For now, Regenerate can only be applied to video or audio in English.

Impossible edits are now possible — and insanely easy — with the introduction of the AI-powered Regenerate feature. This tool allows users to effortlessly smooth out any jumpy, abrupt, or awkward sections in their audio recordings. In this guide, we'll walk you through how to use Regenerate effectively to enhance your podcast or other audio content.

Before you begin

To get started with Regenerate, make sure you have added video or audio to your Descript project. You can do this by importing files or directly recording in the Descript platform. For detailed instructions on importing and transcribing files, recording audio and video, or screen recording, check out our how-to guides:

• Importing and transcribing files 
• Recording audio
• Recording video 
• Screen recording

How to use Regenerate

1. Set a Speaker for the section of text you are looking to regenerate. 
2. Play through your video or audio and identify the section where you want to make a cut or improve the vocal performance. Edits in your script are represented by vertical dashed lines between two words.
3. Highlight the surrounding words or the specific section around the edit mark where you want to regenerate audio.
4. Select Replace >  Regenerate in the selection toolbar above your highlighted section.
   • If you try to regenerate after changing or correcting text in the range and don't have access to the speaker's AI voice, you'll be prompted to create a new AI Speaker.
     
5. The text will begin pulsating as Descript analyzes the audio and generates new audio content.
6. Once the regenerated section is processed, you'll have an AI audio clip, similar to an Overdub clip. You then have several options:
   • Restore the original audio.
   • Play and review the newly generated section.
   • Regenerate the section again—you'll get a slightly different result each time, so you can keep regenerating until you land on exactly what you want.
   • Keep the AI-generated clip, which will convert the clip into audio; then you can make further adjustments as necessary, like adding crossfades or adjusting word gaps.

Other use cases

Besides smoothing out abrupt edit points, Regenerate can be used in a variety of different situations:

• Eliminate sudden noises—if there's like a car honking or your dog or nephews start barking, there's no need to re-record. Simply use Regenerate to remove the annoying noise from that section (if the audio is bad or noisy across an entire audio track, try Studio Sound).
• Change the delivery of a voice performance—the best audio engineer in the world can't fix a moment where a speaker trails off mid-sentence. But Regenerate can. It injects a lively spark into lackluster performances, so you can turn something deathly boring into something truly engaging.

Feature constraints

There are a couple of instances where Regenerate will not be available for use:

• Regenerate is not available for sections that contain any of the following:
  • Multiple edit points
  • A scene boundary
  • Ignored text
• If you try to regenerate after changing or correcting text in the range and don't have access to the speaker's AI voice, you'll be prompted to create a new AI Speaker.
• Regenerate is not available for use on sections that already contain a Regenerate clip or an Overdub clip. If you want to make further changes to these sections, just Keep the AI-generated clip and then use Regenerate.