Under the hood – the power of AI-processing in pure:deess



Featuring some of the analysis and processing technologies from sonible’s smart:deess plugin, pure:deess listens to the input signal (a speaker or vocalist) and identifies different sibilants (s, sh and more). From here, a target spectrum is computed for the particular voice, and individual sibilant sounds are processed towards that target. This means that the plugin can also boost to revive dull sibilants and increase intelligibility.

Suppression: The user controls how much processing is applied using the large central dial. When the Suppression level is low, only the louder sibilants are reduced; turn it up further to tame even the minor details in the recorded audio.

Color: This parameter changes the target character the plug-in aims for. With Color set to Soft, sibilant elements are softened a lot, making this a good setting where intelligibility can be sacrificed (for example backing vocals). In Neutral, the characteristics of the original sibilants can be preserved, with only harsh resonances corrected. Sharp focuses on retaining presence and intelligibility while reducing the unnaturally harshest material.

Shaping: The Shaping parameter in pure:deess adjusts the intensity of spectral processing, shaping sibilants toward the selected Color. Higher values apply stronger shaping for a more consistent sound and lower values maintain natural variations with minimal processing. This ensures precise de-essing while preserving the intended vocal character.