This error appears when Descript’s automated safety system flags your uploaded photo. Most often, this is due to:

• Celebrity likeness or public figure detection
• Inappropriate content such as nudity

Try uploading a different image featuring a human subject, facing forward, with neutral lighting and no image alterations.

This error indicates the content in your script was flagged during generation. This is a separate audit from the image safety check that happens during upload.

Moderation is handled by our model provider. Descript does not control the specific criteria used to flag content. If you encounter this error, revise your script, remove any potentially graphic or offensive content, and attempt generating again. If the issue persists, contact support for further help.

Descript only supports JPEG, PNG, and WEBP image formats for avatar uploads. If your file is another type (like GIF or HEIC), it will be rejected. Convert your file to one of the supported formats and try again.

If your avatar doesn’t appear after assigning it to a speaker, it’s likely hidden behind other layers or not added visibly to the scene editor. We share steps for managing layer visibility in our Layer visibility and order guide.

Any time you make changes to your script, speaker, or assigned avatar, you’ll need to regenerate the avatar.

This includes:

• Editing the script or transcript
• Switching AI speakers
• Changing the underlying audio recording
• Replacing or deleting the avatar image

Once changes are made, click Generate avatar again to re-sync the animation.

If you have just assigned an avatar or made any script edits, click Generate avatar from the scene editor.

Descript avatar generation happens only after you click Generate avatar. Longer videos may take some time to process. You don’t need to stay in the project, closing the tab won’t cancel generation.

You’ll receive an email when your avatar is ready to review.

When exporting your Descript project to editing tools like Premiere Pro or Final Cut via timeline export (e.g., XML or AAF), avatar layers will not appear. This is expected behavior.

Avatars are not represented in the timeline. They must be baked into a video file before export.

To include avatars in your final project:

• Export your composition as a video file (e.g., MP4) from Descript
• Then import that video into your external editor for further work

For more, see Descript’s timeline export guide.

The AI credits required for avatars are calculated based on your total length of spoken audio and not just the parts where the avatar is visible.

Each time you generate, your available AI credits are updated. You can check your balance in the Usage tab of your account settings.

Currently, only image files up to 10MB are supported. Please reduce your file size and try uploading again.

Avatar generation is currently limited to 12 minutes of video at a time. If your video is longer than 12 minutes, you’ll need to break it into shorter sections.

In some cases, you might notice unexpected visual elements, like flickering lines or shapes near the bottom of the screen or around the hands of an avatar during playback.

• Add a gradient background.
• Include a captions layer.
• Crop the frame slightly to remove the affected portion.

Avatars created with the Kling models may include hand movements as part of the avatar generation. See Avatars Overview for more info about available models.