The take beats the gear, and the room beats both
I master my own records on Apple Silicon. I've written about the mix stage and the master stage on this blog before. But none of that matters if the capture is garbage. You cannot polish a vocal that was recorded three feet off-axis in a bathroom with the gain slammed into the red. Mastering is subtractive-friendly and additive-forgiving only up to a point — feed it a clipped, roomy, sibilant take and the best chain in the world just makes the problems louder.
So this one is about the front of the line: how I get a dry vocal at home that my mix and master chain can actually work with. No studio, no six-hundred-dollar rack. A bedroom, a corner, and a signal path that respects headroom.
The actual signal chain (it's shorter than you think)
Four links, in order: mic, interface, monitoring, and the room. That's it.
Mic. You have two honest paths. A large-diaphragm condenser (an Audio-Technica AT2020 or AT2035, a Rode NT1) gives you detail and air — great in a treated space, brutal in an untreated one because it hears everything, including the fridge and the reflection off your window. Or a dynamic like the Shure SM7B, which is the podcast/vocal workhorse for a reason: it rejects the room. It has a tight cardioid pattern and low output, so it hears you and mostly ignores the wall behind you. If your room is bad and you can't fully fix it, the dynamic is the smarter buy. The catch: the SM7B is quiet and needs a lot of clean gain — either an interface with strong preamps or an inline booster like a Cloudlifter or FetHead to add ~+25 dB before the take even hits your interface.
Interface. You need clean gain, not fancy gain. A Focusrite Scarlett, a Universal Audio Volt, a MOTU M2 — any of them at the $150–250 range gives you converters and preamps that are transparent enough that the master will never expose them. What you're paying for is a low noise floor so that when you push the level up for a quiet dynamic mic, you're raising the voice and not a wash of hiss.
Monitoring. Closed-back headphones, always, when a mic is open. Open-back cans and studio monitors bleed into the mic and you'll capture your own playback as a ghost under the vocal. Something like the Audio-Technica ATH-M40x or M50x. Flat-ish, sealed, done.
Pop filter. Non-negotiable and it costs eight dollars. Plosives — the "p" and "b" bursts — hit the diaphragm as a blast of air and produce a low-end thump that no de-esser or EQ move fully repairs. A nylon or metal pop filter two to three inches off the mic kills them at the source.
Why the room beats the mic
Here's the thing nobody selling you a mic wants to say: a $100 mic in a treated corner beats a $700 mic in a bare bedroom. Every time. The reason is reflections. Sound leaves your mouth, hits the hard walls, and comes back into the mic a few milliseconds later as a smeared copy of itself. That's the "boxy," "hollow," "I recorded this in a garage" sound, and it is baked into the file. You cannot EQ out a reflection. It's a timing problem, not a tone problem.
So you kill reflections before they happen. You don't need a certified acoustic vault. You need soft, thick things on the hard surfaces around where you sing:
- Moving blankets or thick comforters hung on the walls behind and beside you.
- A closet — this is the real cheat code. Walk into a closet full of hanging clothes and clap. It's dead. Clothes are a broadband absorber. I've tracked full verses standing in a closet with the laptop on a shelf, and it beats any "vocal booth" I could build for the money.
- If you want to spend a little, actual acoustic panels (Owens Corning 703 in a frame, or off-the-shelf foam as a last resort — foam only handles highs) go on the first-reflection points.
The test costs nothing: clap your hands in the space. If you hear a little zippy ring after the clap, that ring is going into your vocal. Add soft material until the clap sounds flat and short.
Gain staging: leave the headroom
This is where most home takes die. People turn the gain up until the meter looks "loud" and healthy, hitting the red on the loud lines. Do not do that. Digital has a hard ceiling at 0 dBFS and there is nothing above it — a peak that touches 0 clips, and a clipped transient is unrecoverable distortion that follows the file all the way to the master.
Set your gain so your loudest sung line peaks around -12 to -6 dBFS. That leaves 6 to 12 dB of headroom above your hottest moment. It will look quiet on the meter. That's correct. Loudness is the master's job, not the capture's job — I'd rather bring a clean, conservative take up in the mix than try to un-break a hot one. Track at 24-bit and this is free: 24-bit gives you so much dynamic range that a vocal sitting down at -12 still has vastly more resolution than the noise floor. The old "record hot to beat the noise" advice is a 16-bit relic. On 24-bit, headroom is pure upside.
Do a loud test pass first — belt the biggest line in the song — set the gain there, then never touch it for the rest of the session so your takes stay consistent.
Position: close, on-axis, off the plosive line
Get the vocalist close — six to eight inches on a condenser, closer on the SM7B. Close means the direct voice dominates the room, which is the same fight as the reflections, won from the other side. On a dynamic you can work in tight and let proximity effect add chest and warmth.
On-axis means singing straight into the front of the capsule, not up at it or across it — off-axis costs you high end and consistency. And angle the mic slightly so you're not firing plosives dead into the center; singing just past the grille, with the pop filter in the path, is the belt-and-suspenders move.
Then capture it dry. No reverb, no compression printed to the recording, no "vocal preset" from the interface software committed to the file. Record the naked signal. Reverb and compression are decisions I want to make later, in the mix, where they're reversible. A dry, close, on-axis, properly-gained take is exactly the raw material a mastering chain is built to elevate — and the one thing you can never add back after the fact.
FAQ
Do I really need an SM7B, or is that just hype?
You need a good mic, and the SM7B is a genuinely great one — but it's a room-rejecting dynamic, so its real value shows up in untreated spaces. If you've properly deadened your corner, a $150 condenser will give you more detail for less money. Buy the SM7B if your room is bad and staying bad; buy the condenser if you've done the blanket-and-closet work. Either way, budget for the Cloudlifter if you go SM7B — it's quiet without one.
What sample rate and bit depth should I record vocals at?
24-bit, always — that's what buys you the headroom to track conservatively at -12 dBFS without noise. 44.1 kHz or 48 kHz sample rate is plenty for vocals; higher rates mostly just eat disk and CPU for no audible gain on a voice. The bit depth is the one that actually protects your take.
How do I know if my room is bad?
Stand where you'll sing and clap once, hard. If the clap is followed by a short zippy ring or a boxy "room" tail, that coloration is being recorded into every vocal. Keep adding soft, thick material — blankets, clothes, panels — until the clap sounds dead and instantaneous. A closet full of clothes usually passes this test on day one.
Can't I just fix a bad recording in the mix or master?
Only partly, and never the things that matter most. Clipping is permanent distortion, room reflections are a timing smear that EQ can't undo, and plosive thumps are baked-in low-end blasts. Mixing and mastering shape and elevate a good signal — they can't reconstruct information that was destroyed at capture. Protect the take and the rest of the chain has something to work with.