Actions12
Video → Generate From Audio
AI-generatedSummary
Generates a video conditioned on an input audio file and a text prompt using the LTX-2.3 22B model.
Inputs
- prompt (required) — A text description to guide the video generation alongside the audio.
- audioBinaryProperty (required) — The name of the binary field containing the audio file (MP3, WAV, OGG, FLAC). Max 20 MB.
- ratio (required) — Aspect ratio of the generated video (square, landscape, portrait).
- options.firstFrame — The name of the binary field containing the first frame image.
- options.lastFrame — The name of the binary field containing the last frame image.
- options.frames — Number of video frames to generate (49-241).
- options.squareSize — Resolution in pixels for square ratio (512x512, 768x768, 1024x1024).
- options.landscapeSize — Resolution in pixels for landscape ratio (1024x576).
- options.portraitSize — Resolution in pixels for portrait ratio (512x640, 768x960).
- options.seed — Random seed for generation (default -1 is random).
- options.waitTimeout — Maximum time to wait for completion in seconds (30-600).
Output shape
No body — success/failure only
The node uses a wait execution pattern. The generated video is handled via a webhook callback (resumeUrl), not returned directly by this operation.
Examples
Example 1: Generate a 7-second talking-head video in landscape ratio from a speech audio file.
Set prompt to 'A person speaking confidently in a modern office', audioBinaryProperty to 'data', ratio to 'landscape', and frames to 120.
Example 2: Generate a 30-second video with a specific resolution using a seed for reproducibility.
Set options.squareSize to '1024x1024', frames to 241, and options.seed to a specific number.