Vocal Clip Gain Planner
Vocal Clip Gain Planner: Even Out a Take Before the Compressor
Drop a vocal take. Get a phrase-by-phrase clip-gain plan so your compressor stops working overtime. The planner finds every phrase, measures its level, tells you how many dB to move each one, lets you hear the plan applied, and copies it as a list you can follow in any DAW. Everything runs locally with the Web Audio API — the file never leaves your device.
Dry lead vocal, one singer · first 4 minutes analyzed · up to 60 MB
Nothing runs until you choose a file or tap the demo.
Nothing from your file leaves the browser. Decoding and analysis use the Web Audio API on your device; there is no server, no microphone access and no tracking. The only network request this page makes is fetching the demo take when you tap it.
| # | Time | Length | Level (RMS) | Peak | Move | After |
|---|
Phrases are split on silence, not on lyrics, so one musical line can read as two phrases and a loud breath can count as its own phrase. A take with constant background noise or a long reverb tail may be read as one long phrase — gate or trim it first. The numbers are RMS, not perceived loudness. This is a plan: your ears decide the last 0.5 dB, in the mix, with the beat playing.
- Vocal clip gain riding: the step before the compressor
- Vocal volume automation: fader rides after the chain
- Vocal compressor attack and release, explained with numbers
- Serial compression on vocals: two gentle stages beat one hard one
- Compressor Settings Calculator · Gain Staging Calculator
Once the phrases sit level, the compressor only has to work on syllables — and that is where the real chains differ. The Vocal Chain Bible's 88 real vocal chains from hit records show which compressor each engineer used after the gain stage, with attack, release and ratio, plus a free-plugin version of every chain.
See the 88 vocal chains →Try one free chain firstHow to apply the plan in your DAW
| DAW | Where clip gain lives | Fastest workflow |
|---|---|---|
| Logic Pro | Region Gain in the Region inspector, or the Gain tool from the tool menu. | Split the region at each phrase boundary (Cmd-T at the playhead), select a piece, type the dB value from the plan into Gain. |
| Pro Tools | The Clip Gain line on every audio clip (show it with Clip Gain Line in the View menu). | Separate the clip at phrase edges, then nudge each piece's gain with the Clip Gain fader or type the value; breakpoints give you ramps for free. |
| Ableton Live | Clip Gain slider in Clip View; or a Utility device with Gain automation if you prefer a single clip. | Split at phrase edges (Cmd/Ctrl-E), set each clip's gain; consolidate afterwards if you want one clip again. |
| FL Studio | The volume knob at the top-left of each audio clip in the Playlist (also in the clip's channel settings). | Slice the clip at phrase edges with the slice tool, then set each slice's knob; the plan's dB values map to the knob's hint readout. |
| Any DAW | Pre-fader volume automation on the track also works, as long as it sits before the compressor. | Draw a flat lane per phrase at the planned value with 10 ms ramps at the edges — the same shape this tool uses when it plays the plan. |
Clip gain is pre-everything: it changes what the plugins receive. Fader moves are post-chain and do not change how hard the compressor works.
More free vocal tools
Clip gain vs compression: why the level comes first
Clip gain vs compression is not an either-or. They fix different problems. Clip gain is a static move: this phrase, this many dB, done before any plugin sees the audio. A compressor is a reactive move: it listens to the signal and turns it down when it is loud, within milliseconds, and turns it back up as it fades. Use the reactive tool on a problem that is actually static — a chorus sung 6 dB louder than the verse — and it has to hold 6 dB of gain reduction through the whole chorus. That is 6 dB of pumping risk, 6 dB of tone change, and a release that is now fighting words instead of shaping them. Move the chorus down 6 dB first and the same compressor only has to catch the syllables that jump, which is the job it is good at.
The planner above does the measuring part of that workflow. It cuts the take into phrases wherever the singer stops for a quarter of a second, measures each phrase's RMS level and peak, picks a reference phrase and writes down how far every other phrase is from it, rounded to half a dB. You do the moves in your DAW. The compressor that follows will usually need a third of the gain reduction it needed before, and its attack and release can be set for the words, not for the song's structure.
What a flat phrase level actually buys you
- Less gain reduction, less artefact. A compressor doing 3 dB on peaks sounds like nothing happened. The same compressor doing 9 dB because the bridge was quiet sounds like a compressor. The difference is audible even on a laptop.
- One threshold that works for the whole take. If phrases sit within ±1.5 dB, the threshold you set on the verse still makes sense on the hook. Without that, you end up automating the threshold, which is clip gain with extra steps.
- De-esser and saturation behave. Both are level-dependent. A quiet phrase slides under the de-esser threshold and keeps its S sounds; a loud phrase pushes saturation into grit. Level first, then they react consistently.
- Tuning plugins track better. Pitch correction detects pitch more reliably on signal that is not dipping 8 dB between lines, and there is less of the thin, overcorrected sound on quiet phrases.
- Fader automation gets simple. Once the chain receives a consistent level, the fader rides at the end are small artistic moves — a lift on the last word of the chorus — instead of damage control.
How the planner measures
The file is mixed to mono and sliced into 50 ms frames. A frame counts as voice when its RMS level is within 35 dB of the loudest frame in the file; this keeps room tone and the noise between lines out of the picture without needing a gate setting. Runs of voice frames become phrases, with any gap shorter than 250 ms bridged, so the natural stops inside a line do not split it; anything under 150 ms is folded into its neighbour so a lip smack does not become a phrase. For each phrase the planner stores the start and end time, the RMS level in dBFS across the whole phrase and its peak sample. The target is the median phrase by default: the one in the middle when phrases are sorted by level, so you move roughly as many up as down and the take ends near its original average. Loudest and quietest targets are there for when you know which phrase has the sound you want. The move is simply target minus phrase level, rounded to 0.5 dB, and zero when the phrase is already inside your tolerance. Spread is the difference between the loudest and quietest phrase; the "compressor work" number is the gain reduction a compressor would have needed just to close that spread, before and after the plan.
When clip gain is the wrong tool
Phrases that are already within about 1.5 dB of each other do not need a plan, and the tool will say so. A phrase whose peak sits near 0 dBFS cannot be pushed up without clipping: lower the target instead, or let a limiter after the compressor catch it. Takes where the dynamics are inside the phrase — a word that jumps 10 dB in the middle of a line — are compressor territory, or a manual split of that one word. Expressive quiet phrases are also a judgement call: a whispered line that reads 8 dB under the chorus might be the point. Bring it up 4 dB so it survives the mix, not 8 so it sounds like the chorus. The plan gives you the measurement; the musical decision stays yours.
Playing the plan before you commit
"Play with plan applied" renders a copy of your take with each phrase's gain applied and 10 ms linear ramps at the phrase edges, which is the same shape clip-gain breakpoints or a volume lane would produce. The A/B switch flips between original and plan without stopping, so you hear whether the levelling alone already fixed the "the chorus is shouting, the verse is mumbling" problem — before any compressor is involved. If it did, you know your compressor can be gentle. If it did not, the problem is inside the phrases, and you have learned that for free.
FAQ
Should I use clip gain or a compressor to even out vocals?
Both, in that order. Use clip gain for the differences between phrases and sections, which are static, and leave the compressor for the differences between syllables, which are fast. A compressor forced to level whole sections needs far more gain reduction and sounds processed; after a clip-gain pass it usually needs 2–4 dB on peaks instead of 8–10.
Does this tool upload my vocal?
No. The file is decoded and analyzed in your browser with the Web Audio API and never leaves your device. There is no server-side step, no account and no microphone access. The only network request is fetching the demo take if you tap it.
What does "within 35 dB of max" mean for phrase detection?
The planner looks at the loudest 50 ms frame in your file and treats every frame within 35 dB of it as voice. Room tone and the quiet noise between lines usually sit 45–60 dB below the loudest word, so this separates singing from silence without a gate setting. A very noisy take or a long reverb tail can break that assumption — trim or gate first.
Which target should I choose?
Median is the safe default: it moves about as many phrases up as down and keeps the take's overall level where it was. Choose "loudest" when the biggest phrase has the energy you want and you can afford to lift the rest (watch the peak warning). Choose "quietest" when the loud phrases are clipping or spilling and you would rather bring everything down.
Why does the plan say a move would clip?
If a phrase already peaks close to 0 dBFS, adding gain pushes samples past full scale. Floating-point DAWs survive that internally, but anything that renders or exports to a fixed-point format will clip, and many plugins expect sane input. Lower the target, use the quietest phrase as the reference, or accept that a limiter after the compressor will catch it.
Is the dB value here the same as my DAW's clip gain value?
Yes — clip gain in every major DAW is a plain dB offset, so +4.5 dB in the plan is +4.5 dB on the clip. The absolute levels (RMS dBFS) are mono-summed RMS, which may differ slightly from a DAW meter, but the differences between phrases are what matter and those carry over exactly.
Will this make my vocal sound compressed or lifeless?
Not if you keep the moves honest. Levelling phrases to ±1.5 dB removes the accidental jumps, not the expression inside a line; the micro-dynamics are untouched. If a quiet phrase is quiet on purpose, move it less than the plan says. The goal is to let the compressor work less, not to replace it.
Can I use this on a stereo file, a doubled vocal or a full mix?
It works on any audio, but the result is only meaningful on a single voice. Stereo files are summed to mono for analysis; doubles and stacks will blur phrase edges; a full mix will read as one long phrase because there is no silence. Run it per take, before stacking.
Know the problem. Now copy the chain that solves it.
The Vocal Chain Bible is a 187-page PDF with 88 real vocal chains from hit records, as reported by the engineers in public interviews — mic, preamp, plugins and settings, including which compressor each engineer used after the gain stage and how hard they ran it. Every chain also comes in a free-plugin version, with the Vocal Chain Maker app included. Level your take with the plan above, then put a chain built for your genre behind it.
88 chains · every chain in a free-plugin version · Try one free chain firstFree download
Get the free Grammy Sauce PDF
The plugins and settings 100+ Grammy-winning engineers actually use — collected from 5 years of interviews, in one free PDF.
Want the full pro setup? Get the Vocal Chain Bible — 88 real chains →