ComfyUI Node Runs on cloud
Audio Compressor
Turn loud passages down so a clip sits at an even level, or duck it under another clip wired into sidechain, such as music under a voice. Threshold and ratio drag on the curve drawn on the node.
Audio Compressor
- audio
- sidechain
- audio
- max_reduction
◄threshold_db-18.0►
◄ratio4.0►
◄knee_db6.0►
◄attack_ms10.00►
◄release_ms120►
◄detectorpeak►
◄lookahead_ms0.0►
◄makeup_db0.0►
◄auto_makeupfalse►
◄sidechain_cut_hz0►
◄mix1.00►
◄output_db0.0►
CategoryWAS Suite/Audio/Dynamics
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | The clip to compress. | |
| threshold_db | FLOAT | -18.0-60–0 | Level in dBFS above which the gain comes down. -18 = moderate; -30 = most of the clip; -6 = only the loudest peaks. |
| ratio | FLOAT | 4.01–30 | dB in per dB out above the threshold. `1` = off; `2` = gentle; `4` = firm; `10` and up = close to limiting. |
| knee_db | FLOAT | 6.00–24 | dB around the threshold the ratio eases in over. `0` = hard knee; `6` = smooth; `12` = very gradual. |
| attack_ms | FLOAT | 10.000.05–500 | Milliseconds the gain takes to come down. `1` = catches transients; `10` = keeps the punch; `50` = slow, lets drums through. |
| release_ms | FLOAT | 1205–5000 | Milliseconds the gain takes to recover. `50` = fast and lively; `120` = natural; `500` = smooth, for vocals and buses. |
| detector | COMBO | peak | `peak` = reacts to every sample, tighter control; `rms` = reacts to the average level over 10 ms, smoother and closer to what the ear hears. |
| lookahead_ms | FLOAT | 0.00–20 | Milliseconds the gain starts coming down before a peak arrives. `0` = off; `5` = catches sharp transients without clicks. |
| makeup_db | FLOAT | 0.0-24–24 | dB added after compression to bring the level back up. `0` = none. |
| auto_makeup | BOOLEAN | false | `true` = add the gain the threshold and ratio take off at full scale, halved, on top of makeup_db; `false` = makeup_db only. |
| sidechain_cut_hz | FLOAT | 00–1000 | Hertz below which the detector ignores the signal, so bass does not pump the gain. `0` = off; `100` = typical for a mix. |
| mix | FLOAT | 1.000–1 | `1` = fully compressed; `0.5` = parallel compression, punch with the dynamics kept; `0` = the input unchanged. |
| output_db | FLOAT | 0.0-24–24 | dB applied to the result. `0` = unchanged. |
| sidechainopt | AUDIO | Another clip whose level drives the gain, such as a voice ducking music. Empty = the clip drives itself. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | The compressed clip. |
| max_reduction | FLOAT | Most dB of gain reduction on the first clip. |