Podcast Mic Bleed and How to Stop It in a Shared Room

Two people at one table, two microphones, and every word from your co-host shows up a fraction of a second late in your own track. That podcast mic bleed makes voices sound hollow and phasey, and it gets worse the moment you add compression or try to edit one speaker without the other.

I have produced shows in kitchens, spare bedrooms and a converted garage, and bleed was the first problem in every one of them. The good news is that most of it is fixed before you ever press record, with choices about distance, angle and gear that cost little or nothing.

Why podcast mic bleed sounds so bad

Sound travels about 34 centimetres per millisecond. If your co-host sits 1.5 metres from your mic, their voice arrives there roughly 4 to 5 milliseconds after it hits their own mic. When the two tracks are mixed, the delayed copy partly cancels and partly reinforces the original at different frequencies. That is comb filtering, and it is why bleed makes voices sound like they are coming down a cardboard tube.

A little bleed is normal and harmless. The trouble starts when the leaked voice is only 10 to 15 dB quieter than the direct voice. Aim for a gap of at least 20 dB between each host's own voice and the bleed from the other person, and most of the hollow sound disappears.

Fix the setup before the software

Get closer to the mic

The inverse square law is your best friend here. Halve the distance between your mouth and the mic and your voice gets roughly 6 dB louder relative to everything else in the room. Moving from 30 cm to 8 cm gives you more than 10 dB of separation for free. A broadcast style position of about two to three finger widths from the grille is a solid starting point.

Point the dead side at your co-host

A cardioid mic rejects sound arriving from directly behind it. Seat hosts across the table from each other, or at a 90 degree angle, and aim each mic so the back faces the other person. Side by side seating is the worst layout because each mic hears the other voice from its most sensitive side.

Choose a dynamic mic

For shared rooms, dynamic mics like the Shure SM58, Shure MV7X or Rode PodMic are hard to beat. They need more gain, but they pick up far less of the room than a condenser. A large diaphragm condenser in a small untreated room hears everything, including the other host, the fridge and the street.

Keep gain modest

Turning up gain raises the bleed just as much as your voice. Set levels so normal speech peaks around minus 12 dBFS, and let people move in rather than cranking the preamp. Our guide on setting microphone gain before a guest arrives walks through the full level check.

A 10-minute room check before each recording

Seat both hosts and record 30 seconds where only Host A talks.
Repeat with only Host B talking.
Open the files in your editor and look at Host B's track during Host A's section. Measure the peak level of the leaked voice.
Compare it to Host B's own speaking peaks. If the gap is under 20 dB, move mics closer, change angles or add distance between chairs.
Repeat once. Most rooms reach a usable gap within two tries.

Soft furnishings help too. Bleed in a reflective room arrives twice: once directly and again off the walls. The tips in our article on taming a boomy recording room cut that second path.

Cleaning up bleed in the edit

If the session is already recorded, you still have options.

Manual cuts: Lower or mute each track when that person is not talking. Slow, but the cleanest result for short shows. On a 45 minute episode expect about 30 to 40 minutes of work.
Expander rather than gate: An expander set to reduce the quiet parts by 10 to 15 dB sounds more natural than a hard gate that slams to silence. Use a ratio around 1:2 to 1:4 and a release near 150 ms.
Automixers: Multitrack processing services such as Auphonic, and the automatic leveling in some podcast editors, balance the tracks by who is speaking. They work best when bleed is already moderate.
Align the tracks: If you must keep both tracks open during crosstalk, nudging one track by the measured delay (4 to 5 ms in our example) can reduce comb filtering. Do this by ear and check in mono.

The counter-take: do not reach for the noise gate first

The standard advice online is "just put a gate on each mic." I push back on that for conversational shows. Gates chop the start of quiet words, clip breaths, and create a pumping effect where the room appears and vanishes as each person talks. Listeners notice it within a minute even if they cannot name it.

The bigger problem is that a gate hides a bad setup rather than fixing it. Bleed still sits under every sentence while the gate is open, and that is exactly when it matters. Fix distance and angle first, then use gentle expansion only if needed. On a well arranged table, many shows need no gating at all.

When to give up on one room

Sometimes the room is simply too small. If two hosts cannot sit more than a metre apart and the walls are bare plaster, recording each person in a separate room, or one person remotely, often gives cleaner audio than any plugin. The workflow in our guide to recording a remote podcast guest works for a co-host down the hall as well as one across the country.

A quick example from a real session

Last spring I recorded a two host show in a 3 by 4 metre spare room with a hard floor. On the first test, each Rode PodMic sat about 20 cm from its host, hosts side by side, and the bleed measured only 11 dB below the direct voice. The result sounded like a tunnel.

We made three changes. The hosts moved to opposite corners of a small table at roughly 90 degrees, each mic came in to about 7 cm, and we threw a duvet over the bare wall behind them. The second test showed 23 dB of separation. Total cost was nothing, and the edit afterward took about half the usual time because we barely had to touch the crosstalk sections.

The lesson is that small, physical changes add up. Each one bought only a few decibels, but together they turned an unusable room into a fine one.

Frequently asked questions

Is some mic bleed normal on a podcast?

Yes. Almost every in-person recording with more than one mic has some bleed. It becomes a problem when the leaked voice is loud enough to cause a hollow or echoey sound, usually when it sits less than 20 dB below the direct voice.

Do dynamic mics stop bleed completely?

No, but they reduce it a lot compared with condensers. Their lower sensitivity and tighter pickup mean the other voice arrives quieter. Close placement still matters more than mic type.

Should I record podcast hosts on separate tracks?

Always, when you can. Separate tracks let you level each voice, cut coughs and reduce bleed in the edit. A single stereo mix locks every problem in place.