Media Formats Support

Direct file uploads to the Sync API are limited to 20 MB. This applies when you use the create-with-files endpoint to upload video or audio files directly in the request body. If your file exceeds 20 MB, host it at a publicly accessible URL (such as an S3 bucket, CDN, or any web server) and pass the URL in the url field of your video or audio input instead. There is no file size limit when using URL-based inputs — Sync downloads the file from your URL during processing. For large batch workloads, the URL-based approach is recommended regardless of file size, as it avoids upload timeouts and is more reliable for production pipelines. Make sure your hosted files are publicly accessible without authentication headers, as Sync’s servers need to fetch the file directly.

The maximum video duration depends on your subscription plan. Free accounts can process videos up to 20 seconds long. Paid plans support significantly longer videos — the maximum duration increases with each tier, ranging from 1 minute on Hobbyist up to 30 minutes on Scale+ plans. Check the pricing page for your plan’s specific duration limit. Note that react-1 has a separate hard limit of 15 seconds regardless of your plan, as it is designed for short-form expressive content. For videos that exceed your plan’s maximum duration, consider splitting them into shorter segments using the Segments API and processing each segment separately. Longer videos also take proportionally longer to process — see our Generation Times guide for expected processing times by model and duration.

Yes, Sync processes videos in any aspect ratio including vertical (9:16), horizontal (16:9), square (1:1), and any custom aspect ratio. The lip sync pipeline extracts the face region from the video at 512x512 resolution for processing regardless of the input video’s dimensions or orientation. The processed face is then composited back into the original frame, preserving the original aspect ratio and resolution in the output. This means vertical videos from mobile phones, square videos for social media, and standard widescreen footage all work equally well. The output video maintains the same dimensions as your input. For the best face detection results, ensure the speaker’s face occupies a reasonable portion of the frame — in vertical videos where the face may be smaller relative to the full frame, make sure the face is still clearly visible and well-lit.