What mic bleed actually is
Record two people in the same room and every open microphone picks up both voices — just the "wrong" one at a lower level and slightly delayed. Mixed together, that delayed copy combs with the main signal and creates a thin, hollow, slightly phasey quality that's easy to hear even when you can't quite place what's wrong.
Why it's so hard to remove afterwards
Steady background noise — hiss, hum, a fan — sits in a predictable, learnable frequency range that a processor can subtract cleanly. Bleed doesn't. It's another human voice, occupying the exact same frequency range as the voice you want to keep, arriving at a slightly different time. There's no clean fingerprint to remove; anything aggressive enough to touch it will damage the primary voice too.
Fixing it at the source
- Increase the distance between speakers, or between each mic and the other person, however small the room allows.
- Use directional (cardioid) mics angled away from each other, so each capsule's rejection zone points at the other speaker.
- Have quieter mics off or muted when someone isn't actively speaking, rather than leaving every channel hot the whole session.
- Record remotely where possible — separate rooms or separate calls give every speaker a genuinely isolated track, which is the only complete fix.
If it's already recorded
Light bleed in a well-treated room is usually less noticeable than people fear once the episode is balanced and mastered — chasing it out aggressively tends to introduce worse artefacts than leaving it alone. Reserve heavy processing for genuinely bad cases, and treat this one as a lesson for the next session rather than a fight to win in post.
What processing can still do
Optivox won't undo bleed that's baked into a shared-room recording — nothing reliably does — but its noise reduction and speaker balancing still smooth out the rest of the episode around it, and a proper master keeps the whole thing sounding consistent regardless.