NotebookLM Podcast Example: Create a Faceless YouTube Channel with AI video podcast hosts!

Welcome back, everyone! Thanks for tuning in. This is Samier Reviews, and today we're going to do a walkthrough from beginning to end.

We're going to create a Notebook LM video podcast, and we're going to try to do the whole thing from beginning to end. We're going to do a complete podcast, and I'm going to walk you through the steps of this workflow that I created and posted in the last video. We're going to go through each one of these steps and actually create a video podcast using Notebook LM audio overviews.

Let's go ahead and start with Notebook LM. All right, so what we're going to begin with is we're going to upload our source for Notebook LM to create the podcast from. And again, you can use any kind of format, whether it's a YouTube video, a website article, a PDF file—anything.

For us, since we're going to do a book review, I actually have the PDF of a book that's open-sourced right now. Excuse me, it's public domain right now, so I'm going to pick it up in my folder here. We're going to choose our file to upload, and I've already created the folder for it.

Give me one second. All right, so that's the PDF. As you can see, it's already doing the summary.

Now, one of the things we're going to do here is that Notebook LM lets you customize the audio overview so that you can prompt the AI into the style and purpose of the audio overview. If you want to do more of a study review, you can do that; if you want it to have a funnier style, where it's more entertaining, you can do that. So let's go in here, and what I want to customize it to today is to tell Notebook LM to—let's type it in—summarize this PDF of the book into an informative but short review.

Keep it under, let's say, under 10 minutes. Let's aim for, say, under six minutes in length. Let's summarize the PDF into an informative and entertaining podcast, but please keep it short.

Okay, let's go ahead and let it generate that. The reason we want to do this today is that I don't want to create a super long podcast. I wanted to keep it more short-form, so maybe five to six minutes in length.

So let this run, and while it's running, what we're going to do next is create our AI characters. Remember, we're not going to do the separation right now because it's still processing. That was step one; Notebook LM is in progress right now.

Obviously, we can't do step two because this is still in progress. So the next step is to create our AI characters. As you know, I signed up with Replicate.

What Replicate is, is basically a website where you can sign up, and it has multiple AI tools all in one place. When you sign up with Replicate, you don't have to sign up for each individual tool. If you go here in the search, you can look for almost any kind of LLM or, especially for image generators.

It's all under one roof. What you want to do is look for Black Forest Labs Flux 1. 1.

That's one of the best image generators out there. I would say it's very comparable to Midjourney, and today, that's what we're using to create our AI characters. As you can see, I already had the prompt loaded, but I have my prompts here that I pre-created, so I will just copy and paste this into our image generator.

Whenever you're using Flux, the only thing I would recommend is that you can keep all the settings the same. Obviously, you can change the aspect ratio; I'm going to leave it at one by one. But down here, where it says "output format," don't use WebP—just use the JPEG or PNG.

I'm going to go with JPEG here; output quality 80 is fine. Since I already did my prompt for the female host, we're going to create two images: one for the male host and one for the female. Let's go ahead and run this and see what we get.

It doesn't take long, so let's see. Okay, I don't like how this is going into her sleeve. I'm not quite happy with that, so let's go ahead and run it again, and if we run it again, it should give us a different output.

Okay, yeah, I like that one! I like this one a lot. We'll use that one, so let's go ahead and save the image as—make sure it goes to the right folder.

Okay, and then we're just going to save it as "Female Host. " All right, and then we're going to do the same thing with the prompt that I had for the male host. So let me go ahead and copy and paste that, and I'm going to put it right there.

Run that, okay? Now, if you're wondering, uh, these are, um, whenever each time I run it, it's not free, but at the same time, you don't have to pay anything monthly; it's just pay as you go. Um, each time I generate an image, it literally costs me just 4 cents, which is, um, pretty, pretty reasonable.

Um, for this one, I would prefer he looks straight at the camera, so, um, let's run it again. If it's not looking straight at the camera, I might have to put that in the prompt, so let's run that again. Okay, since he's not looking at the camera, he is looking straight at the camera.

All right, let's go ahead and run that because, um, okay, um, I do like this one. The only problem is, um, the mustache is kind of covering up the lips, and I don't know if that's something I want to do with the, um, with the image to video, so let's run it again. I did like that one; that one's not too bad.

Okay, I like this one a lot. All right, this is, yeah, this one's a good one. I do like this, so we're going to go ahead here, click, and save image as mail host.

Save. Okay, all right, so we saved that. All right, now let's go back to Notebook LM and see if it's done processing.

Okay, so the first thing that I realized right off the bat is it did not, um, make it under 8 minutes. Um, it turned into a 12-minute podcast, which is fine. Like I said, there are some limitations with Notebook LM; it doesn't really give you too much control over how long it is.

Um, but it did shorten it for us because typically, um, it can sometimes go as high as 30 minutes. So for this, uh, tutorial, um, we did it for 12 minutes, so let's go ahead and download this. Well, first, let's listen to it and see how it sounds.

Welcome back, everybody. Today, uh, we're going deep on a book that's been around for well over a century. Wow, that's a long time!

It is. It's *As a Man Thinketh* by James Allen. Oh yeah, I've heard that one!

It was published back in 1903, but people still, you know, love it. It's a self-help classic and we're going to be looking at some excerpts, some really key parts from the ebook edition, cool, published by Global Gray. So we're going to get to like the heart of it, you know, the good stuff.

Sounds good to me! So are you ready to unlock the secrets of this book and see how we can apply these ideas? Even, okay, I like that a lot; it came out pretty good.

Um, so let's go ahead and download this, all right? And I'm going to make sure that it goes to the right [Music] folder. Okay, all right, so now that we got it downloaded, um, we're going to go to—let's go back to our overview, our workflow.

So this is completed; step one's completed. Uh, the next step is we have to separate the audio files, the male and female speaker, into two separate audio tracks. Uh, we're not going to do that manually; we're going to use this tool, Speaker Split IO.

We go right here, already logged in. We're going to go over here to upload it, and yeah, that's the right one, so we're going to hit open and we're going to hit process audio. And again, um, this tool will basically go into the audio, um, the Notebook LM audio file, the audio overview file, excuse me, and it's going to separate out the male speaker and the female speaker into two separate audio tracks so that when we go later on to, uh, dub them or, um, basically lip-sync them, they're in their own audio channels.

That makes it super easier when you go to edit, and this does it for you automatically. As you can see, it's already done. Um, and it also gives you the transcript that's diarized, so it has it separated from, you know, who's speaking.

So let's go ahead and download this—uh, that's the male track isolated—and then we're going to download this, and this is the female track. Okay, so now that we've got it downloaded, let's go back to Notebook LM. All right, so this is a 12—let's say it's a 13-minute-long, okay?

Now, our next step, since we're done with step one, uh, creating the audio overview with Google Notebook LM, that's done; we separated the male and female audio—step two, that's done. Now, obviously, um, let's take a moment here to just see what happened. So it hasn't even been like 5 minutes, and we've already created an entire podcast—the audio file—we separated the audio, uh, using AI tools.

So we basically did, um, almost like 3 or 4 days' worth of work in 5 to 10 minutes already, and that's the power of AI tools! So imagine if you had to hire, uh, some voice actors or, um, or a host. If you had to produce a show, if you had to write the script, if you had to edit and narrate the script, if you had to record it, and then if you had to do all the audio engineering stuff at the end, post-production.

Just imagine, um, in 10 minutes, we were able to do all that using AI tools. So it greatly cut down the curve of creating content, and your role in this is like an executive producer. Okay?

Your talent is AI. Okay, your talent is AI. Your Um, your engineering is AI because you know we're using a lot of audio engineering, mixing, and different things like that.

So where you come in is you're like the creative producer—creative. Actually, you're also the executive producer because you're going to think about how to monetize this. So, just keep that in mind, how powerful these AI tools are, and putting you in the position of being a producer or an executive producer for this type of content.

All right, so we're done with that. We generated our AI characters using Flux, so step three is done. Now we're halfway through, guys.

The next step is we're not going to change the audio voices today; there's no need to do that right now. But if we wanted to, we could do that. I do have another tutorial coming up on how to change or remix the Notebook LM audio overview voices, but for now, we're going to skip this step.

We're going to go into Hedra; that's step five. Okay, so let's go to HRA, and this is the app that basically animates our AI characters. Give it a second; it's loading right now.

If you remember, we go back here—it's a 12- to 13-minute long podcast, and the account that I have is the Creator account. So, the maximum length of each CLI is 2 minutes for this plan. Now, if I had the professional plan, each clip I could do would be up to 12 minutes.

Well, guess what—they're 12 minutes and 57 seconds, so it would reduce the amount of clips that I would have to go stitch together in post-production. So just keep that in mind. For today, this is the plan that we have, so we have to stick to the two-minute clips.

All right, so we're going to go back to create, and what we're going to do here is upload. We're going to go back and select the audio files. Oh, wait a minute.

What I didn't do was the downloaded files. Hold on, is this the right one? Yeah, all right, let's change the names because I think he was the “m.

” Okay, let's go back and upload that again. I don't know why I didn't upload it. Okay, I see what's going on.

This is going to be something common you run into with Hedra. Right now, the audio output file that we got and separated—they're too large to upload to Hedra directly because they're longer in length. So what we're going to have to do is convert it from a WAV file, which is really high.

As you can see here, it’s around 67 megabytes. So, what we're going to do is convert it from a WAV file to an MP3, and what that’s going to do is basically reduce the file size so that we can upload it to Hedra, which is pretty simple to do. So I'm going to go to my audio editor.

You can use anyone out there; this is the one that I use. Let’s go ahead and do that real quick. Okay, so this is going to take a second.

We’re going to close that, and if we want to, we can go back here and check. So this is the file that just outputted. As you can see, it reduced it from a 67-megabyte size file to 12.

So now if we were to go back here, go back to Hedra, and choose the MP3, see how it loads up perfectly? Okay, so what we’re going to do is, since the maximum output that we can do is 2 minutes, we're going to go from 0 to 2 minutes. Welcome back, everybody!

Today, we’re—okay, so right here in the Hedra editor, it lets you segment your isolated male audio track. This is for the male speaker, and it lets you segment it into 2-minute slices. You can see I started from 0 to 2 minutes.

Why did I do that? Because that's the plan that I have; the maximum length I could do is 2 minutes. Okay, so I’m going to have to go back here and do that all over.

Sorry about that, guys. Let’s make sure this is exactly 2 minutes. Okay, all right, so we do that exactly 2 minutes, and then we’re going to go up here and upload our male.

Yeah, and what I’d like to do is, even though this ratio is fine, I don’t want to make it too big. See how you can zoom in that close? You don’t want it that close; that’s a little too close.

I think that’s about right, so we’re going to hit generate, and that’s our first two minutes. So what do you do next? Okay, well, since we did the first two minutes, what we’re going to have to do is start at 2 minutes and then go to 4 minutes and do our next segment.

So we’re going to move this to 4. Okay, so does everybody see why we did that? Because remember, the maximum video output that we can do is 2 minutes long.

Okay, there it goes. So once it starts generating, then it’s good. The step is to go here, and we're going to go back to four [Music] minutes from two minutes to four minutes.

Since we're using the same image, we're just going to hit generate. Okay, and what that's doing is it's going to create our next clip for us. When we go down here, this is the first one it's doing, and here's the second one.

So, the second one is queued up. Okay, all right, so while that's working on the male, let's go to the female one. So, what we're going to do next is we're going to go back up here.

Well, we have to go back to the audio and do the female audio the same way. All right, so the female audio is ready. We basically converted the waveform into an MP3, and we're going to go back to Hadra, and that's still working on our image-to-video conversion.

We can go ahead and discard that, and now what we're going to do is upload the female. Yeah, we'll do that. When you crop in your images, I, again, don't want it to be all the way zoomed out.

Even though that does look nice, you want to kind of make sure that it gets the most accurate kind of lip syncing, and sometimes if it's far away, it doesn't look as nice. So, we'll keep it cropped to there and hit generate. What that's going to do is it's now working on her clip.

So, what we're going to do is at the end, when we're going to take this to our video editor, we're going to stitch all of these together, and we have the whole entire podcast. Okay, and so that's queued up. We just want to make sure it starts to process.

There we go! See, when it says generating, that means it didn't fail; it's going through, which is good. Our other clips are almost done: 74 minutes, 51; 74%, 53%.

So it's working in the background. We're going to go here and do the same thing we did with the other one. Since we did from two minutes to two minutes, now we're going to do from two minutes to four.

So we're going to go this—this is going to go all the way to four. All right, so we did two to four, and we're going to do the same thing and hit generate. While that's processing, while these are processing, we're going to go into our sixth step, which is the video editor.

We're going to go to CapCut, and we're going to create our template to stitch together our Hadra AI video clips. Okay, so let's go ahead and open up CapCut real quick. All right, so we're going to start from scratch.

The first thing we're going to do is create our template. Now we're going to keep a similar template to the one that we've been using. So, the very first thing that we're going to start with is a background.

We're going to go into media, we're going to go to library, and then we're going to go to background. The kind of background that I wanted was like an abstract video background where it's—let's do gradient video. Yeah, let's see if it is the gradient video.

Okay, I like this one, so we're going to use this one as our main background. All right, next, once we've got our video background, what we're going to do is stretch—we can copy this out. All right, so since we know that we're going to have at least four minutes of clips to stitch together, we're going to extend this out to about four minutes for this tutorial.

So now we have our continuous video background, and so when it runs, it looks really nice like that. All right, so the next thing we're going to do is, since we're doing a book review, let's find the book cover that we want to use. So what I do is just I'll go online, and I'll look for the book cover of "As a Man Thinketh.

" We're going to go to images, and what I want to do is look for a book cover that looks really nice because there usually are multiple book covers, and we want one that actually looks like a book cover. All right, now what we're going to do is shrink this down, and we're going to add some animation to this. All right, guys, so we put a little animation on the book cover, and I like that.

So, what we're going to do is next, we're going to shrink that down a little bit because we want to leave room for a few of the other elements that we're going to add in. So the next is going to be our title. We're going to go over here to text; I actually do like that one.

All right, so next, let's animate. Let's do a loop; let's see how this looks. Yeah, I like that one right there.

All right, so it's coming together. I like that so far, and you can add different elements, but the thing to keep in the back of your mind is you're going to have to cater this to the style and creativity of what you want to present. The thing with any elements or backgrounds is to keep them dynamic.

Some things that you want to keep still, but for the most part, just keep things kind of interesting and, um, kind of, uh, but not too distracting. You don't want to put too many elements on your video template; you just want a few things to add some visual appeal. Because generally speaking, with podcasts, there's a visual element to it, uh, if you're doing a video podcast.

But for the most part, the people are going to be watching your AI actors, and then they're going to be listening to it, because they could be doing stuff in the background too. But, um, just keep it—keep some things visual, but don't overdo it. Okay, so we've got our title; keep it over there.

We got our book, which we're going to do here, and then I'm going to add another subtitle. So, I'm going to add another subtitle. All right, it's coming along; looks good so far.

So, um, let's go check back on Hedra to see how we're doing with the. . .

Okay, so it looks like our clips are done. So, let's go ahead and download these. Come back, everybody!

Today, we're going deep. All right, so this is the first clip. Now, what we're going to do is, um, the first thing that I like to do, um, with my style—because we have an actual template that we're using—is I'm going to throw a mask.

So, we're going to go in here and do a circular mask; bring this up right there. Okay, we're going to feather the edges, and then go back to scale and adjust the scale. I think that's really good.

Yeah, all right, let's hit play and see. . .

on a book that's been around for well over a. . .

All right, so that can turn out good. Now, let's go ahead and add the second clip, and we're going to stitch that—uh, probably not good things; not a recipe for success. He uses some pretty striking examples in the.

. . Okay, so what happened there is just basically the way that the audio files were clipped.

There was a little bit of a gap, so I was trimming that gap off, so that the transition from when you're stitching together the Hroc clips would be a little smoother. Um, now there is a more advanced way of editing this, um, without using Hedra. I'm going back here; when we were clipping our audio file, we were doing it exactly at two-minute cutoff lengths.

The problem with that is sometimes if you cut it off exactly in the middle of a waveform, when you go to stitch the other one, sometimes there's a gap. So, what I've done before is I will edit my clips into two-minute long clips—or not exactly two minutes, but less than two minutes. But since I can't see the waveform here, I will use an audio editor and do exactly where I want the cut, so that it doesn't cut off anything, uh, when I do a video stitch.

Um, that's too advanced for this tutorial, so we're just going to do what we did with this edit, which is kind of have to manually go in and fix the video edits to make sure that transition is smooth. But that's not a big problem. Um, it does take a little bit of knowledge of audio engineering, um, or audio manipulation in something like, um, you know, any kind of audio editor.

You would basically go here—let me show you this real quick. So when we were in Hedra, we were cutting it off or we were doing two-minute clips exactly, right? So, let's go here and let's say that this is two minutes right here.

Okay, now let's say that on Hedra we cut it off right here because it's exactly at two minutes. Well, look, it’s in the middle of an audio file. So, if we were to cut it off right here in Hedra, then it's going to do this little video clip, and then it's going to start again over here, but it's in the middle of a waveform.

When it does that, sometimes there's a gap. So to prevent that, we could have manually come here and cut it off right there, and it's less than a minute; it's one minute and fifty-nine seconds. Then that would have saved that audio file from getting cut off, and then you can export these little clips and then upload those audio clips into Hedra, if that makes sense.

But, um, going back to the video editing, now let's add the female speaker. Okay, all right, guys, we're back to, uh, Hedra. What we're going to do, we're going to regenerate the female AI host with a different.

. . um, we're going to pick a different AI host.

I think what happened with Hedra is it kind of messed up the lip sync for the, um, for the first female that we did. Um, I think the issue was the way it was cropped, um, and also how small, uh, she was in relation to the whole scene. Now, sometimes, like I said, it will kind of mess up.

Um, sometimes it'll make the lip sync kind of look off, and with hers, it did something weird with it. So, I went ahead and just did another AI model, and I fixed the crop. Kind of, as you can see here, it just takes up a little bit more; the face is more prominent than it is right here.

Just keep in mind that, uh, when you go to do these Hedra AI, um, talking. . .

Head videos, you want to make sure that the head is like 50% of the whole image. When it gets to be back, everybody, today, uh, we're going deep on a book that's been around for well over a century. Wow!

When the head is smaller or it's more cropped in, you can end up having quality that's not as good. It could kind of mess up the lip sync a little bit, and that's what I found out. So, uh, if you don't like the results, you can always go back and start again, which is why you don't want to generate too many clips at the same time.

What you want to do is first test the first one to see how it looks, and if it looks good, then you can go ahead and do all the other ones with the same image. Since we only did two here, um, it's good that we kind of looked at it before we continued with the whole project. So that's the only advice that I'll have.

Then we're going to go back and drop the new AI actor into our project file, and we'll see how that turned out. Go back to our project file, and we're going to import our new host. And that's the power of AI, guys!

Just like that, we have a brand new host. So let's go check to see if the other clip is done, and it looks like it is. So we'll go ahead and download that.

Go right back to our file and import. All right, so much better! I like the female host video a lot better, and the lip sync—hers was a lot better, more fluid.

Um, didn't have any kind of weird thing going on. All right, and in the last step here, what we're going to do is add one more final element to the template, and that's going to be the audio visualizer—that's the waveform that is kind of synced up with the audio. What we're going to use for that is a free online tool called Vi.

What you basically do is you're going to create an account there. Once you create an account, you will upload your audio file and then choose the spectrum that you want. We're going to go with this spectrum right here, and what happens is when you play it back, that's what it looks like.

You just export it out, and it'll tell you what to pick. If it gives you an error message saying the file size is too big, just reduce it to 720p, and you should be able to export it. I’ve already done that, so what we're going to do is come back into our project here and import.

All right, um, we're going to import the audio waveform into our track. Go there, and now real quick what we're going to do is we're going to crop that all the way in because we're only going to do a 4-minute clip of it; we're not going to do the entire podcast. All right, and then as you could see right here, what happens is since it has a black background, it’s covering up everything.

So click on that audio waveform, and then go to remove background. Then we're going to do a chroma key, and then we select the black. All right, for a century?

Wow, that’s a long time! It is! It's *As a Man Thinketh* by James Allen.

Oh yeah, I've heard that one! It was published back in 1903, but people. .

. All right, and all we have to do now is just, uh, now that we've got all of our elements that we want, we’re just going to export, and we'll take a look at what the final product looks like. So we'll do *Video Pod Demo Exported*.

All right, and we'll take a look at it as soon as it gets finished rendering. Welcome back, everybody! Today, uh, we're going deep on a book that's been around for well over a century.

Wow, that's a long time! It is! It's *As a Man Thinketh* by James Allen.

Oh yeah, I've heard that one! It was published back in 1903, but people still, you know, love it. It's a self-help classic, and we're going to be looking at some excerpts—some really key parts from the ebook edition, Cool, published by Global Gray.

So we're going to get to, like, the heart of it, you know, the good stuff. Sounds good to me! So, are you ready to unlock the secrets of this book and see how we can apply these ideas, even though it's from so long ago, to our lives today?

Yeah, I'm in! Let's do it! Okay, let's dive in.

All right, so this book, *As a Man Thinketh*. . .

As a Man Thinketh. Yeah, it's incredible to me how relevant it still is today! I know, right?

Even after all these years, it's like James Allen really hit on something universal, you know, with his central idea that as a man thinketh in his heart, so is he. Like that simple statement has just captivated readers for generations. It speaks to that power we all have to shape our own lives through our thoughts.

I love that he uses the analogy of a garden! Oh yeah, that our minds are like gardens and our thoughts are the seeds that we plant there. It's such a good analogy!

But what are, like, some of the. . .

Weeds that we need to be pulling out of our mental gardens. Yeah, that's a good question: how do we actually cultivate the right kind of seeds for a flourishing life? Well, that's where Allan's emphasis on personal responsibility comes in.

Okay, you see, he's not just saying, "Think happy thoughts, and everything's going to be great. " No, no, he's calling us to be like mindful gardeners, to really look at our thoughts and choose to nourish the ones that align with our values—that makes sense with what we want in life. Think about it: if you're always planting seeds of doubt, worry, and fear, guess what's going to come up in your life?

Probably not good things. No, not good things—not a recipe for success. He is.

. . some, and all right, guys, we're back again!

So that was the final output, and again, we went through each one of the steps to create our final video podcast. This is the workflow; again, I'm going to post this right here, so you could take a screen grab, screenshot it, and yeah, go through each one of these steps. And that's the whole process.

If you guys have any questions about any one of these, drop them in the comments; I'll do my best to get back to each and every one of your questions. And if you have ideas for another video or something along the lines of this workflow or something in more detail, drop that in the comment section too. I'd love to hear your feedback.

Thanks for tuning in! See you guys later. Bye!