Eliminating mic bleed

Barry W. Hull wrote on 4/16/2013, 8:17 AM
In a two or three person interview/conversation, each person hooked up to a lav mic, I’m looking for the simplest, easiest way to eliminate or reduce mic bleed from one lav to another.

I’m not satisfied with any of the solutions I’ve found, hiring a sound mixer during the shoot (over budget), manually panning back and forth during post (a tedious time consuming nightmare every week), noise gate (either too much or not enough), spacing out the participants (awkward), etc.

I’m hoping that software exists, sort of similar to Plural Eyes, that takes synching to the next level, where it determines the original/source audio channel (loudest) and automatically mutes the other channel(s) during that time. If the participants talk over each other, then that section is not muted, and of course, the ability to manual edit any errors. Does anyone know if that software exists? If it doesn’t exist, why don’t one of you audiophile/programmers write it? Might sell like hotcakes, I’ll buy it today.

If no automated solution exists, any other “easy/simple” suggestions to prevent the mic bleed effect? When I mute the non-source channel, the audio sounds terrific, clear and crisp.

If all else fails, we might try a single shotgun mic on an overhead boom. Any suggestions on make and model?



musicvid10 wrote on 4/16/2013, 8:23 AM
You are talking about source isolation. The only way to do that is in a recording studio.
Hypercardioid lapel mics do not work because the volume shifts every time someone turns their head. Shotguns are generally horrible at picking up reflections and air.

You haven't said why you want this. Is the room noisy? Are the participants noisy? You really need to correct the problems on that level first. It's easy enough to open the mixer in post and ride gains as you would for any other program.

Broadcast interviews us omni mics allowing bleedover, but panned tastefully to fit the mix. The other plus is that if one mic goes out, a nearby mic will pick up the talent. Your interviewees need to learn not to smack their lips or clear their throats during the interview.

Maximum isolation in noisy environments is achieved with something like the Countryman E6 Directional mics. Some reasonably priced knockoffs are made by microphonemadness.com
Barry W. Hull wrote on 4/16/2013, 8:40 AM
The mic picking up the other talent is exactly the issue, it adds that slight "echoey", hollow sound, what is what we are trying to eliminate. Yes, I suppose that is the term, we want source isolation.

Neither the room nor the participants are noisy, with only one mic, the audio sounds great. However, both lavs pick up the conversation and add a hollow sound.

When you say "panned tastefully to fit the mix", are you referring to someone working a sound board during the interview? If so, that goes beyond our budget.

Also, this is not live, so if a mic goes out, we'll pause and replace it with a spare.

Are the hypercardioid lapel mics better at avoiding picking up non-intended audio, source isolation? I suppose we could caution the talent to avoid moving their heads away from the mics.

This issue took great audio from one person, and has turned into very marginal audio from two people, not good.

Also, are there mics (not shotgun) that are good placing overhead that can pick up both voices nicely?
Barry W. Hull wrote on 4/16/2013, 9:01 AM

Thanks for the input. I see Countryman makes a "directional" lav. Maybe that will help improve source isolation.
farss wrote on 4/16/2013, 9:09 AM
A single cardiod mic on a boom arm above, just in front of the talent and just out of shot should get you out of trouble.
Lapel mics can be problematic as you've discovered if the two people are close together. I had one horror when the talent kept scratching the mic :(

At a pinch once I've used a Rode NT6 X-Y stereo mic on the floor aimed at the two speakers, sounded better than I thought it would have.

larry-peter wrote on 4/16/2013, 9:48 AM
If you are recording your subjects to individual audio channels and have any plugins that can do expansion rather than hard gating, you could give that a try. I generally use Waves C-1 for this, although I believe Sony’s Graphic Dynamics can do this also.

Second, see if your mix improves with one of the subject’s audio channels phase inverted and both panned to center. It’s a total crap shoot depending on the distance between them and how much bleed is there, but the elements that are directly additive will become subtractive. It could ruin your sound, or improve it a bit. At the very least, if the lavs and transceivers are identical you can somewhat reduce unwanted sound that is hitting both mics simultaneously, such as room echo.

I think you’re going to find this next to impossible to achieve automatically. Someone has to take care of any audio mix, either on-set or in post.
Barry W. Hull wrote on 4/16/2013, 10:29 AM
Bob, thanks, yes, we've discovered the difficulty to being two people being too close with lav mics.

We are using the Audio Technica AT899, which is an Omni-directional lav. It worked great for one person, not so good with two or three people.

I'm wondering if the AT898 is a better solution. It is a Cardioid lav, advertised to cut down on off-axis sound. Until about five minutes ago, I didn't know the difference between Omni-directional and Cardioid, but seems it might make a big difference.

atom12, I tried the phase invert, read that suggestion on another forum, didn't seem to make any improvement, there is quite a lot of bleed, maybe that is due to the Omni-directional lavs we are currently using.

The technology side of me thinks that if this can be done manually, it can be done automatically. The Vocal Rider plugin from Waves is INCREDIBLE, surely some audiophile programmer guru geek can automate this process, or at least get close, then provide some manual tweaking.
larry-peter wrote on 4/16/2013, 11:33 AM
I'm not as familiar with scripts in Vegas as I should be, but it appears there is a Vegasaur script "Voice Over" that will allow you to duck selected audio channels when a threshold level is detected on another. Scripts aren't top of my mind in video, let alone audio, but there may be some available scripts that could do something close to what you need.

Edit: and it's not clear if this ducking is event based or level-threshold based. May involve editing the tracks anyway. There used to be some side-chain plugins that worked with Vegas, but that was several versions back. I haven't even tried for a few years. I remember a thread from way back, where a guy had figured out how to use the Wave Hammer Surround plugin as an automatic ducker. Might search for that.
Jay M wrote on 4/16/2013, 11:42 AM
You could try a head worn mic to isolate each person. You could also try to place them farther apart.

If you are using a multi-track recorder, and once you get proficient, how hard is it in Vegas to just cut out the track on the unused mic? Whenever I do an interview, I do just that. It's time consuming, but the results are well worth it. What makes it time consuming is that you can't just look at the wave form because there are little vocalized affirmations which aren't easily seen that need to be there, such as when the one not talking utters "uh huh", "yes", or "hmmmm".

rs170a wrote on 4/16/2013, 12:08 PM
The technology side of me thinks that if this can be done manually, it can be done automatically.

The device you want is an automixer.
They come in a variety of sizes (for inputs) with prices ranging from around $500.00 to almost $5,000.00 for the really good one.
The gotcha is that you'll have to do a lot of reading and playing to get it to do what you want.

larry-peter wrote on 4/16/2013, 12:15 PM

This is the thread concerning side-chaining and the (rather convoluted) process for using Wave Hammer for this. Might help you out.
musicvid10 wrote on 4/16/2013, 12:34 PM
"When you say "panned tastefully to fit the mix", are you referring to someone working a sound board during the interview? If so, that goes beyond our budget."
I was still referring to post mixing, a practiced operator can automate the gains and pan in real time on the Vegas timeline.

"Lapel mics can be problematic as you've discovered if the two people are close together. I had one horror when the talent kept scratching the mic :("
It's a huge training issue in live theater. Whistling, singing, crying, clapping, and yelling into the other actors' mics is not uncommon. Even had an actor who kept kissing the leading lady's cheekworn mic!

"I see Countryman makes a "directional" lav."
Again, you don't want a hypercardioid lavalier worn on the clothing. Slightest head movement destroys the gains. Headworn or taped to the cheek on the upper jawbone is the only way this works.

This is half the price of the E6 headworn and for interviews (but not preachers/belters) works just as well.
DavidMcKnight wrote on 4/16/2013, 12:36 PM
The VASST UltimateS Pro tool has had the ducking capability for years. They might even have it now as one of their FASST Apps.
larry-peter wrote on 4/16/2013, 12:47 PM
Aren't the Ultimate S tools event based - ducking occurs when an audio event on a particular track is found, rather than a certain level threshold is reached? That's what I always gathered from reading about them, but haven't used them so I could be wrong. In my understanding the OP was looking for a automatic solution that didn't involve trimming or editing the audio tracks.
Chienworks wrote on 4/16/2013, 1:00 PM
Ultimate S is a set of script-based tools. As such, they have no way of operating on the *content* of the events, but only on whether an event is there or not.
Byron K wrote on 4/16/2013, 1:14 PM
Does Vegas Pro have silence detection in the audio editor? I know Cubase can detect silence in an audio track and adjust the threshold when the level is low enough to mute the audio. I use this all the time.

Are each person's audio on separate tracks?
Barry W. Hull wrote on 4/16/2013, 2:57 PM
Musicvid, to beat this dead horse, these video sessions include the same two people, week after week, seated, not moving very much because the cameras are stationary (3 cameras, only one person working them and the sound). They are talking to each other in normal casual conversation. There are the typical hand gestures, probably some movement, but nothing like what happens in live theatre. I am sure they can be taught to put their head in the proper position when they speak.

That being said, I’m wondering if a cardioid lav would satisfy our needs, shutting out off-axis sound, isolating the source, reducing mic bleed. I suppose we could use the mics worn on the face, that just seems a bit odd for a seated conversation. I only see them used on stage.

Thank you Mike, automixers, found the Audio-Technica AT-MX341a SmartMixer, maybe that will help. It has several modes, "Free for all", "Filibuster", and "Chairman".



I use VASST Ultimate S Pro, the ducking feature, works well on coughs, “umms”, scratches, etc., but it is manual.

Jay, correct, it’s not an issue of muting below a certain defined level. Sometimes there are very low volume words that should not be cut. That is why I thought there might be software that can determine… yes, I hear this same sound in both tracks, track 1 is the loudest, so that must be the source audio in this spot, so I’ll attenuate track 2, and vice versa. Actually, seems like that would be easy to do compared to some of the automation I’ve seen.

We are recording using the Tascam DR-680, it records separate channels, does a great job. Listening to each individual channel is clean and crisp, of the source audio. It is when we combine both channels that the audio sounds a bit “echoey” and hollow.

Bottom line, when Talent 1 speaks and it is picked up in Talent 2 mic, I wish there was a simple, non-labor intensive way to cut that out and be done with it, maybe those automixers are the ticket.
musicvid10 wrote on 4/16/2013, 3:11 PM

A cardioid lapel mic just barely covers the width of the face; a 15 deg. head turn can be heard as a volume shift.

Bob's solution of an overhead boom or a pair of stationary overhead cardioid mics just off-camera is a better solution for your situation, now that I understand it better.
Deadening your space as much as possible is a given.
Barry W. Hull wrote on 4/16/2013, 3:59 PM
Got it, thanks. We bought a pile of $8 foam rubber mattress pads, took the echo right out of the place, nice and dead, although it does give the studio a rather unusual, weird look, all those mattress pads.

Just spoke with Audio Technica, they were very helpful, gave the same advice.
John_Cline wrote on 4/16/2013, 4:00 PM
I have had great luck with a stereo microphone placed overhead, slightly in front of and between both speakers if the shot is fairly tight and the speakers are close together. (I use the Audio Technica AT-825 microphone.) Since both capsules are close together, it avoids the out-of-phase "hollowness" caused by a pair of mics spaced further apart. It is also mono-compatible and you can mix the two channels to mono with no problems. As Musicvid pointed out, controlling the acoustics of the room is very important as there are some things that just can't be fixed in post.
Barry W. Hull wrote on 4/16/2013, 4:20 PM
Thanks a lot John, just when I had this sorted out... (joking)
farss wrote on 4/16/2013, 4:35 PM
Stereo microphone, on the floor. Room was small and incredibly live, tiled floor, hard walls etc.

Barry W. Hull wrote on 4/16/2013, 4:45 PM

Sure, there was some sound bouncing around the room, but as they talked and moved, there was no difference whatsoever in sound volume, quality, maybe a stereo mic is a good choice too. Audio is too friggin complicated.
farss wrote on 4/16/2013, 5:22 PM
"Audio is too friggin complicated."

Good audio is quite simple really, in this case you've over thought the problem.
Quite simply if you've got two mics setup such that bleed is a significant problem then you have one mic too many.

Certainly when trying to get excellent audio it is complicated however good location audio for vision is not at all complicated.
In this venture the audience simply needs to be able to hear what is being said without too many distractions. The people are obviously in a "space" and the audience can see them. Their voices don't have to sound like a voice over, in fact if they did it may well work against the production. Certainly less room than in my video is in order but that's simply enough addressed as you've already done. The sound of them moving, turning pages etc is expected.

ushere wrote on 4/16/2013, 7:13 PM
i sort of went from wired lapel > wireless lapel > boom or fixed shotgun.

i also tried with great success on a couple of occasions (thanks to bob) using a proximity mic. worked very well on a three head conversation.

of course, room tone, talent, etc., are all major factors, but certainly the sound from a well placed (GOOD) shotgun is much more preferable to 'echoey' lapel mics.