Yes, and the thing that decides whether it works is outside the house rather than inside it. A guitar and a voice recorded together in a Sherman Oaks living room comes down to three things: how the two microphones are angled so each one rejects the other source, what the noise floor of the street is at the hour you book, and whether you have accepted that neither part can be fixed later without the other moving with it. The room matters less than people expect. Sherman Oaks sits in the middle of the Ventura Freeway corridor where the 101 and the 405 meet, and that is the real variable.
Why record them together at all?
Because the timing between a voice and the hand playing under it is the performance, and overdubbing quietly destroys it.
A singer accompanying themselves pushes and pulls against their own playing in units of 10 to 30 milliseconds. Play the guitar first and sing to it, and the singer follows a fixed grid instead of leading one. The take gets cleaner and flatter at once. For a settled singer-songwriter arrangement the live version is almost always the better record.
The price is worth stating plainly. You cannot retune one vocal line without the guitar bleed moving with it, and you cannot replace the second chorus guitar. Comping is by whole passes rather than by phrase. If the arrangement is still being written, overdub instead.
What does the bleed actually do to the sound?
It creates comb filtering, and the amount is set by geometry rather than by level.
The guitar arrives at the vocal microphone a fraction of a millisecond after it arrives at the guitar microphone, roughly 3 ms per metre of path difference. Two versions of the same source, slightly offset, sum into a series of notches. The audible result is a hollow quality in the guitar that gets worse when you raise the vocal fader, which is why the problem usually gets diagnosed as a bad guitar sound.
Two rules do most of the work. Keep each microphone at least three times further from the source it is not meant to capture than from the source it is, the standard 3:1 spacing, which puts the bleed roughly 10 dB down. And check the whole thing in mono before recording a note, because comb filtering that is inaudible in stereo is obvious the moment the two channels are summed.
What is the setup that works in an untreated room?
Two figure-of-eight microphones, angled so each null points at the other source.
A ribbon or a large-diaphragm condenser switched to figure-of-eight on the voice, rotated so one of its side nulls, 90 degrees off axis, faces the guitar's soundboard. A second figure-of-eight on the guitar, pointed at the join of neck and body around the 12th to 14th fret, its side null aimed up at the singer's mouth. Note that the rear of a figure-of-eight is not a null: it is fully sensitive with inverted polarity, and aiming it at the other source does nothing. The rejection off the sides is far deeper than the rear rejection of a cardioid, and that is 10 to 15 dB of separation from geometry alone before you have spent anything.
Distances: 20 to 25 cm on the guitar, 20 to 30 cm on the voice. Closer on the guitar and every hand movement changes the tone; further and the untreated room arrives.
The side benefit matters in a house. A figure-of-eight microphone hears nothing from directly above or below, so a ceiling reflection in a room with an 8 foot ceiling, the single worst feature of most Valley houses built after the war, largely disappears from the recording.
What does Sherman Oaks specifically do to the noise floor?
It puts a freeway, a leaf blower ordinance and a hillside into the same conversation.
Sherman Oaks is a neighbourhood of the city of Los Angeles in the San Fernando Valley, not a separate city, which matters the moment you start reading local rules: the ones that apply are the Los Angeles ones, unlike in Burbank or Glendale next door. Its main commercial spine is Ventura Boulevard, and the 101 runs the length of it.
Three practical consequences.
Freeway rumble is broadband and constant. Houses in the flats north of the boulevard sit closer to the 101 and pick up a floor that a high-pass does not remove, because it is not only low frequency. Houses in the hills south of it usually have a lower floor and worse rooms: hillside construction means more glass and more parallel hard surfaces.
Gardening equipment is legally quieter than people assume, and it is worth knowing. Under Los Angeles Municipal Code section 112.04, gas-powered leaf blowers cannot be used within 500 feet of a residence at any time, and the person who hired the gardener is liable as well as the gardener. The same section bans mowers, edgers and similar equipment in residential zones between 10 pm and 7 am. Knowing the rule does not stop a neighbour's crew from showing up, but it turns an argument into a two-minute conversation.
Pool equipment runs on a timer. Almost every house on a Valley street has a pump somewhere, and it is the noise source people never check because they have stopped hearing it. Find the timer before the session.
The practical booking answer: a weekday between 10 am and 2 pm, after the gardening crews and before school traffic, with a two-minute silent recording made at the start so the actual floor is known instead of guessed.
When should you drive over the hill instead?
When the record needs a controlled low end, or when the schedule cannot absorb one bad afternoon.
A house take is right for a song that lives on the performance. It is the wrong choice when the arrangement will get bass and drums later, because an untreated room bakes in a low-mid signature around 150 to 300 Hz that fights everything added afterwards, and no amount of subtractive equalisation puts that back.
Sherman Oaks to my room in Burbank is about nine miles, east on the 101 and then the 134, roughly twenty minutes outside peak hours and closer to forty at 5 pm. That drive buys known monitoring and a floor that does not move when a truck goes past. A workable split for people who want both: track the live takes in the house on a quiet weekday morning, then bring the session across for editing, tuning and mix in a room where the decisions can be trusted.
What settings, and what gets recorded?
48 kHz, 24 bit, peaks around -12 dBFS, and a room tone track nobody skips.
Set gain on the loudest thing the singer actually does rather than on a soundcheck strum, then leave it. Record every pass from the top including the false starts. Record two minutes of silence at the beginning and again at the end, because a noise floor that shifts between the first take and the last is what makes a comped vocal impossible to hide.
One more, the most-skipped item on any list like this: put every phone in the room into aeroplane mode, and turn the refrigerator off at the breaker if the kitchen shares a wall.
Frequently asked
What if the room has a hard tile floor? Put anything soft on the floor between the performer and the nearest wall, and get the microphones closer. A rug and a duvet on a stand solve more than acoustic foam does, because the problem is early reflections rather than absorption in general.
Can I fix bleed later with a plugin? Source separation is good enough to rescue a take and not good enough to build a record on. Every separated stem loses transient detail on the guitar, which is what makes a live take feel live.
Is a live take actually better than overdubbing? For a settled singer-songwriter arrangement, usually yes, and the reason is timing rather than tone.
If you have a room and a song
Tell me the street, the ceiling height and roughly what time of day the house is quiet, and I will tell you whether the take is worth attempting there and which two microphones to point where. Full credit list at /credits, background and contact at /about.