Building a Voice Over Booth at Home

Search for a voice over booth and the internet will show you a beautiful, sealed, freestanding box with a window in it, and a price you have to email someone to get. Then it will show you an advert for it on every website you visit for the next three weeks. That is what a Paid Difficulty of 100 looks like from the inside: an awful lot of people who would very much like to sell you a room.

You may eventually want one. You almost certainly do not need one now. What you need is for your recordings to stop sounding like they were made in a bedroom — and it turns out that is a much cheaper problem than the one being advertised to you.

But you cannot solve it until you understand which of two completely different problems you actually have. Almost every beginner gets this wrong, and the industry is more than happy to let them.

Soundproofing and acoustic treatment are not the same thing

Soundproofing is stopping sound getting in — the lawnmower, the flight path, the neighbour's television, the lorry downshifting outside. It is a problem of mass and airtightness. You beat it with heavy, sealed, dense construction: thick walls, dead air gaps, doors that seal. It is expensive, it is structural, and foam does absolutely nothing for it. This is what a real booth is actually selling you.

Acoustic treatment is a completely different job. It is about what happens to your voice after it leaves your mouth and before the microphone hears it. In a small hard-walled room, your voice hits the walls, the ceiling, the window and the desk, and comes bouncing straight back into the mic a few milliseconds behind itself. Those reflections are what make a recording sound boxy, hollow, cupped — like it was made in a shoebox, which it was. You beat this with soft, absorbent, porous material placed where the reflections are. It is cheap. It is the thing you actually have wrong.

Nearly everyone who says they need soundproofing needs treatment. If your recordings sound boxy, more mass will not help you. If a bus goes past every four minutes, blankets will not save you — but neither will a $600 slab of foam, and you would be better off recording at a different time of day. Diagnose which problem you have before you spend a pound. They have different answers.

What is actually going wrong in your room

The mechanism is worth understanding, because once you hear it you cannot unhear it. A small untreated room concentrates its reflections in the 200Hz to 4kHz band — which is, unhelpfully, exactly where the human ear is most sensitive and exactly where the intelligible part of your voice lives. That is the boxy sound. It is not a mic problem and it cannot be EQ'd out.

Reverb also smears the detail. As acoustics people describe it: the tail of the previous syllable overlaps the start of the next one. That is why an untreated read sounds mushy and slightly amateur even when the performance is good. A well-treated room gives you a voice that is clear, free of background noise, and has an audible focus to it — and as one acoustics firm puts it bluntly, that focus is impossible to achieve in a small, untreated room, more or less regardless of which microphone you point at yourself.

And then the compounding problem: an untreated room is usually also a noisy one, and when you raise the gain in post to fix a quiet recording, you raise the noise floor with it. Your noise floor is simply the sound of the room when you are silent. An untreated bedroom typically sits around -40 to -50 dBFS. ACX and Audible want -60dB RMS or better. You do not get there with software. You get there with cloth and with turning things off.

The cheapest setup that produces broadcast-usable audio

Moving blankets on a PVC frame. That is the answer. It is not glamorous and there is no affiliate commission worth chasing on it, which is precisely why it is under-recommended. Six to eight moving blankets and a frame of PVC pipe runs about $100 to $150, it is endorsed on the studio-building forums where actual engineers argue about this, and it will get you inside the specs.

A mattress fort genuinely works too, at $50 to $400 depending on what you already own — mattresses are dense, thick and porous, which is the entire specification. If you want to buy something purpose-made, the middle ground exists: GIK's PIB Portable Vocal Screen is $499 and the PIB Vocal Booth Pro is $999.

What does not work on its own is the clothes-in-a-closet trick everybody repeats. It helps. It does not solve the room: only the very highest frequencies, the ones with the shortest wavelengths, are absorbed by hanging clothes, and the boxiness lives lower down than that. A closet is a decent place to record because it is small and full of soft mass — but the clothes are not the treatment, they are a start.

What a real booth actually costs

So that you can see the whole ladder, here is the top of it, at live prices. A WhisperRoom MDL 4230 S, single-person, is $5,545. Their Work From Home Booth, the MDL 4872 S, runs $7,618 to $8,019. The two-person, 32-square-foot MDL 4896 S is $9,350 in single-wall form, and $15,725 as the double-wall MDL 4896 E with enhanced isolation. Shipping is not included in any of those figures, and these things are heavy.

The other name you will hear is Studiobricks. I am not going to give you a single number for it, because there isn't one to give: the manufacturer has moved to quote-only pricing, and the roughly $3,500 base figure that circulates comes from third-party reviews that may well be stale. The realistic range is somewhere between $3,500 and $8,000 or more. Anyone quoting you a precise Studiobricks price on a blog is guessing.

Notice what you are actually buying at those prices: isolation. Mass and sealing. A real booth is a soundproofing product. If your problem is reflections rather than the flight path, you are spending five figures to solve a problem that blankets solve for a hundred dollars — and you will still need absorption inside the booth, because a small sealed box is, acoustically, the most boxy room you will ever stand in.

The specs, and what comes back

Whatever you build, it is judged against numbers. For ACX and Audible: RMS volume between -23dB and -18dB, peaks no higher than -3dB, and a noise floor no higher than -60dB RMS. Each file needs 1 to 5 seconds of room tone at the head and the tail — room tone being the recorded sound of your silent room, which the editor uses to patch gaps, and which is also, incidentally, the honest measurement of whether your treatment worked. Files go out as 192kbps or better CBR MP3 at 44.1kHz, mono or stereo but consistent throughout, and no longer than 120 minutes each. For broadcast and commercial delivery it is 48kHz / 24-bit WAV, mono, delivered dry with no processing at all.

One honest caveat on that -3dB peak figure, since everybody quotes it as gospel: it is a marketplace convention rather than a broadcast standard. It is an explicit Voice123 rule, so their files survive conversion. No standards body issued it. Hit it anyway, because clients ask for it — but know what it is.

And what comes back? Noise floor over threshold, first and always. Then hum, reverb, plosives, mouth noise, levels that wander between takes. And — the one that surprises everyone — over-processing. More home recordings fail quality control for being cleaned too hard than for being left too dirty. The beginner hears hiss, panics, and buries the voice under noise reduction until it sounds like it is speaking through a wall of water. The human listening on the other end hears those artefacts immediately. Gentle corrections, as few as possible. Fix the room, not the file.

Build the blanket fort. Record thirty seconds of silence and look at the number. Fix what that number tells you to fix, and fix it again next week. If you are still failing after that, then you have discovered you have a soundproofing problem rather than a treatment problem, and you will know precisely what you are buying and why. That is a much better position to be in than the one the adverts want you in — which is holding a credit card and no diagnosis.