Parallel compression is an old trick: a heavily squashed copy of a track sits under the original, so the quiet detail comes up while the transients stay intact. Parallel clipping is the same idea with a different process on the second copy. Instead of a compressor turning the signal down and back up, a clipper cuts the tops off the peaks and turns what it removed into harmonics. Blend that under the clean signal and you get density and edge without losing the shape of the hit or the phrase.
It works best on two kinds of source: drums, where the transient is the whole point and you do not want to flatten it, and vocals, where you want the voice to hold its place in a busy mix without sounding pinned. This post walks through both with TubeClip, a tube clipper, and explains why the blend control matters more than it looks.
Why the MIX has to be time-aligned
A parallel blend only works if the two copies line up to the sample. If the processed copy arrives even a fraction of a millisecond late, the sum is no longer "original plus a bit of grit". It is two copies of the same signal slightly offset, and that offset cancels some frequencies and reinforces others. The result is comb filtering: a hollow, phasey tone that changes as you move the blend, and it is easy to mistake for the clipper sounding bad.
A good clipper has latency. TubeClip oversamples so the harmonics it creates have room above the audio band before they are filtered away, and it holds its ceiling with a 0.7 ms look-ahead guard. At the default 8x oversampling the latency is 3.3 ms (0.7 ms at 1x), reported to the host. The host compensates for that across the whole plugin, but it cannot know that the dry signal inside the plugin needs delaying too.
That is why MIX matters. It blends the clipped signal against the dry one inside the plugin, and the dry side is delayed to match, so the two are time-aligned. Bypass is delayed by the same amount, so switching TubeClip in or out moves nothing on the timeline either. You can set up parallel clipping on the insert itself, with no send, no extra bus and no manual delay to dial in. ## Parallel clipping on drums
The store's starting point for drums is PENTODE on a drum bus or a snare, with GLOW up a little. PENTODE is the crisper of the three tube curves under the TUBE switch: odd harmonics and punch. The transients are clipped into harmonics instead of ducked, and the kit hits harder at the same peak level.
Run that as a full insert first, then bring MIX down and listen to what changes. With the whole signal clipped, the very top of every hit is shaved. With MIX below fully wet, part of the original peak comes back through the dry path, and the clipped copy adds its harmonics underneath. The attack keeps more of its spike while the body of the hit gets thicker.
GLOW is the second half of this. It lifts the body of the waveform by up to 9 dB while the peaks stay exactly where they are, with unity at the ceiling. It is a curve, not a compressor, so it has no attack or release and nothing pumps: the room and the decay of each hit come up without the kick dragging the cymbals down.
DRIVE sets how hard the programme hits the tube, from -12 to +24 dB, and the CLIP meter reads dB over the ceiling. The meter also keeps the clip clock, the heaviest clip so far and when in the song it happened, so if one fill is doing far more than the rest of the track you will see where. Click the meter to clear it.
If the kick and toms need weight rather than edge, FAT is the control. It is a low shelf before the tube and its inverse after: the lows are driven harder into the curve and the balance is restored afterwards. The low end gains harmonics and weight without turning to mud. FAT FREQ, behind SETTINGS, sets where that shelf sits.
Parallel clipping on vocals
For vocals the store's recipe is MIX at half, GLOW up and DRIVE high. The clipped copy sits under the clean one for density and warmth, time-aligned, with the peaks unchanged.
The reasoning is the reverse of the drum case. On a drum you clip mainly to control the transient. On a vocal you clip hard on the wet side precisely because you will only hear half of it. High DRIVE turns the consonants and the loud syllables into a dense band of harmonics, and with the dry voice on top the listener hears a vocal that is fuller and more present, not a distorted one.
TRIODE is worth trying here. It is a warm knee with a touch of asymmetry, giving a long tail of harmonics, where PENTODE is crisper and CLEAN is a polynomial knee with a few odd harmonics that decay fast. The choice changes the character of the blended layer more than its level.
Behind SETTINGS, SILK is a high shelf that rounds the top the harder you clip, and SILK FREQ sets its corner. On a bright vocal driven hard that takes the edge off sibilant harmonics before they reach the blend. BIAS adds even harmonics, and KNEE runs from hard to soft.
Parallel processing makes things louder, and louder nearly always sounds better for a few seconds. TubeClip has two listening aids for this. MATCH holds the output at the input's loudness, so as you move MIX and DRIVE you are comparing tone, not level. DELTA plays only what the clipper takes away, which shows exactly what the wet side is doing. Neither is ever applied to a render. OUTPUT trims 24 dB either way after everything, once you have decided.
To hear it for yourself, start a 14-day free trial of the full version and put TubeClip on a snare with PENTODE, GLOW up a little and MATCH on. Bring DRIVE up until the CLIP meter moves on every hit, then pull MIX down slowly and listen for the point where the attack comes back but the body stays thick. Then move it to a lead vocal: TRIODE, MIX at half, GLOW up, DRIVE high, and switch DELTA on for a moment to hear the layer you are adding.