Dubbing is often pictured as a matter of padded booths, thousand-button consoles and five-figure budgets. Small projects work differently: the first voice of an indie game or of an audio story episode is almost always recorded in a bedroom, on a Sunday afternoon, with three objects sitting on a desk. What separates a usable take from an unlistenable one rarely comes down to the price of the microphone.
We have gone through this several times for our own projects, and the same conclusion comes back every session: a voice over setup is sorted out in one evening, the method takes a little longer. Here are both, in the order that actually matters.
The room matters more than the microphone
A three-hundred-euro microphone in an empty living room will sound worse than a ninety-euro one in a furnished bedroom. The microphone picks up the voice and everything the voice does in the room: reflections off bare walls, tiled floors, a large window. Those reflections arrive a few milliseconds after the direct sound and leave that bathroom quality behind, which no amount of editing really repairs.
The fix costs close to nothing. A rug, heavy curtains, a full bookshelf and a bed already absorb most of it. Plenty of voice over artists record inside a wardrobe, surrounded by clothes: unglamorous and extremely effective. If the room stays reverberant, a duvet hung behind the microphone or a portable acoustic shield on the desk changes the result from the very first take.
Silence counts as much as absorption. Fridge, ventilation, laptop fan, notifications, a neighbour mowing the lawn – recording at ten in the evening in a quiet room beats recording at three in the afternoon with an excellent microphone and a boiler firing up in the middle of the best take.
The minimum gear, in order of importance
The microphone
Two families share the work. The dynamic microphone, the one used on stage and in radio, mostly hears what is placed in front of it: it forgives an imperfect room and asks for more gain in return, which means a decent preamp. The condenser microphone captures more detail – breath, saliva, inhalations – and the room along with it. In an ordinary flat, the dynamic remains the safer choice; in a treated room, the condenser gives a more open voice.
Then there is the connection. A USB microphone plugs in and works, with no interface and no extra cable, which is enough to record a whole season of audio stories. An XLR microphone needs an audio interface, costs more up front, and lasts ten years while everything around it changes. If the project is meant to last, the second option pays for itself quickly.
The headphones
Closed-back, never open-back: open models leak the monitor mix into the microphone, and that leak ends up in the file. Headphones are there to monitor the take as it happens, not to judge the final mix; an entry-level monitoring model does the job.
The accessories that genuinely improve a take
A pop filter, placed four or five centimetres from the capsule, stops plosives – those bursts of air on P and B sounds that overload the input and cannot be repaired afterwards. A stand or a boom arm keeps keystrokes and desk knocks from travelling into the microphone. Both cost about as much as a meal and improve a take far more reliably than moving up a price bracket.
What can safely wait: the high-end preamp, the booth, full acoustic treatment of the room, the second microphone. None of it shows until the method is in place.
Prepare, record, clean up
Before the session
Read the script from a tablet or a screen rather than paper, because rustling paper is audible. Mark the breaths in advance, and note for every line who the character is talking to, what state they are in and what just happened: those three pieces of information are always missing at the moment of recording.
Placement is a matter of centimetres. About fifteen centimetres between mouth and microphone, the capsule at mouth height or slightly above, the face a little off-axis so plosives pass beside it. Too close and the voice swells in the low end through proximity effect; too far and the room comes back into the take.
As for levels, aim for peaks between -12 and -6 dBFS. The headroom looks generous: it exists for the character's first scream.
During the take
Record continuously and run three takes of each line without stopping. Mark a failed one by clapping or simply saying « pick-up »: the waveform is then spotted at a glance during editing.
Thirty seconds of room silence, captured at the start of every session, are worth their weight in gold. That room tone serves as a profile for noise reduction and as a patch for the gaps between lines, where absolute digital silence sounds like a hole.
Efforts – screams, breaths, falls, pain – belong at the end of the session: they tire the voice and ruin anything recorded afterwards. For a game they often add up to more files than the dialogue itself.
Finally, write down the microphone position, or photograph it. A pick-up recorded three weeks later at a different distance will not match.
The minimum clean-up
Four moves cover the vast majority of cases: a high-pass filter around 80 Hz to remove rumble, light noise reduction based on the room tone, gentle compression – roughly 3:1, three to six decibels of reduction – to hold the level swings, then a touch of equalisation. A de-esser joins in if the sibilants sting.
Output level depends on the destination: around -16 LUFS integrated for an episode meant for listening, and peak-normalised files near -3 dBFS for a game, where the engine handles the final mix. Audacity is plenty to start with, Reaper costs little and does the rest.
Act, do not read
The rest is acting, and that is the part that improves fastest. Recording while standing opens the ribcage and frees the gesture; the body is audible in the voice. Keeping a fixed point to look at and addressing the line to someone, even an imaginary someone, pushes the reading tone away.
A game line is performed blind, without the image and without the other characters' lines. Hence the value of recording two or three intentions of the same sentence while you are there: editing will decide, and a pick-up session always costs more. The subject reaches beyond recording, and we covered the writing side in writing an audio story, where the format sets its own constraints. The texts we record often come from our own books, published under the name Allison Lister.
Delivering clean files
Explicit naming saves hours: character, scene, line number, version, along the lines of ELIA_02_014_v01.wav. WAV at 48 kHz and 24 bits is the standard in video and games; using it everywhere avoids pointless conversions. MP3 belongs at the very end of the chain, for distribution, never during the work.
For the web, a plain HTML5 audio player is enough to publish an episode without depending on a platform. For a game, plan for subtitle space and line length from the start: a sentence that outruns its animation is noticed immediately.
Start small
A voice over rig comes down to very little: a decent microphone, closed headphones, a pop filter, a furnished room and a method you stick to: enough to record the voices of a prototype, a demo or a first season. The rest – the booth, the analogue chain, the professional actor – arrives when the project deserves it. In the meantime, the one real mistake would be postponing the recording while waiting for better gear.
0 Comments