The room does most of the work, so the first decision is where the main pair stands, not which microphone goes on it. A pair of omnis about 50 cm apart, roughly 2.5 metres up and 1.5 to 2 metres in front of the front row, gets you most of the record. Everything after that is subtraction: turn the heating off, get the back row physically higher than the front row, and record full run-throughs rather than bars. Glendale has an unusual density of hard-surfaced halls for a city its size, and the hall you pick decides more than any plugin will.
Why does microphone position matter more than microphone choice?
Because distance sets the ratio of direct sound to reverberant sound, and no fader recovers that later.
Every room has a distance at which the direct sound from the source and the reflected energy from the room are equal. Inside that distance you are recording the choir. Outside it you are recording the building. A plaster and stone hall with a mid-band decay somewhere between 1.8 and 2.5 seconds, the T30 figure that ISO 3382-1 defines for performance spaces, sounds magnificent standing in it and turns to mush on a recording made three metres too far back, because consonants arrive with the direct sound and reverberation does not carry them.
So the working method is to move the pair, listen, move it again. Thirty centimetres forward is a real change. If the recording is going to need reverb it can be added later, and added reverb you control is better than room you cannot remove.
What is the actual array, and when do you change it?
Start with spaced omnis and only change if the room forces it.
Two omnidirectional capsules 40 to 60 cm apart, at about 2.5 metres, give a wide unforced image and the low end that cardioids lose. In a hall that is too live, switch to a near-coincident cardioid pair, ORTF style at 17 cm and 110 degrees, which pulls the direct-to-reverberant ratio back without moving the stands. For a larger group across a wide platform, add a centre omni about a metre in front of and slightly above the two outriggers so the middle of the choir does not fall into a hole.
Spot microphones are for soloists, not for sections. A section spot mic close enough to be useful is close enough to make one row louder than the people beside them, which is the opposite of what a choir is for. Any spot mic gets delayed back to the main pair, about 3 milliseconds per metre of distance difference, or the image collapses toward the spot.
What has to happen in the room before anyone sings?
Four things, and only one of them involves audio equipment.
Heating and air off, and off early. Blowers take several minutes to spin down and an old system rumbles below 60 Hz for longer than that. In a church hall, the two forgotten sources are the boiler and, if there is a pipe organ, the organ blower, which is often on its own switch in another room.
Risers, or the closest thing available. Getting the back row physically above the front row is worth more than any EQ decision made afterwards. Two rows on the flat means the back row sings into the heads in front of them, and the recording is missing the voices you cannot hear.
Spacing between singers. 45 to 60 cm each way. Tighter and the blend gets muddy in a way that sounds like a bad microphone; wider and the sections stop hearing each other and the tuning goes.
A noise-floor listen with everyone silent. Two minutes with the recorder running and nobody moving. Fluorescent ballasts, a refrigerator down a corridor, traffic through a single-glazed window. Every one of those is fixable in five minutes before the session and unfixable afterwards.
Why does Glendale have so many usable halls?
Because of who built them and when.
Glendale is a separate city with its own municipal code, not a district of Los Angeles, and the same is true of Burbank next door. That distinction matters when you start calling around about evening access and parking, since the rules are not the ones you read for Los Angeles. What makes it useful for this particular job is the building stock along and around Brand Boulevard: a run of 1920s civic and religious architecture with plaster over lath, hard floors, and volume that nobody would build today. The Alex Theatre at 216 North Brand, opened in 1925 with about 1,400 seats and a 100-foot neon tower added in 1940, is the visible example of that period; the halls that are actually bookable for a session are mostly the churches and community rooms around it.
The other reason is living choral practice. About a third of Glendale's residents claim Armenian ancestry, the result of several decades of immigration to Southern California, and the churches that came with that community sing regularly rather than ceremonially. A hall used by a choir every week is a hall where the acoustics are already understood by the people who run it, and where somebody can tell you which corner rings.
Central Glendale is roughly six or seven miles from my room in Burbank, about fifteen minutes down Glenoaks Boulevard outside peak. That distance is why the split works: record in the hall, then edit, tune and mix in a small treated room where the monitoring is known.
When should the session be, and how long?
Weekday, not Sunday, and shorter than you think.
A church parking lot on a Tuesday evening is empty and on a Sunday morning is not, and load-in with stands, cables and a recorder is the part that decides whether the first take happens at 7:15 or at 8:00. Weekday daytime is quieter still if the choir can do it.
On length: two and a half to three hours including warm-up, and no more. A choir does not behave like a session player. Voices are at their best somewhere around take two or three and then decline, and the fifth take of a difficult passage is almost always worse than the second. Plan three complete run-throughs of each piece plus targeted patches, and get the patches while the voices are still fresh rather than at the end.
No headphones for the choir. They sing to each other and to the conductor, which is how the blend happens. If the recording has to lock to a click, one earpiece for the conductor and nobody else.
What gets recorded, and at what settings?
48 kHz, 24 bit, and enough headroom that a fortissimo entry has somewhere to go.
Set gain on the loudest passage in the loudest piece, not on the warm-up, then leave it. Peaks around -12 dBFS are fine and leave room for the entry that arrives 6 dB louder than rehearsal did. Choirs are dynamic in a way that surprises people who mostly record pop, and a clipped tutti is the one thing on this list that cannot be repaired.
Record continuously through the whole session, including the talking. The moment somebody sings a phrase perfectly while marking it is more common than it should be, and it is only usable if the recorder was running.
Frequently asked
Can I do this with two microphones? Yes, and for most choirs two is the right answer. A third microphone solves one specific problem, a wide group with a weak middle, and creates phase problems if the problem was not there.
What about recording in the Alex Theatre or a similar venue? A seated theatre is a different job from a church hall, because the seats are absorbers and the decay is much shorter. It is easier to control and gives you less for free.
Do we need the organ in the recording? Decide before, not after. An organ and a choir share a room and cannot be separated in the mix, so if the organ is on the record it is on every microphone from the first take.
If you have a hall and a date
Send me the hall, the size of the choir and what the programme is, and I will tell you where the pair should stand, whether the room needs a cardioid array instead, and how long the session realistically has before the voices start going backwards. Full credit list at /credits, background and contact at /about.