Creating a podcast can be a fantastic way to connect with your audience, share valuable insights, and give your brand a voice.
Using AI tools like ChatGPT for scriptwriting and ElevenLabs for the voice can make this process smooth and professional.
These are the same tools we use to produce our Techpresso podcast on autopilot, without any human action, and accumulating several hundred subscribers and thousands of listens.
This tutorial will guide you through the steps to get your AI-generated podcast up and running.
Key Objectives:
- Define and nail down the tone and style for your podcast.
- Craft a structured and engaging script with ChatGPT.
- Produce a high-quality voiceover using ElevenLabs.
- Publish your podcast on major platforms like YouTube, Spotify, and Apple Podcasts.
Educating ChatGPT on tone and style
First, we need to get the tone and style just right. This step ensures that everything we create matches your vision and resonates with your audience.
Prompt:
I'm starting a podcast and need help to define the tone and style. The podcast is called [name] and is for [purpose of the podcast, e.g., business, educational, entertainment]. My target audience is [describe target audience]. I want the tone to be [e.g., formal, conversational, humorous]. Can you help me create a prompt template to set this tone for our future interactions?

Generating the podcast script
Now that we've got the tone and style down, let's move on to creating the content. We'll structure the episode and generate the script.
Prompt:
Let's create a podcast episode. Use the tone and style we defined earlier. Here's the structure I want: [e.g., introduction, main content, conclusion]. The topic for this episode is [topic]. Based on this structure and topic, can you help me draft an outline for the episode? Once we have the outline, we can work on generating detailed segments for each part.

Then let's actually make ChatGPT write the content with the following prompt. By the way, this is the exact prompt we use internally for Techpresso (we spent hours fine tuning it!):
I'd like you to start the podcast like this 'Welcome to your daily Techpresso! In today's episode, we'll cover...'. Then you name each news item and try to be fairly specific about what it is. Next, I want you to do one section per article. Make a transition between each article, and introduce them. For example 'Next, let's talk about...'. Each section should be around 2-3 paragraphs long and 7-8 lines long. Finally, end the podcast by thanking the listener. Keep your sentences simple, and stay factual. There's no need to use transition sounds or 'host'.

And because we don't want any titles, let's remove them with the following prompt:
Now remove the titles

Using ElevenLabs for voice generation
With the script in hand, it's time to bring it to life using ElevenLabs. This part is all about turning our script into a voiceover.
Go to ElevenLabs, create an account or log in.
First, select the voice that you think will work the best for your podcast. After trying a few of them, I decided to choose "Natasha".

Now, paste your script in the text to speech box, and click "Generate speech".

Once it's done, click the "download" button at the bottom right of your screen.
Publishing your podcast
After you've produced your podcast episode, the next step is to publish it on major platforms like YouTube, Spotify, and Apple Podcasts.
To publish your podcast on the major platforms, I recommend that you start by creating a podcast on Spotify for Podcasters.
Here's how it looks like once it's configured:

Once it's done, go to Distribution tab, and from there add it to the other major platforms like Apple Podcasts, Amazon Music...