Aliyun Bailian | 阿里云百炼Install

Summary

Generate a lip-synced video by fusing a source video with a provided driving speech audio, optionally locking onto a specific face with a reference image and extending video length to match the audio.

Inputs

• Model (required) — Select the VideoRetalk lip-sync replacement model to use (only one option available).
• Source Video URL (required) — Public URL of the source video (mp4/avi/mov, max 300 MB, max 20 seconds) whose lip motion will be replaced.
• Source Audio URL (required) — Public URL of the driving speech audio (wav/mp3/aac, max 30 MB, max 20 seconds) to synchronize lip movements with.
• Reference Image URL — Optional public URL of a reference face image to lock lip-sync effect onto a specific person in the video.
• Extend Video Length — Boolean flag to extend the video duration if the audio is longer than the source video.
• Wait for Task Completion — Whether to poll the asynchronous task endpoint until the lip-sync video generation completes or fails. If false, returns task ID immediately.
• Poll Interval (MS) — Delay in milliseconds between polling attempts when waiting for task completion (default 2000ms).
• Max Wait (Seconds) — Maximum time in seconds to wait while polling for task completion before timing out (default 600 seconds).

Output shape

a single record containing the generated lip-sync video task result or task ID

If 'Wait for Task Completion' is true, output includes the completed task response; otherwise, output returns immediately with the task ID. Polling parameters control asynchronous task management.

Links

• n8n community nodes installation guide