ComfyUI Node
HQ 2 · Regenerate Highs
Erase and regenerate: a super-resolution model generates new highs from the song without its top octaves; above the crossover they replace a share of the original's highs, following the original's level. Below the crossover the song stays as it is. Same rate and length.
HQ 2 · Regenerate Highs
- audio
- model
- brief
- audio
- report
◄preset▾►
◄hq_masteringtrue►
◄seed0►
◄model_steps4►
CategoryPlenio/Mastering
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | The song (every item of a batch). | |
| model | PLENIO_AUDIO_MODEL | Load Audio Model (super-resolution). | |
| preset | COMBO | auto: the profile that fits the song's genre (from the connected brief; standard without one). standard, gentle, strong and wide are the same four profiles in HQ 1, 2 and 3; custom shows the fields. | |
| hq_masteringopt | BOOLEAN | true | Off: skipped, the model is not even loaded. In the templates the HQ Mastering switch at the start sets it (off by default): on, fewer AI shimmer and artifacts, about twice the time. |
| briefopt | PLENIO_BRIEF | Optional: Song Brief or Cover Brief - its genre chooses the profile for 'auto'. | |
| seedopt | INT | 00–4294967295 | Seed of the model: the same seed gives the same highs. |
| model_stepsopt | INT | 42–8 | UniverSR's flow steps: 4 (default) about 1.5 s per second of music on an RTX 5060 Ti, 3 about 1.1 s, 2 about 0.75 s - fewer steps, other and quieter highs. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | The song with new highs (same rate and length). |
| report | PLENIO_REPORT | Profile, engine and its input edge, levels before/after. |