How to Use NotebookLM to Make a Podcast From Your Blog Post

NotebookLM is the new AI everyone is talking about because it has one really amazing function: it can create a podcast out of anything you give it.

And I’m not talking about some robotic voice that sounds like Siri mispronouncing every other word.

This actually sounds like real people having a conversation.

I talked about this back in May with the Google IO Keynote where they first planned to release it, and mentioned the implications for bloggers.

I love AI, so I had to try it out … even though I’m a bit on the fence for actually using this one.

I’m gonna show you how to use it first, then we’ll talk about about the issues that might stem from this.

Turn Your Blog Post Into a Podcast

It’s incredibly easy – kinda scarily so – to turn your blog post into a podcast with NotebookLM.

Go to NotebookLM.Google.com to start. You will need to sign into your Google account to use it, but otherwise it’s free.

Inside of Notebook, you’ll want to click on the plus sign beside “Sources” to add any sources. Pictured here:

You have the option to add a Google Doc, Google Slide, URL from a website, Youtube video, pasted text, or to upload any file.

Lyrical Host tends to block AIs, so I went with copied text for my example I’ll show y’all below.

It takes a moment to upload, then the screen will refresh to show you your Notebook options that looks like this:

These include creating FAQs, study guides, Tables of Contents, timelines, and briefs. It summarizes the article, and suggests follow up questions you can ask the AI about it.

But that’s not what we care about right now.

We want to make an “Audio Overview”.

Click to “Generate” and it’ll create a podcast with 2 hosts speaking to each other about your article:

As of right now, it’s only available in English and you can’t customize the host genders or the content in any way.

Once you click generate, it takes about 4 minutes to render a podcast about your content.

Real Life Example

I published an article on my Trending Blog Post Framework a few months back. I keep meaning to turn it into a Youtube video, but I just haven’t had time yet.

I know many people learn better with audio, so I always try to convert my content for all platforms for accessibility.

So I copied and pasted the full article into NotebookLM and generated a podcast.

You can listen to it here:

The podcast has a male and a female host who are discussing my article specifically and pulling on details from it. Rather than speaking like they wrote the article, they talk about it like they’ve read my article.

I will say, they pull my example of a trending topic for dog bloggers as though I actually wrote the post.

But considering it took one click to generate, that’s pretty cool.

Note: You need to make a new “notebook” for each blog post. Otherwise it’ll lump them all together – which can be beneficial, but it’s annoying if that’s not what you intended.

Implications for Bloggers

The Good

One of the pros of this type of AI is the ability to make content more accessible.

As someone with ADHD, I often need to listen to things as I’m doing other tasks in order to focus. Reading a blog post can be challenging to keep my attention.

People with Dyslexia or who are blind will also struggle to read without a screen reader – most of which sounds like Siri is bored out of her mind.

This presents a really cool opportunity to make content more accessible for our audiences without adding a ton of time for us to spend recording it.

It also allows bloggers a potential secondary traffic source if they turn this into a podcast they use to promote their blog.

However, this will depend on podcast platforms allowing AI generated podcasts. Right now, I don’t think they’ve caught on that this is even a thing.

Now, I don’t know the legal implications of that as you can’t edit this. So I don’t believe the blogger would own the copyright to it, which causes issues.

You’d need to look into that more.

I also appreciate that NotebookLM doesn’t train on your data sources. This is really important for people who don’t want an AI to know all their secrets/train on their data. They show that here:

I can see this also being very helpful for course creators wanting to create SOPs or podcasts within their course, but not wanting an AI to know all their proprietary info.

The Bad

NotebookLM was demoed at the Google IO Keynote in May 2024 as a way for students to improve their learning by creating a podcast about their notes to understand it better.

I predicted it would very quickly be used to create podcasts for creative content, too.

Since launching, it’s already happening.

Bloggers on Twitter have already started launching podcasts built with NotebookLM, and a conference I was at even featured it on an AI panel.

Podcasts are one of the most intimate platforms because it’s a real human voice right in your ear.

Having AI jump into that feels wrong, especially if people can’t tell that it’s AI.

I worry that we’re going to see AI coming for podcasters next, and a rise of AI generated podcasts overall.

I think it’s fun to make something for your site as kind of a bonus, but I do really worry about the use of AI recently with new developments to push out human creators.

I predict people will push back against it and want a more human element again. Which should hopefully make the rise of AI podcasts short lived, if at all.

That being said, I think a podcast as a supplementary aid or accessibility resource is wonderful and can be really useful.

But use it cautiously, and always make it clear to your audience when you are using AI.

I don’t see myself using it in it’s current form, as I can’t customize anything like having female voices for “she” knows SEO. And I prefer to record them myself when I do repurpose content from my blog to my podcast.

I expect I’ll use it more like a fad – like I did with Siri when it first came out and then have not touched since.

Or, it’ll adjust to the point where we can customize it and even train it on our voice.

A friend of mine made an app called Dudley Editions that does just that. If you give it 2 minutes of your voice, it can record a full audiobook in your voice to send to friends/family.

Bonus Use Case: Social Media Repurposing

I shared this post in my free Facebook Group and one of my members had a great idea: use the podcast NotebookLM generates for social meida.

I can see this working well, although as Jenny of Alberta Mamas notes, it would require a workflow to be efficient.

Off the top of my head, I could see Vidyo.ai working to help with this. Take the podcast, turn it into an MP4, upload it to Vidyo, and have it clip out audio that would work for Reels.

You could probably set up a Make.com integration to do the reformatting, or just pop it into Canva or iMovie to add some graphic to accompany it for Vidyo’s upload.

Otherwise, you’d need to manually cut it in Premiere Pro or Capcut, which could be time consuming.

You could also use this for Youtube videos, perhaps with less editing since you don’t need to cut it up as much.

I don’t know how effective it’d be, but it’d be an interesting experiment to run.

Will You Try It?

I’d love to know your thoughts on it and if you plan to give NotebookLM a try.

I’ve created a discussion thread here in my free Facebook Group – please remember to answer the questions to join or it auto rejects people. Share your thoughts and how it went if you tried it.

This is a place for open dialogue, so keep it friendly but I am open to negative/positive debate topics.

And speaking of AI for accessibility, have you tried my She Knows Alt Text GPT yet? Get it free here.