Loading REGENT product view.
Freeware / Master Clipper
REGENT is a free clipper plugin for controlled peak shaping, parallel clipping and final-stage processing.
01 / Controls
Drive, curve shape, processed level and blend remain separate throughout the signal path.
INPUT DRIVE
The control adds up to 10 dB before REGENT’s nonlinear stage. The selected transfer curve stays in place and keeps the same shape.
As the input level rises, more of the signal reaches the curved or clamped region. This produces additional peak reduction and harmonic content.
0 to +10 dB
POST-CLIP OUTPUT
CEILING runs after the nonlinear stage. It scales the wet signal without changing the transfer curve, its knee or the amount of input drive.
Use it to place the processed signal before it reaches the dry/wet blend.
-24 to 0 dB
CURVE EXPRESSION
Lower values move either curve toward a harder reference response. Higher values use more of the selected curve’s full knee.
CHARACTER does not add EQ, gain or another saturation stage. The control has no effect when Hard is selected.
ALIGNED DRY/WET BLEND
REGENT delays the dry branch by eight samples to match the processed branch before combining them.
At 0% the output is the aligned dry path. At 100% it is the complete processed path. Positions between those endpoints use a continuous linear blend.
Reported plugin latency: 8 samples
TRANSFER SHAPE AND RATE
Soft begins bending earlier and uses REGENT’s broadest knee. Its canonical transition is continuous through the third derivative (C3).
Solid stays linear for longer, then moves through a narrower knee. Its canonical transition is continuous in value, slope and curvature (C2).
Hard has no knee. It applies an abrupt normalized clamp, and CHARACTER is not used in this mode.
At 1x, the nonlinear stage runs at the host sample rate. The 2x, 4x and 8x settings run it at a higher internal rate before reconstructing the signal at the host rate.
ADAA remains active at every rate. The four processing paths are prepared before realtime processing begins.
DISPLAY MODES
The IN and OUT rails show stereo sample peaks. S. LUFS reports K-weighted short-term loudness over a rolling three-second window.
The history view uses the stereo signal leaving REGENT. Local minimum and maximum values are preserved when the display is reduced, which keeps short transients visible.
History lengths of 0.5, 1, 2, 4 and 8 seconds are available. LIVE updates continuously; PAUSE holds the current view.
The transfer view shows the direct relationship set by CLIP, STYLE, CHARACTER and CEILING. Its operating point follows the current input peak.
This view covers the memoryless transfer model. ADAA history, oversampling reconstruction, MIX and bypass are outside that graph.
02 / Audio Preview
Switch between Before and After without losing your place in the track.
Loading DEMO I before source.
03 / Free Download
Download REGENT Free Clipper Plugin
04 / Technical Notes
How the signal path is implemented
01
REGENT keeps input drive, transfer shape, processed level and blend as separate stages. CLIP applies gain before the nonlinear stage. STYLE selects the transfer curve, and CHARACTER adjusts the Soft or Solid response. CEILING trims the processed path after clipping. MIX then combines it with the aligned dry path.
This structure lets you increase drive without moving the selected curve or tying output level to input gain.
02
Soft uses REGENT’s broadest canonical knee, with continuity through the third derivative (C3). Solid uses a narrower knee that remains continuous in value, slope and curvature (C2). Hard applies an abrupt normalized clamp with no knee.
CHARACTER moves the Soft and Solid responses toward or away from a harder reference. It is not used when Hard is selected.
03
Clipping creates harmonics, and digital nonlinear processing can fold part of that spectrum back into the audible range. REGENT evaluates each active curve with first-order Antiderivative Antialiasing based on the mathematical primitive of that curve.
ADAA remains active at 1x, 2x, 4x and 8x. At the higher settings, the nonlinear stage runs at an increased internal sample rate before the result is reconstructed at the host rate. These methods reduce aliasing but do not guarantee that every alias component is removed.
04
REGENT reports a fixed latency of eight samples. Before MIX combines the two branches, the dry signal is delayed to the same timing reference as the processed signal.
MIX uses a continuous linear blend. At 0% the output is the aligned dry path, and at 100% it is the complete processed path.
05
The IN and OUT rails are stereo sample-peak meters. S. LUFS is a K-weighted short-term loudness reading calculated over a rolling three-second window.
The waveform view uses the post-output stereo signal and preserves local minimum and maximum values when reducing the history for display. The transfer view shows the direct CLIP, STYLE, CHARACTER and CEILING relationship together with the current operating point. It does not model ADAA history, oversampling reconstruction, MIX or bypass.
06
Oversampling paths and working buffers are prepared before audio processing begins. During normal processing, REGENT performs no explicit heap allocation. Parameter state is read through atomics, while meter and waveform data are sent to the interface without locking the audio thread.
CLIP, CEILING, CHARACTER, MIX and bypass transitions are smoothed in their processing paths. If a host requests audio before preparation is complete, REGENT passes that audio through unchanged.