My Voice allows you to privately train custom models on your existing work, and then use those models to make prose sound more like something you’d actually write. Beginning today, My Voice is available in open beta to everyone on Sudowrite.

Getting started with My Voice is simple—just two steps. ✌️

1. Train a custom My Voice on your existing work right here.

2. Select that My Voice model when highlighting text to restyle prose.

The rewritten work shows up in the History column on the right, just like other AI features. Whether you use My Voice on something originally written by the AI or even your human writing partner, it’ll wind up sounding just like you.

FAQ

How much do I need to have written to train a My Voice?
You can train a My Voice model using as few as 1,000 human written words, but there’s also no upper limit (in case you’d like to train on a book you’ve already written). The most important thing is the style you’re looking for is well represented, and that you’re not training the AI on AI-generated prose.

What’s do you mean My Voice is in open beta?
In short, we’re still improving it and the way it works may change a bit. We’re looking forward to your feedback once you’ve had a chance to use it!

Will I lose my trained voices once the beta is over?
Nope—we may still tweak the user interface, or optimize certain things, but your trained voices will remain once we fully launch the feature out of beta.

Is my training data really private?
Yes! Both your data and your My Voices are stored privately in your Sudowrite account. They also won’t be used to train anyone else’s AI!

Does My Voice use credits?
Yes, My Voice uses about the same amount of credits that something like Rewrite does. The early feedback we’ve received suggests it’s well worth it.

Wait a minute, where did Canvas go?
You can still access Canvas from the Plugins dropdown menu.