Fixing Latency & Microphone Bleed in Post

Question

Is there a post-production fix for this?

I’m editing a multitrack podcast. There are three channels, each with one speaker. Say Speaker A Laughs at 0:15 while speaker B is talking. The microphone bleed for Speaker A laughing in speaker B’s mic is heard at 0:18 and I’m not able to edit it out because Speaker B is talking and editing out Speaker A’s laughing would also effectively edit out Speaker B’s talking.

I’ve tried a noise gate but no avail. I just ends up messing with Speaker B’s voice as well.

Any insight on this would be much appreciated.

Answer 1

i suppose you could copy the laugh on "mic A", then invert the signal, paste the inverted signal "below mic B's" recording and maybe shift it forward a few milliseconds until it filters out the laugh. this may end up taking away some signal but probably will just add some unwanted laugh back also... its kinda really hard to fix that after recording...

or or just mute "mic B" when he laughs and just live with the small bleed through on "mic A", at least then the laugh will have much less energy and "mic A" will be more audible. lesson learned get gates and/or shotgun mics or headsets...

Answer 2

I don't have the reputation to put these questions in a comment where they belong, so here goes:

Are all the people speaking in the same room? Where is the sound being produced for the mics to pick it up? Are the people wearing headphones? Or listening on desktop or laptop speakers? I'm not sure if you are asking a question about latency or bleed. In your example, you say that you hear the laugh later at an unwanted time. 3 seconds is a long latency, that won't sound like an echo or delay it will just sound like a repeat. (I get that this is what you are trying to correct, but I'm trying to get a handle on exactly what the problem is.)

If you have a delay like this, I think that is your primary problem, not the fact that it is bleeding into mics. A long delay makes makes conversation nearly impossible. To give you hints on how to correct latency, we will need to know more about your recording set up. Are these skype calls? Google Hangouts?

There are noise isolating headphones. I use Direct Sound EX-29 Dynamic Closed Headphones in my home studio when having a singer record vocals over tracked instruments. They cost $129.

There is a plug-in from iZotope called De-Bleed. It's part of the RX noise removal tool set they have created. From the product description, it sounds like it was BUILT for your problem. But at $400, it might be overkill!

Answer 3

That sounds like a case for adaptive echo cancellation. This relies on adaptively modeling the response from loudspeaker to microphone and subtracting the expected echo from the signal. The residue signal after subtraction is then used for slight adaptive correction of the echo estimate. Parts not correlated to the echo source lead to "overcorrection" but cancel out over time.

This requires the response function to be reasonably constant. For a microphone on a stand (rather than held in hand) and a talker not moving around too much in the path between loudspeaker and microphone, this tends to deliver workable results.

Somewhat funnily, results will be significantly better when speaker A has low-quality recording equipment producing significant noise since the noise will give the cancellation algorithm material to work with.