Cloning your Voice with AI

Day 20 / 366

We have looked at text and image AIs, but AI has been used to generate audio as well. Last year we had lots of AI-generated songs getting viral, where they used the AI model of a popular singer for the vocals of a track they didn’t sing.

Today I looked into what AI voice cloning is, how I can do it, and whether is it worth the hype.

I looked around a lot, and like every other AI thing, there just isn’t enough information out there for beginners. As a coder myself I was looking for a way I can do this myself from scratch, on my laptop. However, I did not find any relevant tutorials. The best thing I found was this —

https://serp.ai/tools/bark-text-to-speech-ai-voice-clone-app/

However I think it would take me more than a day to explore, so I will look into it tomorrow.

But I still wanted to see how good AI voice cloning is, so I decided to check out some of the existing web apps that allowed you to clone your voice. Most of them were paid, which is fine with me, but they required you to get a monthly subscription.

Luckily, I found https://myvoice.speechify.com/

This website allows you to try voice cloning for me. You can either upload a few recordings of your own, or record something right there on the website. Once that is done, your model would be ready and you can type out anything you want to be said in your voice.

The whole process took less than 5 minutes. I recorded about a minute of me reading out some random text, which the website used to train my model. And this is the result -

I am impressed with the results. It’s crazy how good this is with just a minute of input audio, think of what could be acheive with more input data?