Sibilus
A de-esser that tells an 's' from a bright sound, with a detector you can see and check.

Sibilus turns sibilance down across the whole signal or only in its band. In Single vocal mode a classifier combines two measured features, the band's share of the energy and its spectral flatness, so bright vowels and tones are left alone. A key from any strip, lookahead and mid/side are built in.
Hear it, with and without
An excerpt of the demo song Midnight Static
Hear its presets
- Without the effecta spoken voice
- Male vocalVocal — Pop, Rock
- Female vocalVocal — Pop, R&B
- Rap vocalVocal — Trap · Hip-hop
Rendered by Cantus from its factory presets, each matched to the dry take's loudness by a gain alone, with no processing after.
Voice: Britton Rea, public domain (Wikimedia Commons)
What stands out
- Single vocal and Allround detection; the classifier's confidence drawn per column
- Wide band, Split high or Split band (linear phase, leaves the air above alone)
- Band 1–20 kHz dragged on a spectrum; threshold dragged on the detector lane
- Learn threshold from a passage, in one undo step
- Lookahead 0–15 ms reported for compensation; L/R, M/S, mid only or side only
- Listen to Trigger or Removed; events, average and deepest cut per event