Generative AI is being used to create amazing art; first through paid services like Midjourney and now also with free, open source alternatives like Stable Diffusion. Now you can even generate art in a particular style, first with Google's Dreambooth and later with open-source implementations of the same.

Is there a Generative AI program for audio that allows you to create a model/style of a particular voice exemplar?

I am looking to train a voice model in a particular style using the aural equivalent of Dreambooth, and then apply that voice model/style to written text. Ideally I could review several options, and then pick certain ones to expand upon and later upscale.

I have researched this extensively but all of the voice generation services I have found are 1) paid 2) closed source 3) don't allow you to train the software on specific audio samples or 4) don't sound very natural at all.

Are there modern generative AI services for text-to-audio?


1 Answer 1


You can check sample-generator, an implementation of stable diffusion for audio data from Harmonai. I'm playing with it recently and I can say it works pretty well already out the shelf. With some hacks I was also able to reduce the model size and run a training on cpu.

And if you have the proper gears they also provide pretrained weights of some of their models.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .