I came across Hume.ai yesterday and was immediately struck by the demo. The way it handled voice and emotion intrigued me enough to test the free demo myself. I was impressed.

I decided to take a short piece I had published on Medium two days ago — “A Brief Discussion About The Revolution” — and see how quickly I could turn it into an audio drama.

Sadly, after paying $10, I ran into problems . Errors, voices refusing to switch when the script required, and caching issues. Worse, I was being charged every time I had to regenerate a voice — not because of my mistakes, but because their software was broken.

Glad to say, my hopes were resurrected when I remembered ElevenLabs existed. There were some similar issues with voices not switching when they should . But that was easily fixed by clicking the “Apply” button a couple of times. Other than that, the experience was smooth.

Tools Used

• ChatGPT — Defined character tone and personality

• ElevenLabs — Generated character voices and downloaded audio

• Veed.io — Edited the audio

• Canva — Created the YouTube video

• YouTube — Hosted the final product

Process

I fed my Medium post into ChatGPT, asking it to describe the persona, voice, and character of Jules and Marcus. Using the Creative Writing Coach custom GPT, I extracted detailed voice descriptions.

I took the “Voice” description and fed it into ElevenLabs to create custom voices. 

I copied each line of text into ElevenLabs, applied the relevant voices, and generated each line of dialogue as an individual audio file. Once everything was set, I exported the entire project.

I also generated background sound for the café using ElevenLabs’ “Sound Effects” feature. I based the description on the first line of my Medium post, letting the AI create an atmosphere that fit the scene.

Next, I uploaded the dialogue and sound effects to Veed. Once the audio was finalized, I moved it into Canva, selected the YouTube Video format, added an AI-generated image, and hit download.

Thoughts:

This entire process took about an 30 minutes. Far from polished, but the potential is obvious. More time would mean a much stronger result.

What excites me most is the ability to create something reminiscent of old radio dramas from the ’40s and ’50s. Or even an audio adaptation of a book or newsletter. As a writer, I wasn’t sure how platforms like YouTube or TikTok would work for me. But this workflow, or a refined version of it, might just be the answer.