How to Soundproof a Room: What Actually Works (and What Doesn't)

Start With a Number, Not a Shopping Cart

Almost every soundproofing project that disappoints people started the same way: they bought something, installed it, and then tried to judge the result from memory. Memory is a terrible sound level meter. A 4 dB improvement is genuinely useful and completely invisible to recollection a week later.

Take a baseline before you touch anything. Stand in the quiet room, point the microphone at the wall or door the noise is coming through, and record the level during the noise you actually care about — the neighbour's television, the street at rush hour, the washing machine on spin. Note the time of day and where you stood, because you will need to repeat the measurement in the same place to compare.

Try it right now against the room you are sitting in:

Soundproofing Is Not Acoustic Treatment

This is the distinction that decides whether your money does anything at all. They are two different jobs with two different product categories, and the marketing for one is written to be mistaken for the other.

• Soundproofing (isolation) stops sound moving between two spaces. It needs mass, airtightness, and broken physical connections. It is measured in STC.
• Acoustic treatment (absorption) controls sound bouncing around inside one space. It needs porous material on the reflective surfaces. It is measured in NRC.

A two-inch foam panel weighs a few ounces per square foot and is full of air. It is very good at soaking up a reflection and essentially useless as a barrier. Covering a shared wall in foam will make your side of the wall sound drier and deader — which people often misread as quieter — while the neighbour's television comes through exactly as before.

If your problem is that your room echoes, your voice sounds boxy on calls, or your recordings are muddy, absorption is the correct fix and the rest of this article is not what you need. If your problem is a person, a road, or a machine on the other side of a surface, keep reading.

Step 1: Seal Every Gap (Cheapest, Highest Return)

Airborne sound behaves a lot like water: it finds the hole. A wall rated STC 50 with a one-percent open area performs closer to STC 30 in practice. This is why sealing comes first and why it so often beats the expensive step people were planning instead.

Work through the leaks in descending order of size:

• Under the door. A standard interior door has a 1/2–3/4 inch undercut for airflow. That gap is an open channel between two rooms. A door sweep plus perimeter weatherstrip typically buys 3–5 dB for under $30 and an hour of work.
• The door itself. A hollow-core door is a cardboard-cored panel weighing about 25 lb. A solid-core replacement weighs 70–100 lb and is worth roughly 10 STC points on its own — but only once the perimeter is sealed. An unsealed solid door barely outperforms a sealed hollow one.
• Electrical boxes. Outlets and switches are holes cut through the drywall, and in shared walls they are often installed back-to-back in the same stud bay, leaving a near-direct air path. Foam gaskets behind the cover plates take twenty minutes.
• Perimeter joints. Where drywall meets floor, ceiling and adjacent walls, there is usually a gap behind the trim. Bead it with a non-hardening acoustical sealant. Ordinary latex caulk cures rigid, cracks as the building moves, and quietly reopens the leak within a season.
• Ducts and vents. A shared return duct can carry conversation between rooms better than the wall does. Lining the duct or fitting a baffle box is a bigger job, but it is worth identifying before you spend money on the walls.

Step 2: Add Mass

Once the air paths are closed, what remains is the wall itself vibrating and re-radiating sound on the far side. Heavier surfaces vibrate less for a given amount of energy. As a rough rule, doubling the mass of a single-leaf partition is worth about 5 dB — real, but note that it takes a doubling each time, which is why mass alone gets expensive fast.

In practice mass means one of two things. A second layer of 5/8 inch drywall over the existing wall is the cheapest mass per square foot you can buy and the standard approach. Mass loaded vinyl — a dense, limp sheet at roughly 1 lb per square foot — is the alternative where thickness or weight on the structure is a problem. MLV works because it is limp as well as heavy; stiffness lets a panel couple to the sound and carry it, which is why a rigid panel of the same weight performs worse.

MLV is also the material most often installed wrong. Every seam overlapped and taped, every edge sealed to the framing, every penetration caulked — an unsealed MLV installation is an expensive way to learn the lesson from Step 1 again.

Insulation belongs in this step too, though it works differently. Filling a stud cavity with mineral wool stops the trapped air acting as a spring between the two faces of the wall. In an otherwise untouched cavity that is worth around 4–6 STC points; combined with decoupling it is worth considerably more. It is the cheapest upgrade available whenever the wall is already open.

Step 3: Damping — The Cheapest Real Upgrade

If you are adding that second layer of drywall anyway, a viscoelastic damping compound between the two sheets is the best value in the whole field. The compound stays permanently soft; when the sandwich flexes, the two sheets shear against each other through it, and the vibration is converted to a trivial amount of heat instead of being passed along.

Two sheets of drywall screwed together behave acoustically like one thick sheet. The same two sheets with damping compound between them are worth roughly 8–10 STC points more, with the biggest improvements at low frequencies, which is exactly where the neighbour's subwoofer lives and where mass alone struggles.

The cost is two tubes per 4×8 sheet and about ten extra minutes per sheet. There is no version of this project where adding the second layer without damping is the right call.

Step 4: Decoupling — The Biggest Gains

Everything above fights sound travelling through the air. Decoupling attacks the other path: vibration conducted through the solid structure. A standard wall has drywall screwed directly to studs that are screwed directly to the drywall on the other side. That is a mechanical bridge, and no amount of mass on one face fixes a bridge.

The practical fix is sound isolation clips carrying hat channel: rubber-isolated clips screw to the studs, steel channel snaps into the clips, and the new drywall hangs on the channel without ever touching the framing. Done properly and combined with cavity insulation and a damped double layer, this takes a typical STC 33 partition well past STC 55.

Two warnings, because this is the step where projects get ruined. First, a single screw long enough to reach a stud through the channel short-circuits the entire assembly — one screw can cost you most of the benefit you paid for. Second, the new wall must not touch anything: leave a gap at the floor and ceiling and seal it with acoustical sealant rather than letting the drywall bear on the slab.

Expect to lose about an inch and a half of room on that wall, to redo the trim and the outlet boxes, and to spend a weekend per wall. In exchange this is the one step that reliably makes a neighbour inaudible rather than quieter.

Footfall, Appliances and Other Structure-Borne Noise

If the noise is impact rather than airborne — footsteps overhead, a washing machine two rooms away, a subwoofer on a shared floor — walls are the wrong target entirely. The energy is travelling through the building, and the fix is always at the source or at the contact point.

• Appliances: anti-vibration pads under the feet break the path into the floor. A washing machine on spin can be the loudest thing in a building at night, and pads cost less than a service call. Our washing machine noise guide covers what a normal level looks like.
• Floors: a dense felt pad under an area rug does far more than the rug. Half an inch of felt under a large rug is the standard renter fix for footfall complaints from below.
• Speakers and desks: anything that vibrates should sit on something soft rather than directly on the floor or a shared wall. A speaker stand on isolation pads beats an expensively treated wall.

If You Rent and Can't Build Anything

Most of the effective steps above involve screws. If that is off the table, be honest about the ceiling on what is achievable: a few decibels, plus a room that sounds calmer. That is not nothing — 3–5 dB is often the difference between noise you notice and noise you don't — but no arrangement of soft furnishings will make a neighbour inaudible.

What is worth doing, roughly in order:

• Seal the door with a removable sweep and peel-off weatherstrip, and fit outlet gaskets — both are reversible in minutes.
• Put a heavy rug on a dense pad, especially if you are the upstairs neighbour.
• Hang heavyweight curtains floor to ceiling and wider than the window. Buy on weight per panel, not on the word “soundproof” in the listing — no curtain outperforms the glass it hangs over, but a dense one takes a few dB off street noise and kills the reflections.
• Fill the shared wall with furniture. A packed bookshelf is genuine mass, costs nothing if you already own the books, and measurably outperforms foam on the same wall.
• Raise the noise floor. A white noise machine doesn't reduce anything, but it stops intermittent noise standing out against silence, which is usually what actually wakes you. Keep it under 50 dB at the pillow.

Reading the Numbers: STC and What dB Reductions Feel Like

Sound Transmission Class, defined by ASTM E413, condenses an assembly's performance across speech frequencies into one number. It is useful for comparing products and useless as a promise: lab ratings assume perfect installation and no flanking paths, and real walls have both problems. Treat published STC as a ceiling, not an expectation.

• STC 33 — a bare single-layer stud wall. Normal speech is intelligible through it.
• STC 40 — loud speech audible but not intelligible. Roughly what insulation plus careful sealing gets you.
• STC 50 — typical code minimum between dwellings. Loud speech is a faint murmur; bass still gets through.
• STC 60+ — decoupled, insulated, damped double drywall. A television at normal volume is inaudible next door.

The other number worth internalising is how decibel reductions map onto perception. A 3 dB drop halves the sound energy but is only just noticeable. A 6 dB drop is clearly noticeable. A 10 dB drop is perceived as about half as loud. A 20 dB drop is the difference between a problem and a non-issue. Seen this way, the 5 dB from a couple of hours of sealing is a real fraction of the journey — and the reason to measure rather than to guess.

If the noise is loud enough that you are wondering about hearing damage rather than annoyance, the relevant thresholds are in our guide to how many decibels is too loud.

What Doesn't Work

An unusual amount of money gets spent on things that cannot work, because the physics of blocking sound is unintuitive and the listings are written to exploit that.

• Egg cartons. The canonical example. No mass, no meaningful absorption, and a fire hazard on a wall.
• Foam on a shared wall. Covered above. It changes how your room sounds, not what gets through it.
• “Soundproof” paint. A coat of anything adds a fraction of a pound per square foot. The mechanism requires pounds.
• Thin foam or rubber sheet sold as a barrier. If it weighs noticeably less than 1 lb per square foot, it is not a barrier regardless of what the listing calls it. Check the shipping weight against the coverage area.
• Treating one wall and ignoring the door. The most common way to do real work and see no result. Sound goes around, and the weakest path sets the outcome.

The uncomfortable summary: with the exception of sealing, there is no cheap soundproofing. There is only correctly-ordered spending. Do the free and cheap steps first, measure, and let the measurement decide whether the expensive step is still necessary.

Verify the Result

Repeat the baseline measurement: same spot, same distance, same source, ideally same time of day. The comparison only means something if the source is repeatable, so if you can, play a fixed sound — pink noise or a track at a set volume — from the far side and measure that instead of waiting for the neighbour.

Our free online decibel meter is accurate enough to show a 3 dB change, which is all this task requires. For numbers you would put in a letter to a landlord or a noise complaint, use a hardware meter — see online vs. hardware decibel meters for where the line sits, and our sound level meter guide for specific models. A calibrated measurement microphone is the middle option: it pulls phone and browser readings to roughly ±2 dB for about $25.

One caution when measuring quiet rooms: below about 35 dB you are approaching the noise floor of most phone microphones, and the reading stops falling even though the room keeps getting quieter. If your before and after both read around 30 dB, that is a limitation of the microphone rather than a failed project — measure during the noise, not during the calm.

Frequently Asked Questions