Skip to main content
Audio Convolve (IR)

Audio Convolve (IR) · screenshot coming soon

​
What this node does

Audio Convolve (IR) convolves an audio track with a second audio clip used as an impulse response (IR) — a recording of how a space or device responds to a short impulse. The result carries the reverb tail, coloration, or filtering baked into that IR. This is the classic technique behind convolution reverbs and cabinet/speaker simulation. It takes a COMFYTV_AUDIO snapshot on audio (or a COMFYTV_VIDEO, whose audio track is used) and a second COMFYTV_AUDIO on ir, and outputs processed COMFYTV_AUDIO. It has a ▶ Run. An impulse response on the ir input is required — the node errors if it’s empty. To feed native ComfyUI AUDIO in or out, insert a Bridge — see bridges.md.

​
When to use it

  • Place a dry vocal or instrument inside a real hall, church, or room captured as an IR.
  • Simulate a guitar cabinet or hardware device from its impulse response.
  • Extract an IR with Audio Deconvolve first, then apply it to any source here.

​
Parameters

​
wet

Level of the convolved (processed) signal, 0.0–2.0 (default 1.0). Values above 1 amplify the wet path.

​
dry

Level of the original (unprocessed) signal mixed back in, 0.0–2.0 (default 0.0). Raise for a mix of dry + reverberant sound; leave at 0 for fully wet.

​
normalize

Boolean, default on. Normalizes the impulse response so a loud IR doesn’t blow out the output level.

​
Outputs

​
Tips

  • The ir input is mandatory — wire an impulse-response clip (a captured reverb or a synthetic IR) into it.
  • For a natural reverb blend, keep dry near the source level and add wet to taste rather than going fully wet.
  • Turn normalize off only if you want the IR’s raw gain to pass through.
  • Audio Deconvolve (IR) — recovers an impulse response from a swept-sine recording.
  • Audio Sweep (ESS) — generates the sine sweep used to capture an IR.
  • Muse Reverb (FDN) — an algorithmic reverb alternative that needs no IR.