Using AI to Create an Audiobook on YouTube

For a challenge to myself I wondered if it would be possible to create an "audiobook" version of my stories,  but for free. Or at least for the lowest possible cost. Is it possible to create such a thing with little to no experience? 

I've heard a lot lately about using AI text-to-speech to produce human-sounding audio. This got me intrigued.

Video background

Firstly,  I needed a background that was consistent with the story being told. I've chosen my Halloween based flash-fiction piece, Tasty.

I went to Canva.com for this, and used my free account. I don't have a pro account, but I was able to use Canva to create a 20 second loop of a night-time scene. Adding the image of the sweet I created for the cover was simple because I'd used Canva to create that too.

This loop was exported as an MP4 to my computer. 

Creating the Video

Next was the video itself, but where was I getting the audio from? I could narrate it myself,  this is a valid option anyone could use. However, the idea here is to use AI.

That's where Clipchamp comes in. Clipchamp is owned by Microsoft and now comes with Windows 11. Within Clipchamp is an AI reader that allows you to convert text to speech. There are so many choices over language, voice sound from either make or female. You can even select a language and have that applied as an accent to another set off text written in a different language. For example, you could select French and write English and then the speech will be in English with a French accent. 

The tool has rudimentary settings you can apply to amend speech pattern. Choose from speed of the speech, from slow to fast, and add pauses.

In Clipchamp I needed to copy in the video image loop from Canva and then paste in the text from my story. There were a few foibles with the audio generated that I had to change the text in parts.  For example, "and inviting" was read so quickly it came out as "uninviting". I amended the text to say "so inviting" instead. There were two other parts I needed to and for similar issues.

The 20 second loop of video was woefully short, meaning I had to copy and paste it several times to cover the whole audio section. 

The Final Product 

The final video needed then to be exported from Clipchamp. This took several minutes. While the video was rendering, I connected my copy of Clipchamp to my YouTube account. 

When Clipchamp finished the export process, it uploaded the video directly to my channel.  You can see the final thing here: