EmptyAceStep1.5LatentAudio - ComfyUI Built-in Node Documentation - ComfyUI

The Empty Ace Step 1.5 Latent Audio node creates an empty (silent) audio latent tensor for audio generation workflows. It builds a latent with 64 channels whose temporal length is calculated from the requested duration, and tags it as audio data for use by downstream audio nodes.

Inputs

Parameter	Description	Data Type	Required	Range
`seconds`	The duration of the audio to generate, in seconds (default: 120.0). The latent length is calculated as `seconds * 48000 / 1920`, rounded to the nearest whole number.	FLOAT	Yes	1.0 - 1000.0 (step: 0.01)
`batch_size`	The number of latent images in the batch (default: 1).	INT	Yes	1 - 4096

Outputs

Output Name	Description	Data Type
`LATENT`	An empty latent tensor representing silent audio. The tensor has shape [batch_size, 64, length], where length is derived from `seconds`. The output also includes a type identifier of “audio” and a `downscale_ratio_temporal` value of 1764, which is used for temporal downscaling in audio processing.	LATENT

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): bb7120c91ce5d779147cb8553d6f96fa160d87468d4d87550fb6dd4ec89b1557