How to create a Podcast using Artificial Intelligence Tools

Welcome to twenty First Day of my daily blogging challenge. Today is Monday, and theme is MyPod Mondays. Let us take a look at an exciting development in podcasting! Today we will discuss podbot.ai, a site that uses Artificial Intelligence technology to create podcasts. I have written about this tool in the past in other places, but not on this blot. In a way, you can consider this post as an update. And what an exciting update it is!

In This post

podcast episode created by Podbot.ai
Background: AI and Podcasts
GPT-2 and GPT-3
How to create podcast episodes using podbot.ai
How this AI tool for Podcasting works
What are the use cases for podcast created using AI
Podcast Episode Transcript

Listen to the podcast episode created by Podbot.ai

If the below audio does not play, you can visit the podcast page on podbot site.

What Would Happen When The World Runs Out of Coffee?

Background: Use of AI Tools in Podcasting

I had used podbot.ai in the past, probably a little over a year ago. I re-visited it recently to check if the site has any updates. According to their FAQ or Frequently Asked Questions page, podbot is a creation of two persons, Nimo Rotem and Willy Arisky.The underlying technology according to the website uses a variety of AI solutions. the founders mention that the

“app was created to demonstrate the awesome potential in several new artificial intelligence solutions and state of the art neural networks. Among those: Language models (GPT-2), GAN models, text to speech networks, contextual summarizes and more.”

In the two episodes I tested recently, they seem to have made some progress on this. In terms of quality, in terms of technology, we are getting there and mind you this tool uses GPT-2 technology, which is a generation lower than GPT-3. The latter is supposed to be much more agile and offer more bang for the buck. A lot of writing tools such as Rytr, Frase.ai and others use GPT-3.

Areas of struggle for podcasters. blog of Amar Vyas

What do podcasters struggle with ?

Confused between GPT-2 and GPT-3? This should explain it

Remember the Terminator movies? Think of Arnold in the form of T-800 as GPT-2. And Robert Patrick in the form of T-1000 as GPT-3.

How to create podcast episodes using podbot.ai

I created two episodes for the purpose of this blog post. Of course one has to create and account and a podcast feed in order to do so. My “podcast” on that site is called Weekly Seasonal. I had created another show, called seasons last year. Under that show, I had two episodes. One was about the weather part itself, titled Monsoons.

Episodes of Weekly Seasonal

At the time of publishing this post, I have five podcast episodes on Weekly Seasonal. Below is a summary of these shows

The first episode is titled “What Happens when it rains in February” It can be ignored- its output wasn’t the greatest.
The second really was great- it is titled “Top TV Shows on Amazon Prime” Its output was actually quite good!
In the third episode, titled “What would happen when the worls runs out of tea?” I actually made a typo in the title. Instead of writing “World”, I typed “Worls” and hit enter. I cannot edit the title anymore, but the output was relatively decent. There are three parts of the audio that I would like to highlight: First and third are vary close to the topic. The middle portion? Things kind of went haywire there.
The fourth was almost a class Act. It’s theme was similar to the third : instead of tea, I asked “What Happens When the world runs out of coffee?” This episode blew my mind. Barring a few repetitions of words here and there, the rest of the content is quite good.
Finally, the fifth episode is on using images on your website. I have written on this topic a few times on this blog, and it would be interesting to compare the AI generated output with my own research and experience.
There is a sixth episode- which will most likely have a limited audience. There is an old joke among the Linux/ Open Source community, which goes as follows : “Will the next year be the year of the Linux Desktop?” I thought it would be interesting to see what the AI tool thinks about this question!

How this AI tool for Podcasting works

Watch the video to get a quick overview

The only input required from the user is adding a problem statement. In fact, the podbot tool can even suggest a random topic or a theme. The statement you enter becomes the title of the podcast episode. This tool creates the script, narrates using speech to text to speech, adds royalty free background music, cover art, and upload the created mp3 file to the podcast feed. It even creates a dedicated page for the podcast episode, and publishes the transcript. All in a matter of minutes, depending on how busy the server is. Of course there are occasional errors where the episode cannot get published. But leaving that frustration aside, the overall experience is great.

What are the use cases for podcast created using AI ?

Obviously this tool is not ready for prime time. Here is one use I can think of right away. If you are a content creator, or are setting up a test website, you may be familiar with Lorem Ipsum. In early 2022, consider Podbot.ai as the lorem ipsum equivalent of podcasting. This is not dis-respecting the work of the creators, I am simply stating a real time use case. If you want to test a new podcast feed, a few sample episodes fro podbot can do the trick. Same can be said if you are creating a new podcast or audio streaming app, some of the content can be readily available. However, make sure you understand that this is experimental stuff. And the fair use or copyright issues are not clear at this time. Of course, a chat with the creators can be helpful if you are planning to use any content for commercial purposes. I have no such plans.

Lorem Ipsum in English	Lorem Ipsum in Hindi

Transcript of the episode: What Happens When the World runs out of Coffee?

Below you will find a mini transcript of the podcast episode- this is a verbatim copy + paste of the first paragraph. This was generated by the AI tool of podbot, I merely provided the title. The entire transcript is available on the podcast episode page of podbot.ai.

Click to read the Podcast transcript

Researchers at the Royal Botanic Gardens, Kew in the UK The Royal Botanic Gardens, Kew in the UK are warning that climate change, deforestation, drought and plant disease are threatening the future of coffee. The Climate Institute is not the first to warn of a bleak future for coffee beans. While most coffee researchers agree that agricultural production is in crisis, some argue that the focus on climate change and leaf spot is masking real dilemmas and potential solutions to coffee cultivation. Yvette Perfecto, a researcher at the University of Michigan who has been studying coffee for three decades, says it was the shift away from agroforestry, not just climate change, that marked the real crisis for coffee and the livelihoods of coffee growers. . Not only is the supply chain at risk, but the livelihoods of the estimated 25 million coffee farmers who earn their living by growing coffee. From small farmers to large producers, supply chain operators are taking steps to save coffee.

About this post

The theme for publishing blog posts on Mondays is MyPod Mondays, where I write about topics related to podcasting. You can follow my blogging challenge by visiting this page

Word Count:1200. Time taken for this post: 1 hour 30 minutes,
including updates prior to publishing and embedding audio player.

_{This post was updated on 2022-06-04}