Clip Gain Riding: Fix Vocal Levels Before You Compress
Vocal mixing · 4 min read
Clip gain is the volume of the audio itself, before any plugin sees it. Ride it phrase by phrase until the loudest and quietest words sit inside a 3 dB window, with the whole chain bypassed. Do that first and your compressor goes from 10 dB of gain reduction to 3, which is the difference between a controlled vocal and a squashed one.
Target window
+/-3 dB
Across the whole take · peaks near -6 dBFS
The 60-second answer
Bypass every plugin. Find the loudest word in the song and set its peak around -6 dBFS. Split at breaths and ride each phrase until a VU meter on the vocal stays between -3 and +1 VU for the whole verse. Then unbypass and set the compressor for 3 dB.
Why the compressor cannot do this job
A compressor has one threshold and one set of time constants for the whole song. Your singer does not. Verse one is 8 dB quieter than the last chorus, one line is mumbled, one word is shouted. To catch the shouted word the threshold has to sit low, and once it does the mumbled line gets compressed too and the take flattens. Clip gain has no threshold and no time constant — it is a number per clip.
Clip gain, fader and automation are three different tools
Clip gain lives before the inserts and fixes recording problems: mic distance, a swallowed word, a shout. Fader automation lives after the inserts and fixes musical balance against the beat. The fader itself is your one static level for the song. Use them in that order and each has a clear job. Use the fader to fix a recording problem and you drag the reverb send, the bus compressor and every automation move along with it.
How to ride clip gain, step by step
| # | Do this |
|---|---|
| 01 |
Bypass the chain, keep the tuner You want the raw recorded level. Tuning stays on because it changes the waveform. Everything else comes off. Inserts: bypassed |
| 02 |
Set your loudest word to -6 dBFS Scan the whole song, find the single loudest word, and use it as the anchor everything else moves towards. Peak anchor: -6 dBFS |
| 03 |
Split at breaths, never inside a word Cut on the silence or on the breath. A level change either side of a cut inside a vowel is an audible step even with a crossfade. Crossfade 10–20 ms |
| 04 |
Ride whole phrases first Get every phrase inside 3 dB of the anchor. That is 80% of the result and takes about five minutes on a three-minute song. Moves of 1–4 dB |
| 05 |
Then fix individual words Only the outliers: the shouted word, the swallowed one, the consonant that spikes. Usually eight to fifteen per song. Moves of 1–3 dB |
| 06 |
Check with a VU meter, then reset the compressor Insert a VU meter calibrated so -18 dBFS reads 0 VU and keep the needle in a 4 VU band. Then unbypass: threshold now lands much higher. -18 dBFS = 0 VU, GR 3 dB |
The trick: ride to a VU needle, not to the picture of the waveform
The waveform on your screen shows peak level, and peak level has almost nothing to do with how loud a word sounds. An S peaks high and sounds quiet. A chesty open vowel peaks lower and sounds enormous. So if you flatten the waveform until it looks like a brick, you have made the sibilants match the vowels, which is the opposite of what you want. Instead insert a VU meter on the vocal, calibrated so -18 dBFS reads 0 VU. A VU meter has roughly a 300 ms integration time, close to how the ear averages a syllable, so the needle tracks perceived loudness rather than peaks. Bypass the compressor, play the verse and watch: the target is a needle living between -3 and +1 VU that never swings more than about 4 VU. Get there, then close your eyes and play it again. It will already sound mixed.
Stop guessing the chain.
The Vocal Chain Bible is 88 real vocal chains from Grammy-winning engineers — every plugin, every setting, in order, with free alternatives for each one. 5+ years, 100+ interviews, one PDF.
Get the Vocal Chain Bible — $48 →Mistakes that kill it
- Boosting quiet clips by 10 dB. Clip gain lifts the room and the mic hiss with the word. Above about +6 dB, re-record the line.
- Riding clip gain after the compressor is dialled in. Every move changes what the threshold means. Ride first, compress second.
- Cutting mid-vowel. A 2 dB step inside a sustained note is audible even with a crossfade. Cut in the gaps.
- Flattening the performance. A 3 dB window is the goal, not zero variation. If the shout and the whisper match, you deleted the arrangement.
This is the least exciting hour in mixing and the one that separates records from demos. Everything downstream behaves predictably once the input stops moving.
Keep going
FAQ
Where is clip gain in my DAW?
Pro Tools calls it Clip Gain. Logic has region gain in the Region inspector. Studio One uses Event Volume. In Ableton Live it is the Gain slider in the Sample box. FL Studio has a gain control in the audio clip settings.
Is clip gain the same as normalising?
No. Normalising sets one gain for the whole file from its loudest point. Clip gain riding sets a different gain for every phrase, which is the part that helps.
Should I ride clip gain or automate the fader?
Both, in that order. Clip gain fixes the recording before the plugins, fader automation fixes the balance after them.
Who's behind this
Ivan Meshcheriakov · IvanFromRnD
5+ years ago I moved to LA with zero connections and started building from nothing — 14-hour days, meeting people and interning at studios. It took me 5+ years to meet and interview 100+ Grammy-winning engineers — then I turned it all into the Vocal Chain Bible.
Read my full story →