My Voice (Beta) – Sudowrite

Ever wish it was easier to make prose sound more like you?

My Voice lets you train a private, custom model on your work so that Sudowrite can instantly rewrite anything in your voice. It works like magic, whether what you’re transforming was originally written by the AI or a human writing partner.

✌️ There are two simple steps to using My Voice.

Train a custom My Voice on some existing work.

Use that My Voice model to restyle prose.

What you’ll need to create a My Voice model

Existing work: Have a sample of your own, human-written text ready to go. It should be representative of the style you’re looking to replicate.

Sample criteria: Minimum of 1,000 words with no upper limit. The sample quality is the most important consideration.

File types: Upload a .txt file or just copy/paste text directly into the modal.

⚠️

Important note

Avoid using AI-generated text as your training data; it can reduce the quality of the resulting model.

How to train your My Voice model

Head to Sudowrite and open any document.

In the bottom of the left sidebar, click My Voice—or just click right here.

Click Train New My Voice.

Choose one:

Upload a .txt file, or

Paste your sample into the text box.

Click Start training. Training typically takes less than 2 minutes.

When training finishes, you’re returned to the My Voice model list, where your new voice appears.

By default, it’s named after your file (or the first part of pasted text).

To rename it, open the More (⋯) menu next to the model and choose Rename. This is also where you’ll delete a voice.

Use your voice to revise text

Think of My Voice primarily as an editing tool that helps you take a first draft to second, third, and beyond.

Open your project and highlight the text you want to transform

You can select short snippets or large sections (up to 4,000 words).

Select My Voice in the Selection Menu that appears to rewrite the highlighted prose in the selected style.

Alternatively, click the down arrow (⌄) next to My Voice and choose your preferred model (e.g., Regency style). Your most-recently selected model will persist.

A card will appear in the History section on the right with your rewritten prose.

Controlling the My Voice intensity

You may find that the first time you use My Voice to rewrite a passage, the style is either too subtle or too pronounced for your liking. Fortunately, you can easily change the intensity of a My Voice rewrite. To do so, simply tweak the Creativity slider in your Write Settings. (You can find the Write Settings beneath the down arrow next to the Write button in the top toolbar.)

Lower Creativity → lighter edits, less apparent style, closer to the source.

Higher Creativity → stronger more pronounced rewrites with more stylistic change.

Recommended workflow: Start near the middle, then adjust up or down based on your preference.

☝️

Tips & best practices

Match POV & tense: Train with samples in the same point of view and tense as your target work for most predictable results.

Selection length ↔ style strength:

On short selections, high Creativity may cause the model to expand more than you want. If it overwrites or “keeps going,” lower your Creativity.

On long selections (approaching 4,000 words), you may need to raise Creativity to see sufficient stylistic impact.

Iterate to intensify style: For extra impact, copy the rewritten prose from your My Voice card in the History section, paste it back into your draft, and run My Voice on it again for a second pass.

What else to know

My Voices are stored privately in your Sudowrite account, accessible only to you. They are not project-specific, however, which means you can use any My Voice you train in any of your projects.

Troubleshooting

“It didn’t change much.”

Increase Creativity or select a longer passage. Ensure your training sample is strong and representative.

“It changed too much / kept writing past my selection.”

Lower Creativity and select a slightly longer span for more context.

“Results feel off.”

Check that your training sample matches your project’s POV and tense, and avoid AI-generated training text.