Sign in
videoGeneration

Generate video from audio

Generate a video driven by an audio track. Supply dialogue, music, or ambient sound and the model produces visuals synchronized to the audio. Optionally condition on a reference image for visual direction, a last frame for interpolation, and camera motion. Output frame rate defaults to 24 fps; its length is set by the audio.

Returns the generated video directly in the response. For an asynchronous version that returns a job to poll, see v2/audio-to-video.

Billed per second of input audio. See Pricing.

post/v1/audio-to-video

Headers

Authorizationstring required

API key authentication

Request body

audio_uristring required

Audio file to be used as the soundtrack for the video. The audio sets the length of the generated video, so its duration must not exceed the maximum the selected model supports at the chosen resolution.

ModelResolutionMaximum audio duration
ltx-2-5-fast720p, 1080p20 seconds
ltx-2-5-fast1440p, 4K10 seconds
ltx-2-5-pro720p, 1080p10 seconds
ltx-2-3-pro720p, 1080p20 seconds
ltx-2-3-pro1440p, 4K10 seconds

See Input Formats for supported formats and size limits.

image_uristring

Input image to be used as the first frame of the video. Required if prompt is not provided. See Input Formats for supported formats and size limits.

promptstring

Text description of how the video should be generated. Required if image_uri is not provided. Can be empty string when image_uri is provided. If image_uri is provided, this describes how the image should be animated. If no image_uri is provided, this describes the video content.

resolutionstring

The resolution of the generated video in WIDTHxHEIGHT format. When omitted, the resolution is automatically determined based on the input image orientation — portrait images produce 1080x1920 video, landscape images produce 1920x1080 video. If no image is provided, defaults to 1920x1080.

Supported resolutions depend on the model — see Supported Models for the full matrix. ltx-2-5-fast and ltx-2-3-pro support 720p through 4K; ltx-2-5-pro supports 720p and 1080p only.

model'ltx-2-3-pro' | 'ltx-2-5-fast' | 'ltx-2-5-pro'

Model to use for video generation.

fpsinteger

Frame rate in frames per second. See Supported Models for available FPS per model and resolution.

last_frame_uristring

Image to be used as the last frame of the video. Requires image_uri. When provided, the video interpolates between the first and last frame. See Input Formats for supported formats and size limits.

camera_motion'dolly_in' | 'dolly_out' | 'dolly_left' | 'dolly_right' | 'jib_up' | 'jib_down' | 'static' | 'focus_shift'

Apply camera motion effects to the generated video.

Response

Video generated successfully

Changes