Create an Engaging AI Generated Podcast in 3 Easy Steps

For publishers, an AI generated podcast can be more than just an audio format. It can be a convenient way for readers to discover and follow your content, it can drive people back to your website, and eventually it can scale to become a new revenue stream.

It’s also another content distribution channel as publishers make the move from being print-centric to digital-centric.

Of course, producing daily or weekly podcast episodes can be resource-intensive and expensive. This is where the concept of a fully AI generated podcast comes in. By creatively combining a couple of artificial intelligence tools, it’s easy to create a daily or weekly audio podcast summary of all your recently published content with just a few easy steps.

Why Not Just Use Text-to-Speech (TTS) on Articles?

Many publishers, like The Atlantic, use text-to-speech technology on their articles. These systems automatically ingest your content, create an audio version of the article, and then embed the audio version of your article in a player on the article page. Once you have it set up, there typically is no user intervetion or production work necessary.

However, the vast majority of readers landing on an article opt to read it rather than listen to the audio version. This reduces the value of TTS and makes it hard to get the scale needed to monetize it. The cost can also get quite expensive if you use text-to-speech on every article. As a result of these factors, several publishers I work with have decided to remove TTS from their site.

From RSS Feed to AI Generated Podcast

A more practical approach is to produce a daily podcast episode that recaps all stories that you’ve published in the past day. Or, if you don’t produce enough content for a daily podcast, you can always do it less frequently.

The idea is simple:

  1. Use ChatGPT to ingest your website’s RSS feed and create a podcast script summarizing the stories from the past day or week.
  2. Copy/paste the script into an AI text-to-speech tool to create an audio file.
  3. Upload the audio file into your favorite podcasting platform for distribution to Apple, Spotify, Google, other podcast networks, or even embed it on your own website.

Creating the Script and Audio Files

When I first gave ChatGPT an RSS feed to summarize, the script it gave back included superfluous production notes and host references. With a few tweaks to my query, however, ChatGPT created a clean audio podcast script that was surprisingly well-crafted.

I then copy/pasted the script into Amazon Polly and BeyondWords, two of my favorite text-to-speech tools. Amazon Polly can only use one voice at a time, so while the audio quality was good, the single voice got a bit monotonous. With BeyondWords, however, you can have multiple voices for each audio. I chose to alternate a male and female voice for each story summary which helped keep the audio interesting.

Once you have the audio file, you can then upload it to any podcasting platform. I prefer Buzzsprout, but any podcasting platform will work.

What’s Next for AI Generated Podcasts

Using AI streamlines the process for generating daily or weekly article summaries and can create a compelling podcast for any publisher. It only takes a few minutes for a production person to generate the script, create the audio file, and publish it via your podcasting platform.

But this can be taken even further. The entire process can be automated where a publisher won’t even need a production person. You could set a schedule for which days you want the podcast summary to run, and everything else can happen automatically.

Fully automating an AI generated podcast is not quite there yet, but we’re actively working on solutions to this right now and the technology is getting closer every day.

Share this post …

Sign up for my newsletter

I send digital revenue, audience and technology tips for publishers once per week.