Frequently Asked Questions

That is exactly what the sound check separates. It scores three things independently: how much the room is colouring your voice, how far your speech sits above the background noise, and whether your recording level is usable. Rough audio almost always comes from one of those three, and people consistently misattribute the first one to the microphone.

Measure first. Your voice reaches the mic twice — once directly, then again milliseconds later off the walls, the window, the desk. A more sensitive microphone in the same room captures those reflections in more detail, not less, which is why the upgrade so often disappoints. If your reverberation time is above roughly 0.4 seconds, a rug and a few panels will do more for your sound than any microphone in that price range.

That is the most common story we hear and it is not your fault. Thin foam absorbs well at high frequencies and very little below roughly 500 Hz, which is where a voice picks up that boxy, closed-in quality. Thickness and the air gap behind a panel matter far more than how much wall you cover. Our absorption coefficient tables show the difference plainly, and the panel calculator will tell you outright when a product cannot reach your target at any quantity.

Use them — on a recording that already exists they are the only option, and they are good. But they are a repair you pay for every episode rather than a fix you make once, they leave audible artifacts on heavily reverberant material, and they cannot rescue a guest who recorded in a worse room than yours. We do not de-reverb, and we are not trying to: treating the room is the thing that stops you needing to.

Not to the recording, no — but you can to the next one. Send them the sound check before you record. It takes ten seconds, needs nothing installed, and the fix on the other end is usually a duvet and a corner of a room rather than a purchase. A remote interview is two rooms, and the worse one sets the quality of the episode.

Because we are measuring time, not level. Reverberation time is how long sound takes to decay, and that decay is the same shape whatever microphone hears it — so it does not need the calibrated response a frequency measurement would. That is the trade that makes this possible with the gear already in front of you, and it is also why we publish the error bars instead of claiming lab accuracy.

No. Reverberation is normally measured by popping a balloon or playing a sweep and recording the result. We recover it blind instead, from the decay in the short gaps between ordinary words, using a maximum-likelihood estimator published by Ratnam and colleagues in the Journal of the Acoustical Society of America in 2003. You read four sentences out loud, sitting where you normally record, and that is the whole procedure.

Against speech mixed with rooms of known reverberation time it lands within about 0.03–0.07 seconds on a clean recording. Two limits are worth knowing. It cannot resolve below roughly 0.2 seconds, so a well-treated booth comes back as “under 0.25s” rather than a precise figure. And background noise hides the tail of the decay, which pushes the estimate up — so when your noise floor is high we tell you the number is probably an overestimate and mark the result low confidence. Treat the band (dry, controlled, live, very live) as the reliable output and the exact figure as an indication.

No. The audio is analysed in your browser and never leaves your device. There is no account, no upload endpoint and nothing stored server-side. You can disconnect from the network and the sound check still works, which is the only version of that claim worth making.

Not today. Everything on this site runs in a browser tab, on a phone or a laptop, with nothing to install. If a native app ships we will say so here — and until it does, nothing on this site will imply otherwise.

Any current version of Chrome, Edge, Firefox or Safari, on desktop or mobile. The tools need Web Audio and, for the ones that measure, permission to use your microphone — which the browser asks for and which you can revoke at any time.

For spoken word, around 0.3 seconds or less. Broadcast voice booths sit nearer 0.2. An untreated spare bedroom usually lands between 0.4 and 0.6, and a room with hard floors and large windows can pass 0.8 — which is where voices start to sound like they were recorded down a hallway. You do not need a booth. You need to get from where you are to about 0.3.

Something soft, large, and behind or beside the microphone — a duvet on a clothes rail, a thick blanket over a door, a full bookshelf. A rug on a hard floor is usually the single biggest surface you can treat for the money. Try what you already own tonight and re-measure; it costs nothing and it tells you how far the free moves get you before you buy a single panel.

It depends on the room, the target and the panel, which is what the panel calculator is for — it uses the reverberation figure you measured and shows what fewer panels would get you as well as what reaches the target. As a rough guide, a spare bedroom in the 0.5 second range usually wants four to six thick panels at the first reflection points rather than a wall covered in thin ones.

Most of what works here does not need a drill. Rugs, heavy curtains, a full bookshelf, upholstered furniture, a duvet on a rail behind the mic and freestanding panels on stands all move out with you. Command strips hold light panels on most wall finishes. The physics does not care whether the absorption is screwed to the wall.

First reflection points first — the spots on the side walls, the ceiling and the wall behind the mic where sound bounces once on its way from your mouth to the microphone. Behind the mic matters more than most people expect, because that surface is directly in the pattern of a cardioid. Spreading absorption around beats concentrating it in one place.

Measure again — it is ten seconds and it will tell you whether the number moved. If the room score improved but the recording still sounds rough, the problem was probably never the room: check the noise sub-score for a fridge or a computer fan, and the level sub-score for something recorded too quietly and then turned up. Fixing one thing routinely reveals the next one under it.

The tools are free, there is no account and nothing is metered. We earn a commission if you buy a product through one of our links, and we say so where it applies. That is the whole model. It is also why the calculators show you what a cheaper option gets you rather than only the configuration that hits the target.

No, and we would rather say so than take the traffic. Stopping sound getting in or out of a room is soundproofing — mass, decoupling and sealed gaps — and it is a construction problem you largely cannot solve in a room you rent. Everything here works on how your voice sounds inside the room. If noise is getting in, the noise meter will confirm it, and then we are the wrong site.

Not properly yet, and we would rather not pretend. A listening room's problem is bass — room modes and modal ringing below about 200 Hz — and speech carries almost no energy down there, so the measurement on this site cannot describe it. Doing that honestly needs a swept-sine measurement through your own speakers, which is the next thing we want to build. The room calculator already finds the modes for a room's dimensions if that is what you are after today.