How to Remove Room Sound From Vocals (What Actually Works)
Vocal mixing · 7 min read
You cannot EQ room sound out of a vocal, because the room is made of the same frequencies as the voice, arriving a few milliseconds later. What you can do is shorten it: an expander trims the tail between words, a transient shaper trims the sustain inside them, a couple of narrow EQ dips take the ring out of the room's resonances, and a deliberate short reverb covers what is left so it reads as a choice. Dedicated dereverb plugins do more, and they are paid; even they have a ceiling before the voice smears. Prevention at the mic is worth more than every tool combined.
What actually works
Shorten, then mask
Expander · transient shaper · 2 EQ dips · a better reverb on top
The 60-second answer
Downward expander first in the chain: ratio 1:2 to 1:3, threshold just under the quietest word, range −10 to −15 dB, attack 1 ms, release 100–200 ms. That trims the tail in the gaps without the chop of a gate. Next, a transient shaper with sustain at −3 to −6 dB to pull the room down inside the words. Then sweep a narrow bell between 200 and 600 Hz for where the room rings and cut 3 dB there, Q 3. High-pass at 100 Hz. Finally, send to a short, dark plate, 0.6–0.9 s, pre-delay 20–30 ms, at a level you only notice when you mute it. If the room is still obvious, the fix is the next recording, not another plugin.
Why EQ cannot remove reverb
Reverb is the voice itself, reflected and delayed. A 500 Hz vowel bouncing off a wall is still 500 Hz when it reaches the mic 4 ms later. There is no "room band" to cut. What a room does have is a few frequencies where it stores energy longer than others, and those show up as ringing at fixed spots on every take. EQ handles that part only, which is why the dips here are narrow: you are removing the ring, not the room.
Everything else is a time problem, and time problems take time-based tools. An expander turns the vocal down when the direct sound has stopped and only the tail remains. A transient shaper separates a note's attack from its decay and lets you turn the decay down. Dereverb plugins model what is direct and what is reflected across the whole spectrum; the model is the product, which is why they cost money. Run your take through the Vocal Problem Finder first; if it flags low-mid ringing as well as room, you have an EQ problem and a time problem, in that order.
What each tool actually removes
None of these stages removes "the room". The expander only works between words; during a held note the direct sound keeps it open. The transient shaper works inside words but also softens the voice's natural sustain, so past −6 dB the vocal sounds clipped off. EQ dips only remove ringing resonances. The masking reverb removes nothing; it hides an ugly short tail under a nicer, longer one, which works far more often than it should. Dereverb tools genuinely subtract reflected energy, with artifacts arriving as you push past a few dB. Stack the free tools in that order and the realistic result is a vocal that sounds like a smaller, drier room. That is the ceiling.
Reducing room sound step by step
| # | The move |
|---|---|
| 01 |
Clip-gain the phrases level first An expander threshold only works if quiet and loud lines are within a few dB. Level them by hand before any dynamics. Phrases within ±3 dB |
| 02 |
Expander, not gate, first in the chain Set the threshold while looping the quietest word so it stays open. Limited range so the gaps get quieter, not silent. 1:2–1:3 · range −10 to −15 dB · attack 1 ms · release 100–200 ms |
| 03 |
Transient shaper, sustain down Pulls the decay inside each word down. Stop when word endings start sounding cut; back off 1 dB from there. Sustain −3 to −6 dB · attack 0 |
| 04 |
Find and cut the ring Narrow +8 dB sweep between 200 and 600 Hz on a word with a long tail. The room's resonance will howl. Cut it; check once more above 1 kHz. Bell −3 dB · Q 3 · HPF 100 Hz |
| 05 |
Compress after, not before Compression raises the tail. Put it after the expander and shaper so it lifts a drier signal, and keep makeup modest. 3:1 · 3–5 dB GR |
| 06 |
Mask with a reverb you chose Short dark plate on a send, with pre-delay so the direct sound stays clear. The ugly tail hides under the pretty one. Plate 0.6–0.9 s · pre-delay 20–30 ms · LPF on the send 6–8 kHz |
The trick: duck the room with the voice
A room tail is most audible in the 100 ms after each word ends. Instead of one expander with a fixed release, add a second, gentler one keyed from the vocal itself with a slow release of 300–500 ms, so the moment the direct sound stops the level starts falling and keeps falling until the next word opens it again. Most stock expanders can do this. The tail now fades the way a treated room would, rather than holding for 200 ms and dropping. Then duck the masking plate from step 06 under the vocal too, using the method in how to duck reverb under vocals; the listener hears one controlled space instead of two fighting ones.
Prevention: what the next take needs
Every stage above costs something: expansion costs breaths, shaping costs sustain, EQ costs body, masking costs clarity. A take with less room in it costs nothing. Mic at 10–15 cm so the direct sound dominates. Cardioid pattern with its rear pointed at the most reflective surface. A duvet behind and beside the singer, not behind the mic. Out of the room's center, away from parallel bare walls. The full setup is in how to record vocals in an untreated room; twenty minutes of it does what no free plugin can.
Chains with the reasons left in.
The Vocal Chain Bible is 88 real vocal chains from hit records, as the engineers described them — mic, preamp, every plugin and setting, in order, with a free-plugin version of every chain. Every chain comes with the reason behind each setting, so you know what to move for your voice.
Get the Vocal Chain Bible — $48 →Mistakes that kill it
- Using a hard gate. Full range makes the room appear and vanish with every word, which draws more attention to it than leaving it alone. Limit the range to 10–15 dB.
- Compressing before expanding. Compression brings the tail up toward the voice. After that the expander has nothing clean to separate.
- Wide EQ cuts to "remove the room". A −6 dB scoop across 200–800 Hz removes the body and leaves the reverb. Dips should be narrow and aimed at a ring you found by sweeping.
- Adding a long hall to hide a short room. A 2.5 s hall on a roomy vocal is a roomy vocal in a hall. Mask with something short and dark; the vocal reverb settings article has the starting points.
Keep going
FAQ
Can you remove reverb from a vocal recording?
Reduce, not remove. Expansion, transient shaping and narrow EQ dips can make a roomy take sound like a smaller, drier space. Dedicated dereverb plugins go further but are paid and still have a ceiling before the voice smears. Nothing makes an untreated room sound like a booth.
Is there a free dereverb plugin?
Not a dedicated one that competes with the paid tools, at the time of writing. The free route is the chain in this article: a stock expander, a free transient shaper, your DAW's EQ and a short plate on a send.
Gate or expander for room sound?
Expander. A gate is an expander with an infinite ratio and full range, which is exactly the setting that chops breaths and word endings. Ratio 1:2 to 1:3 with the range limited to 10–15 dB is quieter in the gaps and invisible on the words.
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 →