's trying to produce video content at the pace that I'm trying to produce video content that's not, not using AI is, you know, probably not sleeping very much and spending a ton of extra time on these, these projects where I let AI do a bunch of the work up front and then I just come in and do the finishing steps to get it online.

I love that. And we're trying to embrace AI across the company and we're going to Definitely be using AI to create the thumbnails for the YouTube channel for this particular video. So folks that are watching this on YouTube, you can bet your bottom dollars. Created by AI with my graphic designer, of course, tweaking it. So let's assume that some of the listeners right now are just about to get started with AI tools. And they love what you just said, they're buying into your vision. But maybe there's some things they need to be aware of before they get started. Like what do they need to be thinking about before they actually just jump into this AI world that we're talking about?

Yeah. So, you know, I mentioned that it gets you that first 80%. It's good at that rough draft, but it's not always perfect. Right. I think that's something that a lot of people need to keep in mind. If you're going to have AI write a blog post for you with, with, let's say, factual details, make sure you double check the facts on it. Because AI has been known to what we call hallucinate, where it just, when it doesn't actually know the facts, the facts aren't trained into its data set. It just makes up the facts. Right. So you're going to want to double check anything where the AI is trying to be factual. Right. That's definitely important when it comes to some of these AI tools that we'll get into a little bit more into this episode where it sort of clips out chunks of your video and maybe turns them into shorts for you. You're going to want to double check the shorts, make sure it gets like the right portion of the video. Sometimes it might, you know, cut you off mid word, mid sentence, and the video won't come out perfect. There's a lot of little things where AI is still very imperfect. And I feel like right now AI still needs a babysitter. It still needs that, that person to come in, follow up and just sort of double check its work and that goes for, you know, AI audio, AI video, AI text, all of it. You're. You're still going to want to double check it.

Yeah. And, you know, it's funny because I've been art directing our designer in house on some of these thumbnails, and I'm using ChatGPT with Dall E3. And it's just like, man, it takes me forever to get Dall E3 to get it right. You know what I mean? And no matter what I tell it, like, I'll say create a thumbnail with no text. It'll have text. And then it will tell me it doesn't have text. And I said, yes, it does. And it says, oh, I'm sorry. And then it'll create another one with text, you know, so just be prepared that you're going to need to do a lot of, like, refinement process. But what I'm excited about is that the creative possibilities are absolutely endless. What we're here, really today to talk about is video. So let's assume we have a video, and maybe it's a recording that we took on our iPhone. Maybe it's a podcast episode that we're filming in video, like we're doing right here, right now in Streamyard. But let's say that we have a video and we want to take that video and we want to slice and dice it into all sorts of different forms of content, but we want AI to help us with this. Where do we start? What do we need to be aware of? What do we start with?

Yeah, AI is the perfect repurposing technology. So the first thing I would probably do is I would take that video and I would get the transcripts from the video. So you can either pull the Transcripts directly from YouTube. If you upload a video onto YouTube, it will automatically transcribe it. I believe Vimeo does as well. I think Mikey told me that. That they do as well.

Yep. You'd probably have to have a premium account, just so you know, on Vimeo. But yes. So, real quick, does YouTube make it easy for you to pull the transcript out of it?

They do, yeah. So if you look at a video and then you open up the description of the video, there's a little transcript button in the description. One thing to keep in mind is when you upload a video to YouTube, the transcription is not usually available immediately. So it could be several hours after the video is uploaded before the transcription shows up. But if you open the description on any video on YouTube, you'll see a little Transcription link towards the bottom of the description. So if you click that button, your transcript will open up on the right sidebar of YouTube. It'll allow you to either show the transcript with timestamps or without timestamps. For our use cases, I always use it without timestamps because I'm just trying to get the content. I don't really need the. Where we mention each piece of content and you can copy that transcript. Now, I don't think that YouTube is always the best at transcribing. I actually tend to do my transcriptions with a tool called descript. Descript actually uses AI to do the transcription, and it tends to work a little bit better than YouTube's transcription for some reason. So a lot of times I will pull it into descript. Descript is also a really, really good editing tool, but we can probably get into that more in a minute. But you grab the transcript, take that entire transcript, and then you can copy and paste it into a tool like Claude or ChatGPT and literally tell whatever chatbot you're using to rewrite this in the form of a blog post and then give it some additional context. Tell it. Don't forget to include bullet points, don't forget to include headings and subheadings. And you can even tell it. Make sure to generate ideas for images throughout this blog post. So that's something else that you can kind of add to the prompt is when I publish this, I want images in the post. Suggest where the images should go and what those images should be, and then Claude or ChatGPT, whichever your tool of choice is, they're both fairly on par with each other these days. Will write up a blog post for you with formatting, with ideas for images, and you can essentially copy and paste that blog post. Double check it like we mentioned, make sure that the language and the wording sounds like you, but you have a 80% of the way there rough draft for a really solid blog post based on the content that you already produced.

And just a little context for people that have Vimeo accounts that are, you know, I don't know what level we have. We have a premium account, I think you don't have to remove the timestamps when you're using Claude because it's smart enough to figure it all out. So we just download whatever that file is called, closed captioning file, and then we upload it in there and then we can ask it to do the writing. Now, what's really intriguing is I think it's probably fair to say Matt, that we probably don't really need to proof because the AI is smart enough when. When there's typos and stuff like that. Or do you proof these transcripts before you upload them into the AI? What's your thoughts on this?

Yeah, I don't proof the transcript itself. I double check the output of the AI though the transcript itself, it will miss some stuff, right? Like if you're talking about a company's name or a person's name or something like that, some of that stuff will probably be misspelled or it'll misunderstand and think you said something else. But AI is really good at finding the context around the entire thing. So it's going to get the messaging right, but the output that you get, you're going to want to double check it for misspellings of proper nouns, essentially. But other than that, you don't have to check it for spelling or grammar or things like that. The chatbot is going to be smart enough to figure out all of that kind of stuff for you.

Okay, so so far what we've talked about is we've talked about taking a transcript out of a video, and I would imagine if it's a YouTube video, you could unlist the video and you're still going to get the transcript probably within a day. Right. It doesn't have to be published for there to be a transcript, does it? Because I know a lot of YouTube creators will upload it unlisted and then they'll schedule it for later. Have you found that?

Yeah, I believe it still does. The transcript. I'm not 100 sure on that because I usually hit publish the same day I upload, but I'm pretty sure it will still do all the transcripts behind the scenes for you.

Let's stick on the discuss of creating a blog post, right? So we're using a tool like do you use cloud or you do you use chat GPT, which is your preferred for creating long form written content?

I kind of use them both. I sort of jump back and forth. Lately I've used clot a little bit more. I think Claude is a little bit better at sort of longer context, longer information. So if you're uploading something that's say 45 minutes long, I think Claude's going to do a better job at sort of understanding that whole thing than ChatGPT will when it comes to large language models that these chatbots use. Underneath, they have what's called a context window, which is how much text between the input you give it and then the output is going to give you back. Can you do. And Claude has 228,000 context windows, something like that, which is roughly 180, 190,000 words between the amount you can input and output. When you're using ChatGPT, it's a much smaller context window. I believe it's only 32,000 context window, which is. It's really annoying that you have to think of how many words but also a context window. And they're not the same numbers, but it's basically 75% of the context window.

Let's talk about filling the gap between AI potential and AI results. I'm sure you've tried ChatGPT. Are you seeing real ROI? Most marketers are barely scratching the surface of what's possible with AI. And it's because they're missing the training and the frameworks that will allow them to be very successful in their careers and in their future. AI Business World allows you to get that training. When you attend, you get two full days and 20 expert practitioners, many of the people that you've heard on this very show, that will teach you exactly what you need to know to build the workflows and and to succeed as an AI enhanced marketer. It takes place April 29th and 30th in Anaheim, right alongside Social Media Marketing World. And here's the good news. You can save $300 when you register by January 16th. Head to AI BusinessWorld Live right now. This discount ends soon. So this is intriguing. Okay, so we're using it to create a possible article that we may or may not edit. Let's assume we edit it and we paste it back in and then we ask it to recommend where to put images. Right. So when we know where to put the images. Now let's talk about creating the images. So how do we create images based off of what like Claude is recommending to us?

You can actually in the prompt when you're telling it to write you the blog post and when you're and telling it to include where I should include images and what the images should be. The more detail you give to the prompt, the better. So you can actually give details in the prompt of the. Write the prompt for me that I should use to get this image. So suggest where the images should go throughout this blog post and then write the prompt for the image that should go in that spot. And then when you get the prompt, whatever image generator you choose is probably going to be fine. I tend to jump back and forth between Dall E3 mid journey and then stable diffusion. Stable diffusion is a little bit more complicated. Dall? E3 you can use directly inside of Chat GPT or you can use it for free over on Bing's image creator. And then Midjourney is I think the cheapest plans, like 10 bucks a month. And as of this recording, you still got to go into Discord to generate images. But they are actually moving it out of discord, like really, really soon. But yeah, just within the prompt of telling it to create this blog post. Also ask it to generate the prompts for the images within the blog post and it will give you sort of detailed prompts that you can copy and paste.

Okay, so let's talk about this for a minute. Specifically, I want to talk about the pros and cons of Dall? E 3, which is probably the most people are familiar with because it's integrated into Chat GPT versus Mid Journey versus Stable diffusion. So talk to us a little bit about the pros and cons. Like which one do we want to use for what kind of images?

Dall? E3 is probably the best for very complicated images where you want a lot of elements in the image. So some of the examples I've done in past demonstrations were like, give me a three headed dragon wearing a fedora, watching tv, eating nachos on a hot sunny day, with a window in the background, right? You put a lot of things into that prompt. And Dall? E3 is really good at hitting every single one of those elements. And almost all of my tests, you can put like six or seven elements in there. And Dall E3 will get all of them Mid Journey, on the other hand, is not quite as good at getting all of those elements in there. But if you're looking for something that's fairly realistic, Midjourney is the best at that right now. So if you want an image that looks like a realistic human on a city sidewalk wearing glasses and a three piece suit or whatever, that is going to generate a fairly realistic looking image. So I love going to Midjourney when I need the realism factor, and I love going to Dall E3 when I need it to be all of these things in a single image. I want to make sure all of these elements make it into the image. So that's where I kind of use those two. They both also kind of have their own sort of style to them. I've gotten to a point now where I can look at an image made by Dall? E and go, okay, I can tell that was Dall? E and an image generated by midjourney and go, okay, that was midjourney. Midjourney has deeper contrast. They're more colors, they're more vibrant images. Dall? E3 tends to use a little more muted tones. They never really go for realism in Dall? E. It's really, really hard to generate something realistic in Dall? E. And then when we're talking about stable diffusion, stable diffusion is a lot more complicated. It's a lot more in the weeds to use. A lot of times to use it, you've got to install software locally on your computer. You got to have a decent GPU to run it. There are some websites out there that let you generate stable diffusion through a SaaS platform like Leonardo. AI is one of them that uses stable diffusion. Underneath the hood. A stable diffusion is when you want to get super dialed in. You have this picture in your mind of exactly what you want this image to look like. Stable diffusion, you'll probably be able to get it because stable diffusion has all sorts of add ons and plugins. It's almost like a WordPress of AI art generators. You can add a whole bunch of things onto it. So they have things called like control nets where you give it a reference image and it will try to make another image that looks like the reference image. Or you can give it a pose. Like I want it to be like a pose of somebody resting their chin on their hand, right? You give it an image and say, follow this pose and it will generate a brand new image with the context and details you give it. But following that exact pose that you put in there. Then they have these things called loras. And loras are like sort of add ons to your image that give like special effects to it. There's like lightning loras that make it look like lightning striking the image. And it's very open source. There's a lot of people developing it. There's a lot of. They're not really called plugins, but there's a lot of plugin type things for stable diffusion.

It's got a learning curve though. It sounds like it's got a.

It's got a big learning curve though. Yeah, exactly.

So just so we're clear, Dall? E3 is only available with the paid ChatGPT account. But did you also mention that you can use bing.com create to also get it? Is that correct?

Yes. So if you use ChatGPT paid, you get it directly within that chat window. But over at bing.com/create I think it lets you generate up to 100 images a day over at bing.com create and then even after you generate 100 images a day, it'll still let you continue to generate. It just slows down the generation. So you just can't generate at the same pace. But if you're generating 100 images a day, I mean, probably recommend going to one of the people.

Is it exactly the same quality output? Okay, yeah.

Same exact underlying model. Yep.

But Mid Journey in particular, did you say that's the one where you have to go into a discord, Is that correct?

Yeah, Mid Journey is in discord. That one's also paid, only they don't have any free way to use it. So I think the lowest plan is $10. But it is in discord. To use it, you do have to sort of learn the commands. There's a few different commands to get the aspect ratio and to get different styles and things like that. Dall E3 and Midjourney are both fairly accessible to anyone. They're both. The learning curve is fairly low on those.

What's cool, I use Dall E3 pretty extensively for the work that I do. You can upload a reference image, I don't know if you're aware of that, into ChatGPT, and then you can ask it to model it, which is cool. So I found one that I really liked and I asked it to model it, and it does an okay job. Okay, so any particular tips on how to do image prompting? Because obviously it is a mad science, right?

Yeah, it really depends on the tool too. So Dall E3 and Midjourney have made it simpler and simpler and simpler. So like with Midjourney, you can go and give it a prompt like make me an image of a cat, and you'll probably get a pretty dang good looking image of a cat. Dall E3 is kind of the same way. What Dall E3 does, though, is underneath the hood, it actually reworks your prompt for you. So if you went and told Dall E3, hey, generate an image of a cat for me, Dall E3 is actually going to send to the generator a much more detailed prompt. It'll say, generate a cat for me that's orange with green eyes, you know, sitting on a chair or whatever. So if you just give it a short prompt, it actually sort of shores it up for you. Right. It actually sort of improves the prompt to get you a better output. When you're using Dall E Stable diffusion, there's a whole science to prompting on that one. Right. You'll See people prompting with additional information like artists names and adding HD and 4K and Blender and, you know, there's all sorts of like little codes and words that people have figured out that if I add this to the end of my stable diffusion prompt, it generates better images. Again, stable diffusion, pretty big learning curve. Mid Journey and Dall? E, honestly, you can pretty much put simple prompts in there and get pretty good images. If you are going to use Mid Journey and you want like a really realistic image, you can put Dash Dash style raw at the end of it. And if you tell it to generate a raw image that tends to give it a more realistic look. You can also pick your aspect ratio. I like to create images in 16:9 because that's the perfect sort of YouTube thumbnail. It's also the perfect featured image for a blog. So I always like to make sure that it's generating in the right aspect ratio. But over time, these image generators have gotten to a point where they wanted to make it simpler and simpler and simpler so you don't really have to think about prompt engineering for your images. Stable diffusion, there's still a little bit of that. But Dall? E and Mid Journey, honestly, like just generate a bunch of images until you get the one you like. I mean, that's what most people do. AI is sort of like a roll of the dice. You roll the dice, you get a few images. That's not what I was looking for. Roll it again. Maybe the second time around, you get the image you're looking for. That's really the game right now.

And it's definitely going to come out square. Generally speaking, on chatgpt/dall? E3, unless you tell it to go 16 by 9. I heard, and I know that we might be dating ourself on this episode, but I heard that Dall? E 3 with editing capabilities, but I'm not yet found how you can do it. So how in the world do you edit an image in dall e3 in chatgpt? Is that complicated?

It's not. It's actually out now. So when you go into Chat GPT and you ask it to generate an image for you, you can actually click on that image. And when the image opens to like a fuller screen, what you'll notice is as you move your mouse around the image, it will highlight different areas of the image. So it's automatically looking at the image and trying to segment out the different pieces of the image. So as you move your mouse around, let's say there's a person that's sort of the foreground of that image. If you hover your mouse over the person, it will highlight that person. If I click on the person, it will basically let me generate a prompt to edit just that person. Or I can say remove this person, and it will actually remove the person.

That is huge right there. That is a big breakthrough because so many people have had the experience I've had where the image looks beautiful, but they added text in it and it's like, how do I get rid of that text? Right. I got to ask it to generate it again, or I have to use Photoshop generative fill, which is not my ideal scenario. That is a really big deal. I don't understand why they don't make that more clear. So you just click in on the image and you hover over it, and somehow it's smart. And it will select that section of the image. It comes up with a prompt, and it allows you to re prompt it. And can you keep re prompting it until you get what you like?

You can. Yep, exactly. So it's. It's real easy. They call that inpainting. It uses a computer vision model to look at the image and try to figure out all the elements of the image. And then as you move it around, it'll highlight elements. So if there's like a bird in the sky and you don't want the bird, move your mouse over the bird, it'll highlight the bird, and then you can just prompt it to get rid of the bird. And it'll get rid of the bird.

Yeah. And see, this is the struggle. Right? Like, maybe I don't like the color palette. Right. And in the past, I would say do the same thing in a different color, and it would come out with a completely different design. Would this now allow me to essentially just change the color, or is it not quite that sophisticated yet?

I don't think it's quite there to change the color. And Dall? E has a very sort of muted color palette. Overall, I feel like all of the images kind of have the same color palette. If you're trying to get more like, vibrant, contrasty images, to this day, I still kind of pull them into Photoshop and brighten them up myself. That's one of the reasons I actually go to Mid Journey fairly often. Still is. I just like Midjourney's color palettes a lot better than Dall E. So if I can't get the colors looking just right in Mid Journey, I'll go and do it. Or in Dall? E, I'll go and do it in mid journey to get that color palette.

Okay. So up to this point, we've talked about taking a video, grabbing the text out of the video, the transcript of the closed captioning, creating an article, and we didn't even touch on the fact that you could obviously create social text posts out of it as well. And that's kind of a given. You can. You could ask ChatGPT or Claude to create you all sorts of different length summaries from it. And then we spent a lot of time talking about images because I think that's a little bit more of the mad science kind of stuff. Like, we can edit text, but we don't all know how to edit images. Right. Which is a little more complicated. So you and I, when we were prepping for this, said, hey, you can actually create original audio also out of the text that comes out of it. Talk to me a little bit about that.

Yeah, so there's a tool out there called 11 Labs that's really cool. Basically, you can train your own voice in it. And I think it only requires somewhere between 30 seconds and two minutes of an audio clip. So you can just, you know, open up Audacity and record yourself talking for two minutes, upload that audio into 11 labs, and 11 labs learns your voice. And now anything that you type into 11 labs, you can have it output an audio of you speaking it in your voice. Now, you've already got the audio from your original video anyway, right? But we all know that there's some issues with just scraping the audio from a video and then saying, here's a podcast. Right? Because maybe you're showing stuff on the screen, maybe you're saying, if you click over here, you get this result, and if I click here, I get this result. So it's not always ideal to just rip the audio and say, all right, here's an audio version of the same piece of content. But since we took these transcripts and we had our ChatGPT or Claude turn it into a blog sort of written content piece, well, I can take that blog now, copy and paste the whole thing into 11 labs, and then have it generate the whole thing back to me in my own voice. And then it will Give me an MP3 file. The MP3 file will sound just like me speaking out the blog post. And now I can embed that audio version at the top of my blog. I can go and use that as a podcast episode. And every single piece of content that I make in video form, I can then convert to a blog and then convert to an audio podcast that actually is something that would be acceptable to listen to in audio without all of the sort of visual references that you might be sharing on screen during your video version.

Well, and if we think about what this could unlock, let's just say you are recording audio only and you're just kind of rambling a little bit and then you're letting AI clean it up. Or let's say you've written something that you love, but you never created a video out of it. You could at least use this to create an audio out of it that could be used in a podcast. So there's lots of applications here. Now, on this 11 Labs thing, does it only work with a two minute sample or can you give it a bigger sample? And the reason I'm asking is because, as you know, there's lots of vocal variations and inflections that we use when we're animated and when we're not animated. And if it's only 60 seconds to two minutes or whatever, we gotta get that audio sample to be representing us truly. Right. So can you give it a longer file or is that not how it works?

No, you can, you can. And that's exactly how it works. The longer the file you give it, the better it's going to be. You can even give it multiple files. So if you have 10 files that are two minutes each, you could feed it all of those and it will learn on all the different files. But yeah, absolutely, the more audio you give it, the better it's going to be. And then when you actually upload your text, when you, when you copy and paste the text into 11 labs, there's actually a little slider that allows you to adjust sort of. I don't remember the exact word that they call it, but it. The inflections in your voice, if you drag it one way, you'll tend to be more monotone. You drag it the other way and it'll tend to make your voice a little more animated and a little more random. So there is actually a little slider that you can play with to get a little bit more of those inflections and things like that into your voice.

Is it smart enough to know to provide emphasis in certain areas and not in other areas, or do you have to kind of work sentence by sentence to do that?

To a degree, it's usually pretty good. I'm typically pretty impressed. Sometimes it'll put the emphasis on the wrong words here and there's. But for the most part it's pretty good. It's pretty smart. And it's getting smarter over time. I mean, these algorithms behind the scenes are just getting better and better and better. Eleven Labs also has translation now as well. So you can actually tell it to take that same piece of content that we just put in there and make a Spanish version, and it'll actually sound like your voice speaking in Spanish. It's not like a new voice. It's my voice speaking in Spanish that came out of it. So huge implications in the translation side as well. But, yeah, I mean, overall, it does a pretty good job of finding where those inflections should be and the context of what you're saying. It does a pretty decent job at that. Not 100%. Again, all of these AI tools are like 80% there, but you still want to double check it and clean it up a bit, probably.

I'm very fascinated by this. What other kind of audio things can you do? What about sound effects and background music tracks and all that other kind of stuff? Let's talk about that a little bit.

Yeah, so 11 labs, amazing, amazing tool because it does sound effects too. You can go into 11 labs and there's actually a separate tab for generating sound effects. And you can say, I need a. An ambulance siren and it'll give you that. Or I need a dog barking in the background, or, you know, whatever you can think of. I've. I've made, like, explosion sounds in there to put in the backgrounds of some of my videos, or like the sound of a person falling over. You know, whatever you can think of, you can actually have it generate those sound effects.

Are they real sounds or are they generated sounds?

They're generated. They're all AI generated sounds. So it's trained on a large data set of audio samples, and then it does a good job at going, okay, well, out of all of the sounds that sound like a siren that we have trained into our data set, this is what the average sound of a siren sounds like. So here you go. It essentially works is the same way as, like, a large language model like ChatGPT does, except it's trained on audio instead of text.

What would you do with these special sound effects? Like put them in your videos or something like that?

Yeah, for me specifically, I would use them for videos or, you know, maybe some podcasts here and there. You can do, like, little transition sounds in between sections of a podcast. For me, though, mostly video. Right. If I pop a meme up onto the screen during one of my videos, I like to have a sound effect to complement that meme. So a lot of times I'm doing it to just to add some little extra dynamics to the video.

Let's talk about music. I know this is kind of an exciting frontier. What do we need to know about music?

Yeah. So there's two really, really cool tools to check out when it comes to generating music with AI. There's Suno S U N O and then there's Udo UD IO. Both of these tools are really, really good at generating whatever kind of music you can think of. Right. You can go in there and say, give me an early 2000s pumpkin anthem about why I should attend Social Media Marketing World. And I actually did that. Part of my presentation at Social Media Marketing World was me creating an AI generated jingle for why people should attend Social Media Marketing World. And I played that on stage.

Is it giving you the lyrics and the music?

You can have it do it either way. So you can either tell it to generate the lyrics for me with AI. And here's the concept, right? I could say I want it to be a jingle for why you should attend Social Media Marketing World, and it will write the lyrics for you using AI. Or you can open up a text box, type in your own lyrics, and then it will put music to the lyrics you give it. So you have both options. There's also a third option of no lyrics. You turn off the lyrics altogether and then it just gives you, you know, a background music track with no lyrics. And that's what I tend to use when I'm. I'm generating videos and I put music in the background of videos. I'll use a tool like Suno or Udio. Tell it no vocals and then just say, this is an action scene. Give me something that feels like a rising action scene that's cinematic and energetic and it will go and make music that sounds like background music to a movie that's cinematic and energetic and sort of crescendos up to ramp up the energy. So I. I've used it for both. As far as, like the actual lyrics. Right now, the I feel like the main use case is memes. Right. You can make really, really fun, funny songs, but there's not a whole lot of use cases for it yet. It's just really fun. But these tools are also great at just turning off the vocals and letting it generate a good song to be this sort of sound bed for a video or podcast.

What's the differences in your opinion between these two tools, Suno and Udo?

They're both very, very similar. But Suno will probably sound a little more AI generated. When you listen to a Suno song, you can kind of tell it's AI generated. Some of the lyrics will sort of blur together, and there's a little uncanniness to the vocals where Yu Gio sounds ultra realistic. You'd probably listen to a song that came from Yu Gio, and if I played that for somebody, they would just assume that that was a real singer singing that song and it wasn't AI. It is really, really so good that you can barely tell it's AI Suno. You can kind of tell. However, I actually think Pseudo generates more poppy, catchy beats than Yu G. Oh, does Yu do the songs it generates? They're not songs that I tend to, you know, bob my head to or get too excited about, but it sounds really realistic. Suno. I really, really, really like the music that it makes, but also I can. You can kind of hear the AI ness of it.

Okay. We've covered music, sound effects, turning text into audio. We've talked about turning video into text. This is a crazy question that I didn't have planned, but I'm just wondering, like, sometimes we want to have cool images in articles that are maybe a little more complicated than what these tools can create. Have you seen any interesting images that can create, like, bar charts and tables and, I don't know, just can take screenshots and make them look sexier? I mean, I'm. I'm just thinking about people that create articles tend to want to put support images in there. And are there any that have come across your horizon that can do this kind of stuff yet?

There are some tools that you can upload a bunch of information or a blog post and it'll make, like, infographics for you? Oh, unfortunately, I don't know the name of the tools off the top of my head, but there are some. Actually, if I go to the Future Tools website and I type infographics, I might be able to find you.

Yeah. While you're doing that, one tool that I did want to mention that Matt and I had on our roster here was Opus Clip, which is pretty cool. Opus Clip will allow you to take your video and the AI will essentially slice it up and it will make recommended videos that you could produce as TikToks shorts and reels. It's pretty interesting. They sponsored social media marketing world, full disclosure. But what's cool is their AI tries to find the best clips and they recommend it and they rank it for you. That's something else. We didn't talk about. Were you able to find that tool while I was.

Yeah. So the tool that I'm thinking of is called Infography. I N F O G R a p h y.in okay, so infography.in and that tool, basically you upload a blog post or you link it up to a blog post and then it tries to create like infographics from the content of the blog post.

Well, since you mentioned this, if people want to find a tool to do something, there's a good chance they're going to find it on your website. So why don't you tell them more about where to find that and if they want to connect with you on the socials, where would they find you? And if there's any other contact points you want to mention, please do mention that.

Sure. Yeah. So my website is called Future Tools. I. Oh. And it's a site where I curate all of the cool AI tools that I come across. Right now it's getting close to 3,000 tools on there. I've been, been adding to this database for two and a half years now. So it's, it's building up all of the cool tools I come across. And then it's also got an AI news section. So if you watch my Friday AI news videos, a lot of times I'll leave out some news that I feel is kind of too insignificant to talk about in the video. Pretty much all the interesting AI news that I come across I also share on future tools. So both of those are on future tools. You've got the tools, you've got the news. It's, it's designed to be a one stop shop to kind of stay looped in on what's going on in the AI space. I've also got my YouTube channel, which you can just find by searching out Matt Wolf on YouTube. And then I'm on X fairly frequently. I'm Mr. E Flow on X. M R E F L O W on X. So those are really the three hangouts that I spend the most time future tools, YouTube and Twitter, slash X.

And for folks listening, Wolf has an E on the end of it. W O L S. Matt, thank you so much for coming on and sharing your insights with us.

Absolutely. This is fun and I'm happy to do it anytime.

Hey, I know there's a lot of stuff we talked about and we've got all the notes for you over@socialmediaexaminer.com a6 and be sure to follow this show on your favorite podcast app if you're not already. And would you do me a favor? Would you let your friends know about this show? If you want, you can tag me. I'm at Stelzner on Facebook and Ike Stelzner on X Twitter. And if you love this show, we've got a bunch of other shows. We've got the Social Media Marketing podcast hosted by me. We've got the Marketing Agency show and the Social Media Marketing talk show. This brings us to the end of another episode of the AI Explored Podcast. I'm your host, Michael Stelzner. I'll be back with you next week. I hope you make the best out of your Journey day, and may AI help you become more successful.

The AI Explored podcast is a production of social media examiner.

AI Business World, early bird pricing ends on January 16th. You can save $300 on your tickets. Register at AI Business World Live today.