1. Learn
    2. Dynamics
    3. Threshold and ratio

    Compressor threshold and ratio

    The threshold is the level where a compressor starts turning a sound down, and the ratio sets how hard it turns down whatever goes over. At a −20 dB threshold and 4:1, a peak at −8 dB is 12 dB over and comes out 3 dB over, at −17 dB. Anything quieter than the threshold passes through unchanged.

    Threshold
    The level where compression starts, in dBFS
    Ratio
    How much of the level over the threshold gets through
    At 4:1
    A peak 8 dB over comes out 2 dB over
    Starting point
    3:1 or 4:1, with 3 to 6 dB of gain reduction
    Played at the same loudness
    Step 1 of 8

    Hear the vocal as it is

    Press play. The loudest word peaks at about −1 dB, so a threshold at 0 dB catches nothing and the amber lane stays flat. Attack, release, knee and mix are locked for this lesson.

    Press play and listen.
    Most turned down0.0 dB
    Loudest peak out−0.9 dBFS

    Look for: The amber lane is flat: nothing is turned down

    0 dB
    4:1
    1.0 ms
    100 ms
    0 dB
    +0.0 dB
    100%
    Train your earsThe lab picks hidden settings and you name what you are hearing. The controls are covered while you play, and the levels are matched so loudness never gives it away.

    Recording by Paranoye on Freesound, public domain (CC0).

    What are threshold and ratio?

    They're the two controls that decide how much a compressor does. The threshold is a line drawn at one level: below it the compressor leaves the sound alone, and above it the compressor starts turning it down. The ratio decides how far down. Every other control on a compressor changes how quickly or how smoothly that happens, but these two set the amount.

    Say you set the threshold to −20 dB and the ratio to 4:1. A word in a vocal that peaks at −8 dB is 12 dB over the threshold, and at 4:1 every 4 dB over becomes 1 dB over, so it comes out 3 dB over, at −17 dB. The compressor has taken 9 dB off that word. A softer word at −24 dB never reached the line, so it comes out at −24 dB, exactly as it went in.

    The loud word and the soft word were 16 dB apart before and they're 7 dB apart after. Shrinking that gap is what compression is for, and the compression lesson covers why you'd want to and what the other controls do. This page stays on threshold and ratio, and on how to set them for a real sound.

    Using the lab on this page

    The lab above is the same compressor as the main compression lesson, set up for this one subject. Attack is locked at 1 ms and release at 100 ms, fast enough that the compressor follows the vocal closely, so what you hear and see is down to the threshold and ratio. The knee is locked at 0 dB (a hard knee), so the transfer curve bends exactly where the worked numbers on this page say it should.

    The Transfer curve picture shows the transfer curve, which draws both controls as one line. Input level runs along the bottom and output level up the side. The dashed amber line is the threshold. Left of it the pink line sits on the diagonal, meaning what goes in comes out. Right of it the line bends flatter, and the higher the ratio, the flatter it gets. While the loop plays, a white dot rides the line at the current level.

    The Whole loop picture shows the loop with the compressed version in pink over the original, and the amber lane along the bottom shows gain reduction moment by moment. Above the pictures, the Turning down readout gives that gain reduction in decibels, and Loudest peak out shows the highest level coming out of the compressor.

    The lesson starts on a vocal because its words land at lots of different levels, so you can watch the threshold catch more of them as it comes down. Drums, bass, acoustic guitar and the full mix are there too, and Your own sound lets you load a file of your own.

    What does the threshold do?

    The threshold sets the level a sound has to reach before the compressor reacts. It's measured in dBFS, where 0 is the loudest level digital audio can hold, which is why thresholds are nearly always negative numbers. A −18 dB threshold means 18 dB below full scale.

    When you lower the threshold, more of the sound goes over the line, so the compressor works more of the time. Whatever was already over is now further over, and it gets turned down by more. In the lab, moving the vocal's threshold from −10 dB to −20 dB takes the most gain reduction from about 4 dB to about 12 dB, and the share of the loop being compressed goes from roughly a tenth to most of it.

    The threshold is relative to the level of the sound coming in, and that catches people out. Turn a vocal up by 6 dB before the compressor and every word is 6 dB further over the same threshold, so the compressor does a lot more. That's why a threshold copied from a preset or a settings chart rarely does the same job on your track: your vocal might be recorded 10 dB quieter or louder than the one the preset was made for. Set the threshold by watching the meter, and keep levels into your plug-ins sensible, which the gain staging lesson covers.

    Some classic compressors don't have a threshold knob at all. The Universal Audio 1176 has a fixed threshold and an Input knob, so you get more compression by pushing more signal into it, and an Output knob to bring the level back down. The Teletronix LA-2A has a Peak Reduction knob that sets how much it compresses. Both come to the same thing as lowering a threshold, because more of the sound ends up over the line.

    What does the compression ratio mean?

    The ratio compares how far the input goes over the threshold with how far the output is allowed to go over. At 4:1, an input 4 dB over comes out 1 dB over, and an input 12 dB over comes out 3 dB over. At 2:1 those become 2 dB and 6 dB. A ratio of 1:1 means the output always equals the input, so the compressor does nothing however low the threshold is.

    If you'd like it as a sum, take how far the peak is over the threshold and divide by the ratio. That's how far over it comes out. The gain reduction is the difference between the two, so a peak 12 dB over at 3:1 comes out 4 dB over and loses 8 dB.

    Each line is a different ratio at the same threshold. Below the threshold they're identical. Above it, a higher ratio gives a flatter line, so loud peaks come out closer to the threshold.

    The table below runs that sum for one peak at different ratios. Look at the right-hand column, because it shows something the knob doesn't: going from 1.5:1 to 4:1 adds 5 dB of gain reduction, while going from 4:1 all the way to ∞:1 adds only 3 dB more. Most of the change happens at low ratios, and past about 10:1 turning the ratio up hardly changes anything.

    A peak 12 dB over the threshold at different ratios
    RatioComes outGain reduction
    1:112 dB over0 dB
    1.5:18 dB over4 dB
    2:16 dB over6 dB
    3:14 dB over8 dB
    4:13 dB over9 dB
    8:11.5 dB over10.5 dB
    10:11.2 dB over10.8 dB
    ∞:10 dB over12 dB

    On the vocal in the lab, with the threshold at −20 dB, the loudest words lose about 8 dB at 2:1 and about 14 dB at 10:1. At 2:1 you can still hear which words were sung harder. At 10:1 the loud words land only a few decibels above the quiet ones and the performance starts to sound flattened.

    Turn the lab's ratio all the way up and the label reads ∞:1. Strictly it's 20:1, which is close enough that a peak 10 dB over comes out half a decibel over. A limiter behaves like an infinite ratio, letting nothing out above its threshold. The compression vs limiting lesson covers where one becomes the other.

    How do threshold and ratio work together?

    Neither control sets the amount of compression by itself. A low ratio with a low threshold and a high ratio with a high threshold can take the same number of decibels off the loudest peak and still sound very different, because they treat everything below the peak differently.

    Try it in the lab. With the threshold at −12 dB and the ratio at 10:1, only the biggest words cross the line, and they lose about 7 dB. Everything else is left alone, and the compressor works on under a third of the loop. Now move to −22 dB at 1.6:1. The loudest words lose about the same 6 to 7 dB, but most of the vocal is over the threshold, so the compressor is pressing gently nearly all the time.

    The first setting is peak control. It keeps the shape of the performance and stops the few moments that jump out, which is what you want on a vocal with one or two shouted words or a bass with the odd note that pops. The second setting is levelling. It squeezes the whole sound into a narrower range and it's harder to hear working, which is why low ratios with low thresholds turn up on mix buses and groups, where 1 or 2 dB of gentle, constant compression helps the parts move together.

    Pick by asking what's wrong with the sound. If a few moments are too loud and the rest is fine, use a higher threshold and a higher ratio. If the whole part wanders up and down in level, use a lower threshold and a lower ratio. On a vocal you can use both in a row, a fast compressor catching peaks followed by a gentle one evening out the rest, and the parallel vs serial compression lesson covers that.

    Which ratio for which sound?

    These are starting ranges. Where you set the threshold depends on your track's level, so it isn't in the table; aim it for the gain reduction in the right-hand column on the loudest moments.

    Starting ratios and gain reduction by source
    SourceRatioGain reduction on peaks
    Mix bus1.5:1 – 2:11 – 2 dB
    Drum bus2:1 – 4:11 – 3 dB
    Acoustic guitar2:1 – 3:12 – 4 dB
    Lead vocal3:1 – 4:13 – 6 dB
    Bass4:1 – 6:13 – 6 dB
    Kick or snare4:1 – 6:13 – 6 dB
    Catching stray peaks8:1 – 20:1a few dB on the worst ones

    Hardware often gives you a handful of fixed ratios instead of a knob. The 1176 has buttons for 4:1, 8:1, 12:1 and 20:1, and the SSL bus compressor has a switch for 2:1, 4:1 and 10:1.

    How to set threshold and ratio, step by step

    Set the ratio first and the threshold second, because the threshold is the control you'll fine-tune by ear. The last step in the lab ends on the settings this method gives on the vocal.

    1. Choose a ratio for the job4:1 for most single sounds, 2:1 for a group or a mix. You can change it later.
    2. Start with the threshold at 0 dBPlay the loudest section of the part, and check the gain reduction meter reads nothing.
    3. Lower the threshold until the meter movesBring it down slowly until the loudest moments just start to register. Now you know where the line sits against your sound.
    4. Keep going to the amount you wantFor a lead vocal, stop when the loudest words show 3 to 6 dB of gain reduction and the quiet ones show little or none.
    5. Adjust the ratio by earIf the loud moments still jump out, raise the ratio. If everything sounds squashed, lower it and bring the threshold down a little to keep the gain reduction.
    6. Match the level and compareAdd makeup gain until the compressed sound is as loud as the original, then switch the compressor off and on. The makeup gain lesson explains why the levels have to match.

    Detection, knee and timing

    The compressor in this lab compares each sample's level with the threshold, so the very tops of the waveform count. That's peak detection. Many compressors average the level over a short window first, called RMS detection, which reads lower than the peak level on spiky sounds like drums. Put a drum loop through an RMS compressor and you'll need a lower threshold to get the same gain reduction, because the short transients hardly move the average. The peak vs RMS compression lesson compares the two.

    The knee changes the corner where the transfer curve bends. With the hard knee locked here, compression switches on at the exact threshold. A soft knee eases the ratio in over a range of levels around the threshold, so a sound just over the line gets less than the full ratio. The compressor knee lesson has its own lab.

    Attack and release change how closely the real gain reduction follows the transfer curve. With a slow attack, a short peak can get well past the threshold before the compressor catches up, so the ratio on the knob only applies once it has. That's covered in the attack and release lesson.

    Common mistakes with threshold and ratio

    Reaching for the ratio when the threshold is the problem

    If only one word in a verse is setting off the compressor, turning the ratio up just squashes that word harder. To get the compressor working on more of the part, lower the threshold.

    A threshold so low the compressor never lets go

    When everything is over the threshold, the gain reduction meter never returns to zero and the compressor turns down the quiet parts along with the loud ones. The sound gets flat, and breaths and room noise come up once you add makeup gain. If the amber lane on the lab's Whole loop picture covers nearly all of the loop, raise the threshold until the quiet passages fall back under it.

    Copying numbers instead of reading the meter

    "Threshold −18 dB, ratio 4:1" tells you nothing until you know how loud the track is. The same settings can do nothing on a quiet recording and crush a loud one, so copy the ratio if you like and set the threshold by the gain reduction it gives.

    Judging it at a different level

    Lowering the threshold makes the output quieter, and quieter usually sounds worse, so heavy compression can seem to dull a sound when most of the change is a few decibels of level. Too much makeup gain fools you the other way. Match the level before you decide, which the lab's Original and Compressed switch does for you.

    A high ratio on a whole mix

    On a mix bus, 4:1 with a threshold low enough to work on most of the song turns every loud kick into a dip in everything else. Stay at 1.5:1 or 2:1 there and keep the gain reduction to a couple of decibels.

    Train your ears

    The quiz under the lab hides the sliders, sets a compressor and matches the levels so loudness can't give the answer away. Every round asks you to pick the ratio from 2:1, 4:1 and 10:1, which comes down to how far the loud moments stand above the quiet ones. The gap is easiest to hear on the drums, so start there before trying the vocal. Your best score is saved in your browser.

    Questions people ask

    What is a good compression ratio?

    For most single instruments and vocals, 3:1 or 4:1 is a sensible start. Go lower, 1.5:1 to 2:1, on groups and whole mixes, and higher, 6:1 to 10:1, when you want peaks held firmly. Past about 10:1 you're close to limiting.

    What threshold should I set?

    There's no correct number, because the threshold depends on how loud the sound arriving at the compressor is. Lower it until the gain reduction meter shows 3 to 6 dB on the loudest moments, then judge by ear.

    Does a higher ratio mean more compression?

    For anything over the threshold, yes: a higher ratio takes more off. But a high ratio with a high threshold can compress less of the sound than a low ratio with a low threshold, because fewer moments go over. The two controls set the amount together.

    What does ∞:1 mean?

    An infinite ratio means nothing comes out above the threshold, however far over the input goes. That's how a limiter behaves. In practice a compressor set to ∞:1 still lets a little through on fast peaks, because the attack takes time.

    Why is the threshold a negative number?

    Digital levels are measured in dBFS, where 0 is the loudest possible level, so everything below it is negative. A threshold of −18 dB means 18 dB below full scale.

    What's the difference between 4:1 and 8:1?

    For a peak 12 dB over the threshold, 4:1 lets it out 3 dB over and 8:1 lets it out 1.5 dB over. That's 9 dB of gain reduction against 10.5 dB, so doubling the ratio here only adds 1.5 dB.

    Keep going

    Zac Dobos

    Zac runs Sounds Like You, which makes sample packs, preset packs and the Murmur granular synth plug-in. The lessons in Learn are free, and every one has a working version of the effect in it.