There is a moment in almost every mix where I want more reverb than the vocal can survive. Push the send until the track sounds like a record, and the words go soft. Back it off until every syllable is clear, and the vocal sits flat against the speaker with nothing behind it. Sidechain reverb is how you stop choosing. The tail ducks out of the way while the singer is singing, then swells back into the gaps between phrases. You hear the space without losing the words. Below is how to set it up by hand in any DAW, the settings I actually use, and where the manual version stops being worth the effort.
Why reverb and vocals fight in the first place
Reverb is a copy of the vocal, spread out in time and stacked on top of itself. That copy occupies the same frequencies as the voice that made it, which is the whole problem. When the tail from the last word is still ringing as the next word starts, the two compete in the same part of the spectrum, and the newer one loses definition. Engineers call it masking. You hear it as mud, or as a vocal that seems quiet no matter how far you push the fader.
The usual fixes attack the symptom. High-pass the return, shorten the decay, turn the send down. They all work, and they all cost you the thing you wanted. Ducking is different because it works in time instead: keep the big tail, and simply get it out of the way during the moments when the voice needs the room.
The classic setup: a compressor on the reverb bus
This works in every DAW and costs nothing. The routing is the part people get wrong, so it is worth being precise about the order.
- Put the reverb on its own aux or send, never as an insert on the vocal. Ducking needs the wet signal on a separate path so you can compress it without touching the dry voice.
- Add a compressor on that return, after the reverb. Before it does nothing useful: you would be compressing the input, not the tail.
- Set the compressor's sidechain input to the dry vocal. In Ableton and Logic this is the sidechain dropdown. In Pro Tools it is a bus send with the key input set to it. In FL Studio you route the vocal to the compressor's sidechain input on the mixer.
- Check you are hearing it: solo the return and play the vocal. The reverb should visibly drop each time the singer comes in and rise back in the pauses.
Settings that actually work
Start at a 4:1 ratio, the fastest attack the compressor has, and a release somewhere between 40 and 100 ms. Then set the threshold so the tail drops about 4 to 6 dB while the vocal is going. That is usually enough to clear space without the reverb audibly pumping.
The release is the control that matters, and it is the one worth spending time on. Too fast and the reverb jumps back in between syllables, which chatters. Too slow and the tail never returns during the gaps you wanted it to fill, so you get all of the cost and none of the payoff. Tune it against the actual cadence of the performance: a slow ballad wants a longer release than a dense rap verse, because the gaps are longer.
One thing worth knowing before you spend an hour on it. Because the compressor reacts to the whole vocal, it pulls the entire tail down, including the low end and the air that were never in the way. That is why heavily ducked reverb can sound like it is disappearing rather than moving aside.
Level ducking versus spectral unmasking
These get talked about as if they are one technique. They are two, and knowing which one a plugin does tells you what it will sound like.
Level ducking turns the whole wet signal down when the dry signal is present. That is what the compressor above does. It is simple, it is predictable, and it treats every frequency the same, so a bright vocal pulls down the low reverb it was never masking anyway.
Spectral unmasking splits the tail into bands and ducks each band by how much energy the dry signal has in that same band right now. Sing a low note and the low part of the tail steps aside while the top stays. It keeps more of the reverb audible for the same amount of clarity gained, because it only removes what is actually in the way.
Neither is better in the abstract. Level ducking is easier to predict and easier to automate. Spectral unmasking preserves more of the space you paid for. Several plugins now do one, and a few do both.
Where the manual method stops being worth it
I still set up a sidechained compressor regularly, and for one vocal on one send it is fine. The friction shows up when the session grows.
Every new source that needs it means another return, another compressor, and another sidechain route. Ten tracks in, that is a lot of plumbing to keep straight. The compressor is also blind to frequency, per the section above. And if the compressor you reach for uses lookahead, you have quietly added latency to a send, which matters more than people expect on a busy session.
None of that makes the manual method wrong. It makes it a technique worth knowing and worth automating once you find yourself doing it on every track.
The reverbs that duck themselves
Enough plugins now build this in that it is worth knowing which mechanism each one uses, because that decides what it can and cannot do for you. Prices below were checked against each maker's own store or an official retailer in July 2026, and the ones that discount heavily are marked, because reverb plugins go on sale constantly.
TAL-Reverb-4 is the surprise of the group. It is free, it has a ducking mode, and it takes an external sidechain input, which some paid plugins do not. If you want to hear what ducked reverb does before spending anything, start there.
iZotope Aurora ($49) does spectral unmasking with attack, release, amount and sensitivity controls, plus a delta button that plays you only the frequencies it is removing. It listens to its own input by design and does not take an external sidechain, so it cannot duck a send under a different track. FabFilter Pro-R 2 ($199) includes level ducking inside a reverb most people buy for its decay-rate EQ and its sound. Baby Audio Crystalline ($99, discounted often) has a Ducker with a standard and a gentler mode. sonible smart:reverb 2 ($129 list, frequently much less) unmasks the direct signal through its Clarity control.
AURA is the one I make, so weigh this accordingly: its CLARITY knob runs a level duck and a six-band spectral unmask together, and its EXT key switches the whole thing to listen to a second input, which is the case Aurora explicitly does not cover. It reports zero latency, and there is a free AURA Lite if you want the mechanism without the price. It is also on version 1.0.0 with no track record behind it, and there is no AAX build, so Pro Tools users are out.
Which one to reach for
If you are doing this once, on one vocal, use the compressor. It is free, you already own it, and understanding the routing makes you better at the plugin versions anyway.
If you are doing it on most tracks, a reverb that ducks itself removes a per-track chore. Check whether it takes an external sidechain before you buy, because that decides whether it works on a shared send fed by several sources. And if you mostly need it on one lead vocal at a time, that limitation may never come up.
Reverbs with built-in ducking (July 2026)
| Plugin | Price | Mechanism | External sidechain | Free version |
|---|---|---|---|---|
| TAL-Reverb-4 | Free | Level ducking | Yes | It is free |
| iZotope Aurora | $49 | Spectral unmasking | No | Trial only |
| Baby Audio Crystalline | $99 (often less) | Level ducking | Not stated | Trial only |
| sonible smart:reverb 2 | $129 (often less) | Spectral unmasking | Not stated | Trial only |
| FabFilter Pro-R 2 | $199 | Level ducking | Not stated | Trial only |
| AURA | $34.99 | Both, on one knob | Yes (EXT) | AURA Lite |
Conclusion
Sidechain reverb is not a trick so much as a correction to a bad tradeoff. You were choosing between a vocal that cuts through and a mix with depth, and there was never a reason to. Set up the send, put a compressor after the reverb, key it off the dry vocal, and spend your time on the release rather than the ratio.
Once it is on every track in the session, it is worth letting the reverb handle it. Which one depends on whether you need an external key and whether you care about keeping the parts of the tail that were never in the way. The same idea works on delay, and the ducked delay technique is one of seven in the vocal delay guide if you want to hear it on repeats instead of tails.
Frequently Asked Questions
What is sidechain reverb?
Sidechain reverb is reverb whose level is controlled by the dry signal feeding it. A compressor sits on the reverb return and is keyed off the dry vocal, so the tail drops while the singer is singing and rises back in the gaps between phrases. The result is a mix that sounds spacious without the reverb burying the words. It is also called ducking reverb.
How do you sidechain reverb to vocals?
Put the reverb on its own aux send rather than as an insert. Add a compressor on that return, positioned after the reverb. Set the compressor's sidechain or key input to the dry vocal. Start with a 4:1 ratio, the fastest attack available, and a release of 40 to 100 ms, then set the threshold so the tail drops about 4 to 6 dB whenever the vocal plays.
What is the difference between ducking reverb and spectral unmasking?
Level ducking turns the entire wet signal down when the dry signal is present, treating all frequencies the same. Spectral unmasking splits the tail into frequency bands and ducks each band according to how much energy the dry signal has in that band at that moment. Unmasking keeps more of the reverb audible for the same gain in clarity, because it only removes the parts that are actually masking the source.
What compressor settings should I use for ducking reverb?
A 4:1 ratio, the fastest attack the compressor offers, and a release between 40 and 100 ms is a reliable starting point. Set the threshold for roughly 4 to 6 dB of gain reduction while the vocal is playing. The release is the setting worth tuning: too fast chatters between syllables, too slow means the tail never returns during the pauses you wanted it to fill.
Is there a free reverb plugin with ducking?
Yes. TAL-Reverb-4 is free, includes a ducking mode, and accepts an external sidechain input, which some paid reverbs do not offer. It runs on Windows, macOS and Linux in VST, VST3, AU and AAX formats. AURA Lite is also free, though it is a simplified four-control version of a paid plugin rather than a standalone free reverb.
Do I need a special plugin to duck reverb?
No. Any compressor with a sidechain input will do it, and the routing is the same in every major DAW. A reverb with built-in ducking saves you from repeating that setup on every track, and some offer frequency-dependent unmasking that a plain compressor cannot do, but the technique itself does not require buying anything.
Does ducking reverb work on drums and synths?
Yes, and it is common on drum busses where a long room reverb would otherwise smear transients. The settings change more than the method: percussive sources usually want a faster release than vocals so the tail recovers between hits. On sustained synths the effect is subtler, because there are fewer gaps for the reverb to return in.
Why does my vocal sound muddy even with reverb ducking?
Ducking only addresses masking during the moments the dry signal is playing. If the reverb is still muddy, the usual causes are a decay that is too long for the tempo, no high-pass on the return, or too much send level overall. Ducking works best combined with a high-pass around 200 to 300 Hz on the reverb return and a decay time that clears before the next phrase lands.
Try AURA free for 14 days
AURA ducks and unmasks its own tail from one knob. The trial is the full plugin, on Windows, macOS and Linux.
Or start with AURA Lite, free forever.