How I Create a Podcast From a Blog Post Using AI

Have you ever wanted to share something, but not in your own voice? That was my biggest obstacle to starting a podcast for a long time. I wanted to make one. But I couldn’t do it without giving up the one thing that makes this project work.

Saul Barden is anonymous. That’s not just a trick — it’s the whole idea. If you could hear me talk on a podcast, I wouldn’t stay anonymous for long. My voice is as recognizable as my face. So for months, the idea of turning a blog post into a podcast episode sat on my someday list, right next to “learn the accordion.” Plenty of other bloggers already know how to create a podcast from a blog post — I just hadn’t gotten around to it. It was a nice idea, but I didn’t think I’d actually do it.

Then I started using AI to create the audio, and suddenly that obstacle was gone.

Why This Matters Right Now

Podcasts aren’t just a side project for blogs anymore. They’re where a large part of your audience already spends time. According to Edison Research’s Infinite Dial study, 58% of Americans now listen to podcasts every month. That’s a record high, and a big jump from just two years ago. That’s about 167 million people who’d rather listen to something than read it themselves.

There’s a real reason for this beyond convenience. Michael Kraus, a researcher at Yale School of Management, published a study in American Psychologist. It found that people understand emotion more accurately through voice alone than through video or text. Tone of voice carries something written words can’t. It’s also much harder to fake emotion with your voice than with your face. If audio beats video for conveying feeling, it’s not close. Audio easily beats plain text with no vocal cues at all.

So I was stuck. The format that connects best with people emotionally was exactly the one my anonymity rule wouldn’t allow. AI is what made it possible to have both.

The Throughline: AI Didn’t Just Make My Podcast Easier — It Made It Possible

I want to be clear about this. A lot of anonymous or pseudonymous creators believe podcasting just isn’t for them. I felt the same way until AI voice generation came along. Once it did, I had no excuse left.

It’s not only the anonymity piece. I’m not a scriptwriter. I was never going to sit down and write natural-sounding two-person dialogue from scratch every week. I don’t. I have Claude handle that part, which is really what makes this process doable for someone who isn’t a professional writer. If you have a blog and not a screenwriting background, that’s probably you too.

The Real Process: How I Create a Podcast From a Blog Post

I don’t write a podcast script from scratch. I use a post that’s already published on the blog and build the episode from there. Wondercraft, an AI audio platform, handles the actual voice work. Here’s how I do it, step by step.

1. Turn the Post Into a Short Excerpt and Show Notes

Before anything touches Wondercraft, I pull two things from the original post. There’s a short excerpt — a few sentences, for the podcast’s own summary field — and a longer set of show notes that link back to the full post. Claude is already connected to my WordPress site, so it can pull the content directly rather than me copying and pasting it in. This is also where I decide what the episode is actually going to cover. You don’t want to just read the blog post out loud for ten minutes. You want to talk through the highlights the way you would with a friend.

2. Ask Claude to Draft the Two-Host Script

I think this step matters most for anyone reading this who isn’t a writer by trade. I don’t draft the Saul-and-Lydia conversation myself. I hand Claude the excerpt, the show notes, and the highlights I want covered. Then I ask it to write a natural back-and-forth between two hosts discussing the post, in plain-text, speaker-tagged format. It reads like two people actually talking, not a script performing enthusiasm. And I’m not the one who has to make that happen from a blank page every week.

3. Get the Format Right — Plain Text, One Line Per Speaker Turn

Whether Claude writes it or you do it by hand, the formatting matters more than you’d think. Keep it plain text, no markdown, one line per speaker turn, with the speaker’s name in brackets followed by a colon:

[Saul]: Dialogue text goes here.
[Lydia]: Their response.

No bold, no headers, no stage directions like [INTRO] in brackets. Wondercraft treats anything in brackets as a speaker name, so only real speaker names belong there. I tell Claude to follow this format directly, so there’s nothing left to clean up. Then I save the result as a plain .txt file.

4. Hand the Script to Wondercraft’s AI Assistant, Not the Auto-Generate Flow

Wondercraft’s built-in assistant, Wonda, will offer to write you a brand-new script from a topic prompt. Skip that if you already have one — it’ll discard your work and write something generic instead. Use the attachment control in Wonda’s chat to upload your finished .txt file directly. Wonda reads it and builds the episode timeline automatically, picking up your speaker names straight from the [Name]: tags.

Why I Create a Podcast With AI Voices Instead of My Own

I’ll be honest: I don’t have a great voice for podcasts. I’m not being falsely modest. I’ve heard myself on old voice memos, and it just doesn’t sound right for this. AI voices, on the other hand, are genuinely good — warm, clear, and consistent episode after episode. My own voice, on a tired, scratchy-throat morning, never would be.

But the real reason isn’t vanity — it’s the anonymity rule I mentioned earlier. I couldn’t have made this podcast without AI. It’s not just that it would’ve been harder — it would’ve been impossible, full stop. My real voice is one of the few things about me that would reveal my identity right away. Two consistent, entirely synthetic voices solve that completely. And as a bonus, they sound better than the alternative I was never going to use anyway.

Assigning those voices is its own small step. Click into a speaker’s line in the timeline to open the “Edit speech” panel, then use the “Change voice” button there. Skip the little name pill above the line — it just toggles between two defaults, and it had me confused the first time. Listen to a sample before you commit to one.

The Turn: It Sounds More Personal, Not Less

Here’s what surprised me. I assumed a synthetic voice would feel like a step down from something more “real.” I pictured a lesser version of what a face-and-name podcast would be. It turned out to be the opposite. Listeners don’t need to know it’s genuinely my voice. They just need to feel like Saul and Lydia are talking with them, not at them. The warmth comes from the tone and the writing, not from whose vocal cords actually made the sound. Once I stopped worrying about that, the show got better, not worse.

Pro tip: once you’ve produced one episode, don’t start the next one from scratch. Copy the finished one and ask Wonda to update the script to match your new blog post. Keep the same voices, structure, and music in place. It inherits everything instead of making you rebuild it.

Finishing the Episode

Once the script and voices are set, Wondercraft’s Generate (or Generate all) button renders the audio. Export gets you the finished file. I add a short intro and outro music bed from Wondercraft’s built-in library the same way each time. It plays once at the very start and once after the last line, with the volume faded down under the dialogue.

For the cover art, I go back to a different AI tool entirely and generate it with ChatGPT. I keep the same visual style across every episode. That way the show reads as one consistent thing in a podcast app, not a new design every week.

The last step happens back on WordPress. I upload the finished audio into a Jetpack Podcast Episode block. Then I set the season and episode number and publish it to the Podcast category. WordPress.com’s own podcasting support docs walk through the block itself if you’re setting this up for the first time. That part isn’t AI at all — just plumbing.

Bottom Line

I used to think a podcast required either showing up as yourself or not showing up at all. And that either way, you needed to be a decent writer and a decent speaker to pull it off. AI voice generation and Claude drafting the script between them took both excuses off the table. I could stay anonymous and still have a better-sounding podcast than my own voice would’ve given me. And I didn’t have to be a scriptwriter to keep the emotional connection that makes people want to listen. If you’ve been wondering whether you can create a podcast from a blog post without giving up your privacy or your confidence about your own voice — you can. It doesn’t have to cost you a chunk of your week writing dialogue either. It wasn’t a close call for me, and it probably won’t be for you either.

I’ll be honest that I prayed about whether to even start this whole pseudonymous project before any of it existed. Voice included. 1 Peter 4:10 says to use whatever gift you’ve been given to serve others. I’ve come to believe that includes borrowed gifts, not just the ones you were born with. A voice I didn’t grow up with is still a voice, if it helps someone hear something true.

Seven episodes in now, and I still open every new one the same way I opened the first. Start from a post, write the conversation, let the AI voices carry it the rest of the way.

One More Thing Before You Go

If you want more content like this, subscribe here. You don’t have to get every post in your inbox. Just pick Tech or any categories that interest you, and I’ll leave out the rest.

What’s held you back from starting a podcast, or from putting your own voice out there at all? I’d genuinely love to hear it.

You Might Also Like

• How I Use AI Every Day (And Why You Should Too)
• Using Claude With WordPress
• How to Connect Claude With WordPress (WordPress.com and Self-Hosted)

Follow along on Instagram: https://www.instagram.com/saulbarden