Okay, perfect.

Introduction

Hi everyone. Today I'm going to do very practical talk. So I'm going to show how to create a podcast with your digital avatar. So I would like to start with this short video.

Hello everyone. Yes, I may look real, but I'm actually an AI avatar here to kick things off.

Creating a Podcast with Your Digital Avatar

In this session, We will focus on creating a podcast featuring your digital avatar using the tools HeyGen and Notebook LM.

So, let's start. Okay, so let's start.

My name is Adas Frank. I'm MindStone partner, CEO, and founder of NextGen AI. I'm an AI strategic consultant and lecturer.

Preparing the Podcast Material

So as you saw in the video, what we're gonna do, we're gonna make podcast with AI and also the visual for the podcast. But to do a podcast, you need to have the material. What do you want the podcast will be about?

So in our, in this case, You need to have a URL code or a PDF that contain the material for your podcast. And after that, we're going to upload the material to Notebook LM.

Then we're going to generate the podcast. Then we're going to upload the audio to HiGen to create the visual for the podcast.

Generating the Audio

So I'm going to show you now, step by step, how to generate the audio for the podcast.

Now, Notebook LM, it's a tool that basically what it does, manage your notes. You can ask questions about your PDF or links to YouTube videos or links link to website or even research paper that you want to ask questions and so on.

So you go to Notebook.LM and then you create a new note. And then, as I say, you need to upload your source. Now it can be from Google Drive. It can be a link. or paste text from Cloud or ChatGPT.

You have up to 50 documents you can upload. So that's a lot.

After we upload the source, then we're going to start to generate.

But in this case, I'm going to use the blog of Google about Notebook LM. That will be my podcast material.

So I copy the link. I paste it, and it's going to generate for me the podcast. So as you see here, that's how it's going to look like.

You're going to have the summary. You can ask question, have conversation. But what I want to focus about is this, deep dive conversation.

So you have two hosts. It's available only in English. This is 100% for free. Any one of you can do it today even.

So now this is pretty new. You have customized, you can customize your broadcast and you can give a specific prompt, you can say focus on specific source or focus on specific topic and then you're going to generate. And it takes about a minute or two.

and then you have an option to speed it down and then we have the audio. Now, have in mind that you can do it a couple of times, it doesn't matter, but every time you do it, it can come a little bit different and also the time. If this time it was almost 10 minutes, when you do it in the next time, it's going to be four minutes or 16 minutes. So each time it's going to be a little bit different.

Creating the Visuals

So after we have the audio, we're going to take the audio we download and we're going to upload it to HiGen. So HiGen, it's a platform that's allowed us to create avatar videos. We're going to focus today on the avatar video.

I know you can do a translation and all that. But we're going to focus for the podcast on this avatar video. So you're just going to click on Create New Video.

And then you choose the layout. And hold on. And as you see here in the beginning, you have the Create Avatar.

So I'm going to talk soon about this window. But basically, if you click on the Create Avatar, As you see, I created for myself an avatar, that's Frank.

But you have here public avatars, professional, casual. So you can choose also avatar that available on the platform itself. So as you see, you have a lot of options.

But I will recommend to start with the script. So the script can be a record of yourself recording, say, whatever you want, or a script. And what's nice, you can even generate inside HiGen.

So you can use ChatGPT and to create the script for your podcast. But in this case, I want to upload my podcast audio. This take a little bit time, depend how long it is, how heavy the file.

So I would recommend to start with the script and then you can choose your avatar, take your time until it's upload. So I upload my audio. As you see, I have a couple options.

I chose the best one and it takes time. So now you have templates. So with iGen, yes, you can create avatars videos and all that, but you can use it also for your work, for marketing, to sell your company or whatever.

So here's some looks that I film and I can add up to 100 looks for free. And hold on. And as you see, I have a lot of options, and they tell you how much looks each avatar have.

I would recommend to choose avatar that have as much looks, like 18, 10, because then you have options.

You can change their outfits and their camera angles, all that prerecorded. So in this case, he have 19 looks. videos already. So you see you have different angles, different outfit.

So it's nice to have more options. So now, as you see here, Here, down here, it's the audio of our broadcast that I already uploaded. Here is the character I chose, and that's his voice.

So when you record yourself doing the avatar, basically it's come with a voice. So it's also record your voice. But you can change the voice. It doesn't really matter.

And I'm going to edit the video. It's very easy to make it. It's not a rockin' science.

While I edit the podcast, I listen to the audio to know how long the guy talk, when the other host coming in. And you see I'm changing, I'm editing it, and I make the changes. and it takes a little bit time listen not everything is second like that and i go and choose the other character so i choose the look you see she have also a lot of options and i choose her because um both of them have almost the same background so in the podcast it just can be nice and i keep editing the video as you see here and base who talks

And when I finished, I just simply submit it. And in this case, I got a notification when I submitted it. So in the notification that you will see, Just a second. Okay, this one.

So basically my audio was very long compared to the video I created. So basically what they tell me is like whatever you left, it just won't be safe, which is fine. It's just for today, I'm not going to make a podcast of 20 minutes. One minute will be enough.

So that's, so, oh, just a second. So if you want to do your own avatar, like you saw in the beginning, you just simply go here to create avatar. So I already have my own.

It's the same process. You just edit and you look and then you have two options or you record from your phone, professional camera or from your webcam. I would recommend, to be honest, do it from your phone or a professional camera. It looks way better than your computer.

But when I tried with my phone, they're very specific. Like, the quality is not 100% try again. And I find that with a computer, it's just way faster and way easier. So that's what I did.

I turned my camera and my mic. And then you will see myself here setting up and try to talk for myself about 30 seconds. Nobody cares what you say in the recording. Yes, they just need to record the camera like you talk and they record your voice.

So when you have the avatar, it will also sounds a little bit like you. I won't say 100%, but you can say about 90%. And then I'm talking, and you see I'm talking to myself a lot, talking, talking, until I finish.

Now, what I really recommend to do when you talk, talk slow, but not too slow. Don't be like, hi. No, be like, hi, my name is Adas Frank. Because you need to have break between sentences, and also in the tutorial, they tell you that. Otherwise, it won't come really good.

And then you check the boxes and you submit it. You check also the video as well and you like it. The video is good. I look in the camera. I do some hands move, like I move. I move my head and I submit it.

And in the email, I will get a notification about a minute or two. Your video is ready. You have another look.

So you basically can record yourself walking outside in your office, library, whatever. And then when you want to have a video with your avatar, you have options. So that's the final product.

Deep Dive into Notebook LM

Today we're going to take a deep dive into something that I am so excited about. Oh yeah. It's Notebook LM.

It's pretty cool. Yeah, you might have heard about it. It was in an experimental phase, but now it's officially graduated and it has some really cool new features.

Yeah, I think what is really captivating about Notebook LM is how it like supercharges note taking with AI. Right. So imagine a world where your notes can actually like understand and synthesize the information for you.

So it's not just like a digital notepad. Right. It's more like an AI research assistant that lives in my computer. Exactly.

And the thing that's so unique about notebook LM is it becomes an expert on your sources. OK. So you upload your documents like research papers whatever you're grappling with and it just dives in and becomes intimately familiar with the content.

So if I was trying to like wrap my head around something super complex like the latest advancements in quantum computing I could feed notebook LM all of my research papers and it would become like my own personal quantum computing tutor. So that's an example of what you can do. It's pretty cool. It sounds very good also.

Practical Applications

But just to summarize what we did, so we created the audio, Notebook LM, then we took it and we upload to Hygiene to create the visual like you see here for the podcast. Now, probably some of you will ask, OK, that's nice, that's cute, but how this is practical for my work? This is example of what you can achieve.

You always need to be creative, especially in this AI new generation, you can say even. But you can use Notebook, for example, even generate a podcast inside the office for your team to teach them new things or a new product that the company launch. You can use Hygen for marketing even.

to sell your product, advertising. Sometimes on Instagram, you can see avatars trying to sell you stuff, or even people have now the digital creators on Instagram, TikTok, and all that. But you really can be creative and use it for your work, but that's just an example of what you can achieve.

Conclusion

So thank you very much.

Now, if you have any question, I will happy to answer. Yes.

Q&A Session

Because the avatars on Meta, the celebrities, they were contracted to sell their personalities.

This tool is free, right? Am I giving away my personality rights to this agent tool, or...

Okay, so with iGen, it's not 100% free. You have a free plan. So the free plan is about, like, you can create three AI avatars videos a month, and the video is about up to three minutes.

Now, it's not available. Let's say you record your own AI avatar. They don't sell it. It's private. It's only your avatar.

The public avatar you see, it's people that work with the company. They're probably being paid to do it. So a lot of people using it. And from what I know, whatever you created, your avatar, it's only for you.

Yeah. Yes. .

So it's actually worked pretty well, I don't know, like in my computer. I don't know why here it's had a little bit delayed, but there is other AI tools available that do more the matching with the lips. So in the worst case, let's say you don't like it, and you think you need to work on it a little bit more,

There is more tools available that will help you with that. But I'm sure they will keep working on that to improve it. But I can show you on my computer. It looks more pretty real.

Yeah. Yeah. Yes.

One thing when it comes to the script that they're using, I think sometimes LLMs can take a source of data and then talk about it in ways that are maybe not super appropriate, like they'll make claims that are a bit outlandish, or they'll speak with a lot of confidence. So are there ways to monitor what the script looks like, edit it in any way, or see analytics on how different scripts and videos are performing after you deploy?

Okay, so in this case, with Notebook LM, you can customize the script. You can ask, and you can always retry like doing it. You can try again and again until you like it, but they're usually very focused on the specific link or source you provided.

They don't give on data they have. Based on what you upload, that's what you will get. with the script inside Hygen or you write for yourself the script or you use ChatGPT.

Now, if you use, of course, any LLM that you want to use, you need to go over it and you need to make sure that you like the output and It's fine. But it's always good to review before you publish or send it.

But in a second, I finished the video. Nobody really can edit it. I done. The video is out there.

So it's not like another AI can change the script or something like that. But always go through it. Yeah. Yes.

what kind of stream you're producing, potentially another AI can pick up and pick apart the stream So let's say in our case, like the podcast, I can upload it even to YouTube, to social media. I don't know, to be honest, like as far as I know, in a second you put it out there, it's like a video.

You put it on social media, nobody really changed your video unless somebody took the video you upload, change it and then upload on their platform or whatever. on different pages. So this is like a real video like you put out there.

Yes, I use the AI to create it, but this is a normal video in the end when it's come out. It's video information with audio and video streams. Potentially another software can take it and modify it.

Yeah, maybe, maybe. To be honest, I can't tell you this information. I don't know.

But from what I know, like you put it, as I said, on YouTube, for example, yes, if somebody take it and do it, then if you brought iGen, then it's your... you own it, then it's a different story. Then you can come to the person and say, hey, this is my product. This is my video.

There is a problem. Yeah. Okay.

This is it. Thank you.