How to make the quiet parts of a recording louder, without squashing it.

You recorded a great take, but the soft lines disappear and the loud ones jump out. Here are four ways to even it out, from free and slow to quick, and what each one does to the sound.

Why it happens

A live performance moves. A singer leans in and out from the mic, a pianist plays a whisper and then a full chord, a podcast guest drops their voice at the end of every sentence. A difference of 10 to 20 dB between the quiet and loud parts is normal. In the room your ears adjust to it. On headphones or a phone speaker, the quiet parts get lost under the noise of a bus or a kitchen, so listeners turn up the volume, and then the loud parts are too much.

The goal is to make the quiet parts audible and the loud parts comfortable, while keeping the shape of the performance: a phrase that grows should still grow.

1. Volume automation by hand (free, precise, slow)

Every DAW lets you draw the volume of a track over time. In Logic and GarageBand, press A to show automation and draw the volume line; in Ableton, use clip gain or the mixer volume envelope; in Audacity, use the Envelope tool; in Reaper, item volume envelopes. Raise the soft phrases, lower the loud ones.

Good: nothing is changed that you didn't decide, and it never sounds processed. Less good: it takes a long time for a long recording, and it's easy to overdo the jumps, which you hear as the room noise going up and down.

Tip: work phrase by phrase, not word by word, and make the changes gradual. A few decibels is usually enough.

2. A compressor (quick, but it changes the sound)

A compressor turns the sound down whenever it goes over a threshold, then you raise the whole track with makeup gain, so the quiet parts end up louder. It's fast and every DAW has one.

The catch is that it reacts to every moment as it happens. Push it hard and it flattens the playing, the attacks of notes go soft, and the background noise breathes up in the pauses ("pumping"). On a piano or an acoustic guitar you hear it quickly.

If you use one for levelling, be gentle: a low ratio (around 2:1), a slow attack so note beginnings survive, a medium release, and only a few decibels of gain reduction. Two light compressors in a row sound more natural than one heavy one.

3. A leveling or "vocal rider" plug-in

These plug-ins move a volume fader for you while the track plays, aiming for a target level. They're smoother than a compressor because they ride slowly instead of grabbing peaks. But they still hear one moment at a time, so they can't know that a soft line is about to grow into a climax, and you find out what they did afterwards.

4. Read the whole recording first, then draw one curve

This is what Gain Curve does, and why I made it. Instead of reacting as it plays, it looks at the entire take first and draws one smooth curve of how loud each moment should be. You see every move before you hear it, and you can redraw any part with the mouse (or with your finger in the iPhone and iPad version).

  1. Open the recording in the app, or put Gain Curve on the track in your DAW.
  2. Press Auto-draw and pick the detail: every note, every phrase, or whole sections. Phrase is a good start for vocals and piano.
  3. Press Smooth once or twice to round off sharp jumps, and use Amount to decide how much to even things out: 100% makes every part equally loud, 50% halves the differences.
  4. Fix any spot by hand, then press Make it ready to set the loudness for where you're releasing.

It's the same idea as hand-drawn automation, only drawn for you in a few seconds, so you keep the control and skip the hours.

Then set the final loudness

Evening out the performance and making it loud enough for release are two different steps. Once the quiet and loud parts sit right, set the overall loudness for where it's going, with a limiter keeping the peaks safe. Spotify and YouTube aim for about −14 LUFS, Apple Music for −16. See what LUFS your song should be.

Quick checklist

Paul Voia-Tipei, PVT Studio · October 2026