Muse Reverb (FDN) · screenshot coming soon
What this node does
Muse Reverb (FDN) adds artificial reverberation using a Feedback Delay Network — a bank of delay lines with feedback that produces a dense, natural-sounding tail without needing an impulse response. You control reverb time, room scale, pre-delay, the balance of dry/early-reflection/late energy, frequency-dependent decay, modulation, and stereo width. It takes aCOMFYTV_AUDIO snapshot (or a COMFYTV_VIDEO, whose audio track is used) and outputs processed COMFYTV_AUDIO. It has a ▶ Run. A source is required.
To feed native ComfyUI AUDIO in or out, insert a Bridge — see bridges.md.
When to use it
- Add hall, room, or chamber ambience to dry recordings without hunting for an IR.
- Sit a vocal or instrument in a defined space with controllable size and decay.
- Sculpt the reverb tone by making highs decay faster than lows (or vice versa).
Parameters
reverb_time_ms
Overall decay time (RT) in ms, 100–10000 (default 2200). Longer = bigger, more sustained space.room_scale
Size scaling of the delay network, 0.5–4.0 (default 0.8). Larger values spread the reflections out for a bigger room feel.predelay_ms
Gap before the reverb begins, in ms, 0–500 (default 10). Adds a sense of distance from the walls and keeps the dry signal clear.Levels
- dry_db — dry (unprocessed) level, -60 to +6 dB (default 0).
- late_db — late reverberation level, -60 to +6 dB (default -14).
- er_db — early-reflections level, -60 to +6 dB (default -14).
Frequency-dependent decay
- time_low — low-band decay scaling %, 50–200 (default 100).
- time_high — high-band decay scaling %, 50–200 (default 60). Lower = brighter highs die faster (more natural).
- xover_low_mid — low/mid crossover Hz, 50–500 (default 400).
- xover_mid_high — mid/high crossover Hz, 1000–10000 (default 4000).
- feedback_top — top-end damping frequency Hz, 500–20000 (default 8000).
Modulation & width
- mod_freq — tail modulation rate Hz, 0–10 (default 1.0).
- mod_amp — modulation amount, 0.0–1.0 (default 0.2). Adds subtle movement to avoid a static, metallic tail.
- stereo_spread — stereo width %, 0–150 (default 100).
quality
Network size:8 or 16 delay lines (default 16). 16 is denser and smoother; 8 is faster.
Outputs
Tips
- Keep the highs decaying faster than the lows (time_high below time_low) for a natural room; equal values sound artificial.
- A little mod_amp (~0.2) stops long tails from ringing metallically.
- If the reverb muddies the mix, raise predelay_ms or lower late_db to keep the dry signal forward.
Related nodes
- Audio Convolve (IR) — reverb from a real recorded space instead of an algorithm.
- Audio Echo — discrete echoes rather than a dense tail.
- Audio EQ — carve space for the reverb, or de-mud it after the fact.