Skip to main content
Mirelo SFX 2.0 generates sound effects. It works in two directions: it watches a clip and produces Foley timed to what happens on screen, or it takes a text description of a sound and builds it from nothing. In ComfyUI the model runs through two partner nodes. Mirelo Video to SFX takes a clip and returns the mixed soundtrack plus the dialogue on its own output. Mirelo Text to SFX takes a prompt and a duration and returns one audio clip. Both are cloud API nodes: they need a Comfy account with credits and download no model files locally. Output is 48 kHz audio. A video run covers 1 to 600 seconds, and text to SFX covers 0.1 to 600 seconds. Video to SFX bills by the duration of the clip you send, and Text to SFX bills by the duration you set.
To use the Partner Nodes, you need to ensure that you are logged in properly and using a permitted network environment. Please refer to the Partner Nodes Overview section of the documentation to understand the specific requirements for using the Partner Nodes.
Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.If nodes are missing when loading a workflow, possible reasons:
  1. You are not using the latest ComfyUI version (Nightly version)
  2. Some nodes failed to import at startup

​
Available workflows

​
Video to SFX

Add sound to a finished cut. The model reads the picture and times each sound to the action in frame, so footsteps, impacts, and ambience land on the movement instead of on a guessed timeline. Mirelo SFX 2.0: Video to SFX workflow preview

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “Mirelo SFX 2.0: Video to SFX” in Template Library
Input material

mage_crash_discovery_video.mp4

Load into the LoadVideo node
The node has two audio outputs. audio carries the finished mix, and speech carries the dialogue separated from the generated effects. When the clip has no speech, the speech output is silent.

​
Text to SFX

Build a sound from a description. The prompt and the duration are the only inputs, so this is the mode for a sound you cannot record or for filling an effects library. Mirelo SFX 2.0: Text to SFX workflow preview

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “Mirelo SFX 2.0: Text to SFX” in Template Library
Example prompts
  • Heavy rain on a tin roof with distant thunder
  • Soft footsteps on wood floor in a small quiet room
  • Sharp knife chop on a wooden cutting board, close mic
  • Metal door slam in a concrete hallway with short echo

​
Node controls

​
Prompting tips

  • Name the sound, not the story. Source, material, space, and intensity in one line: boots on gravel, metal settling, soft cavern wind. Concrete nouns you can hear beat abstract mood words.
  • Start Video to SFX with an empty prompt. The picture already carries the timing. Add one line only for off-screen sound, such as distant church bells, a dog barking two streets away, or to push a single layer.
  • Use the negative prompt to strip layers. music, wind, crowd removes a score from an AI-generated clip. Keep speech out of the negative list when you want the dialogue kept.
  • Match the duration to the event. A single impact does not need 30 seconds of rain.
  • Audition a few variants. Re-run with a different seed and change one dimension at a time: material first, then space, then intensity.

​
Get started

  1. Update ComfyUI to the latest version
  2. Go to Template Library, search for Mirelo