VideoRetalk synchronizes lip movements with audio, generating a new video from an input video and an audio file.
This document applies only to the China (Beijing) region. To use the models, you must obtain an API key from the China (Beijing) region.
Model overview
Example
Input example | Output example |
Character video: Voice audio: |
Billing and rate limits
Model | Unit price | Free quota(View) | RPS limit for task submission | Number of concurrent tasks |
videoretalk | CNY 0.08/second (pay-as-you-go, by generated video duration) | 1,800 seconds | 1 | 1 (one task runs at a time; others are queued) |
Calling the model
VideoRetalk requires API calls (pay-as-you-go) — it cannot be tested in the Model Studio console.
Call VideoRetalk with a clear, front-facing character video and an audio file to generate lip-synced output. For more information, see VideoRetalk video generation.