xAI has just released an API to clone your voice. It is pretty simple, read a script, and then an API where you can have text to speech in that voice.
Here is the python code after you have cloned your voice.
I need to figure out my audio set up a bit better (my mic set up is probably not optimal and it produces some echo). But does a good job imitating my boring voice right out of the box!
And here is an example for longer speech from my intro to LLMs book:
The LLM intro messed up *Application Programming Interface* section (start listening at 50 seconds in). But otherwise it is very nice.
For those worried about security, xAI did something smart here — you need to input text live into the API given their prompts. You cannot have a pre-recording audio input to do this. So cloning someone elses voice is pretty hard.
Costs are around $4 per million characters in the text to speech API. So say narrating my entire book should be under $10 I believe.
Took me a total of less than an hour to set up a voice, create the python code, and write this blog post!
Posted by Andy Wheeler on May 4, 2026
https://andrewpwheeler.com/2026/05/04/xai-voice-cloning-api/