Descript Makes Eleven v4 Its Default AI Speech Model.

Eleven v4 is now Descript’s default AI speech model. It turns typed edits into speech in the user’s voice while retaining their accent and pacing.

In its LinkedIn announcement, Descript used a familiar recording problem to introduce the model: thinking of a better introduction after recording has finished. The replacement can be typed into Descript, which reads it in the speaker’s voice.

Descript’s generated voice now runs on Eleven v4 by default. The model carries the original speaker’s accent and pacing into the generated speech, rather than only reproducing the words from the typed script.

The feature connects a text edit to a spoken result. A creator can write a new introduction after the original recording, then have Descript produce that line in their voice. The generated delivery includes the vocal characteristics named in the announcement: accent and pacing.

Making Eleven v4 the default also means the updated model sits behind Descript’s AI speech generation rather than being presented as a separate model for this use case.

The workflow starts after the recording session. When a line needs changing, the revised words can be typed into the project. Descript then reads those words in the user’s voice using Eleven v4.

This applies to the specific example Descript showed: replacing an introduction that no longer works. The same text-to-speech behaviour produces spoken audio from the typed replacement while carrying across the speaker’s accent and rhythm.

The speech update adds another text-led editing option alongside Descript Underlord’s natural-language video editing.

What is Descript Eleven v4?

Eleven v4 is Descript’s new default AI speech model. It reads typed text in the user’s voice and carries their accent and pacing into the generated speech.

Does Eleven v4 preserve a speaker’s accent?

Yes. Descript says Eleven v4 brings the user’s accent and pacing into speech generated from typed text.

Which Descript plan includes Eleven v4?

Plan and pricing details were not included in the announcement. Eleven v4 is described as the default AI speech model.