Removing Slapback Echo?

  • Thread starter Thread starter SouthSIDE Glen
  • Start date Start date
SouthSIDE Glen

SouthSIDE Glen

independentrecording.net
Allright boys and girls, I have a tough one (for me, anyway :o )...

I have a stereo digital recording of a wedding reception made via the built-in stereo mics on an MiniDV camcorder, which was positioned on a tripod off-axis from the dual PA mains (i.e. the camcorder was closer to the right main than the left one). There are both speeches and live music on the recording. Because of the nature of the setup and the room (a VA hall), there is a really nasty slapback delay or echo on the recording.

Anybody have any tricks or techniques up their sleeves as to any way of removing - or at least minimizing - the echo without doing too much damage to the main impulses?

I'm tending towards applying some kind of phase inversion to try and cancel out the echo, but I'm not sure how that could be applied to affect the echo only.

Any ideas or leads would be much appreciated. :)

G.
 
Oh man, that's a messy one. If the echo is more to the sides, it's possible to minimise it by using what I call "reverse M+S" processing :D Basically, in a multitrack DAW you make a copy of your stereo track, then, on the second one you reverse the L and R channels, and reverse polarity (or phase)... Mix both channels at the same level, and bounce. This will give you the sides w/o the center. Nothing new here. Now, mute the reversed channel (or get rid of it, just keep the original), and load the "side" that you created in the first step. Stereo flip that, and phase reverse. You should now have pure mono signal, w/o any of the side info. Hopefully that will be good enough.

Another possibility is to split the L and R channels and see if the echo is less objectionable in one of them.
 
Well, after showering on it (I do some of my best thinking in the shower since I can't sing for sh*t, even in there ;) ), I'm thinking something starting along the lines of duping the track, shifting the dup track to line up it's main impulses with the delayed impulses on the original. Then lower the volume on the dup to try and match the volume of the duped mains with the original delays. Then invert the dup track.

This wouldn't be perfect, by a long shot. I'm sure the frequency response of the delayed signal is off, perhaps by enough to nullify the effect of the inversion. Alsothis method would probably have a detrimental effect on the main impulses as well (though I don't see any way around that at this time.)

I have a few days before I have to physically address this problem as I'm still editing the 3-camera shoot of the ceremony itself, (audio is not an issue on that part of the project, I have good stuff there), but I'll be coming up to this problem sometime before or near Thanksgiving probably, and I wanted to get it out there to you fine folks now to give the problem enough time to germinate here.

Chess, baby, I tried the "Yikes" solution, but it locked up my PC. No go on that one. :D

G.
 
noisewreck said:
If the echo is more to the sides, it's possible to minimise it by using what I call "reverse M+S" processing :D
Click to expand...
I'm getting a little dizzy following all that in one reading :D, but I think I get the basic, very general idea of where you're going with that, noisewreck.

I'm not sure just how directional the echo is (I haven't yet really LOOKED HARD at the track yet, I just got it into my computer this evening), but I think most of it is from the room itself, possibly even early reflection from the hard dance floor between the PA and the camera, and I *think* - though honestly I'll have to look into it better when it's not approaching midnight after a long day - that there is a pretty even dose of slap on both channels.

I'll look into that angle (no pun) further, though. Thanks for the lead and the interesting technique. :)

G.
 
You might try to duplicate the slapback using a delay box or delay plugin. Then you would record the slapback to another track, reverse that and hope that it negates the slapback on the original track. The advantage to this approach is that you could record only the slapback to a track, not any of the original signal. This might make it a little easier to work with.

Another approach would be to attempt to mask the slapback by adding some early reflections or maybe some heavy but short reverb. Okay, that's kind of a desperate idea, but whatever happens, this isn't going to be easy getting rid of that slapback. So I'd be kind of desperate if I were you. :eek:
 
SonicAlbert said:
You might try to duplicate the slapback using a delay box or delay plugin. Then you would record the slapback to another track, reverse that and hope that it negates the slapback on the original track. The advantage to this approach is that you could record only the slapback to a track, not any of the original signal. This might make it a little easier to work with.
Click to expand...
Yeah, I was thinking of that myself, but what's blocking my thinking there is that I can't (I don't think?) create a clean slapback because I don't have a clean source to create it from. There is overlap in the original signal because the delay time is shorter than the length of many of the spoken (or sung) phonems. I'd think I'd be creating slapback of the slapback along with slapback of the main impulses and just pushing the problem down a series of chinese boxes. I might be able to gate out some of the secondary slap stuff maybe...I'll have to think about that more...

SonicAlbert said:
...whatever happens, this isn't going to be easy getting rid of that slapback. So I'd be kind of desperate if I were you. :eek:
Click to expand...
Yeah, I have no illusions that there is a perfect solution; I'll have to live with either some echo or some degredation of the good stuff, or both, probably. But it will be interesting to see how it shakes out and just how muh I can or cannot do.

But for right this moment my best course of action is to sleep on it. My brain feels even mushier than usual right now :p . Thanks for the useful input! :) And I'll cacth up with you guys tomorrow...

zzzzzzzzzzzzzZZZZZZZ*snork*zzzzzzzzz...

G.
 
I actually had this posted earlier but deleted it because I thought it was leading you down the wrong road, but seeing as how other people are leading you down more or less the same path, I figure I'd repost it. (Good thing I saved it in notepad first :))
-----------------------------------------------------------------------------------

I don't see either of these working but it's the best I could come up with and if the only other responses you get are "you're screwed" the worst that can happen is that you waste your time. :D Oh and you also said any ideas. You didn't say "no stupid ideas allowed" :D.

Way 1)
Gate the signal with the attack set to the time of the echo so the echo is lower than the primary signal. Compress a copy of the signal so the primary signal and the echo are at the same level then invert the copy.

Here you're basically trying to make the primary impulse louder than the echo so you have something left when you do the inversion. If the compressor or the gate change the signal too much this probably won't work.


Way 2)
Make many copies of the original signal. Delay them timed to the echo. Then invert. Hopefully you'll end up with the delayed signal delayed even more. Repeat with the delayed signal. Keep doing this until you push the echo to an easy point to chop off.

Here you're trying to make it possible to just remove the echo by pushing it to a point where there is silence. If the echo isn't the same as the primary sound, this probably won't work.



But, like I said, I doubt that either of these would work. I've just always hated the "there's nothing you can even try" replys to tough questions so I figured I'd throw something out there. If you actually do try either of these you should know pretty quickly whether or not they are going to work. But I should also say that you may have to automate the attack/delay times if the echo times are different.
 
SouthSIDE Glen said:
Yeah, I was thinking of that myself, but what's blocking my thinking there is that I can't (I don't think?) create a clean slapback because I don't have a clean source to create it from. There is overlap in the original signal because the delay time is shorter than the length of many of the spoken (or sung) phonems. I'd think I'd be creating slapback of the slapback along with slapback of the main impulses and just pushing the problem down a series of chinese boxes. I might be able to gate out some of the secondary slap stuff maybe...I'll have to think about that more...
Click to expand...

You can gate that, or using a ducking delay type setup. Once you have the slapback by itself on another track, you have more flexibility as far as editing it.

You can't worry too much about clean slapbacks and whatnot, it's going to be messy no matter what. The point is to reduce the problem to where it is at least somewhat acceptable. What you are really dealing with here is a blunt instrument, a bludgeoning type of solution to a situation that is never going to be perfect.
 
I'm beginning to believe there's more to Chessrock than meets the eye. The only 3 words in this thread I understand came from him. :)
 
in the future, if you have the option, ask the videographer to put a tape directly on the pa line outs so all pa - music, speech, etc... is captured in its entirety so you just have to sync up the audio to tape later rather than depend on the camera mic... my nephews wedding the guy hooked up a dat (running elp so it lasted 4 or so hours) to the pa line out and everyone speaking used a wireless mic - results were stellar and the dat had smpte timecode so sync to the video was a breeze...

in this case, i think you hit it right, you are going yo have to try to line up the channels to see if you can get the slap to be inverted and reduced in level. another option - if you have noise reduction software, you might try to see if you can get the NR tool to sense the slaps and maybe they'll be reduce enough that way...
 
RAMI said:
I'm beginning to believe there's more to Chessrock than meets the eye. The only 3 words in this thread I understand came from him. :)
Click to expand...
Nah, he just likes trying to get a rise out of people. He especially likes poking me in the ribs. I've learned to just gate that noise. ;)

G.
 
Kryptik: Sorry for the fast typing last night. You're input is appreciated :)

Sonic Al: Yeah, I agree that this will be like trying to remove a small brain tumor with a tablespoon. But I think at least *some* improvement can be made.

Noisewreck: Still have your "reverse M/S" idea on the plate. Just waiting to analyze the recording a bit better. Uh oh, is the revere of "M/S" "S/M"? I know whatever the solution, there will be pain involved, so maybe you're on the right track :D.

Gulfo: I was/am the videographer. The problem was that the reception part of it was a last-minute thing. The ceremony part (which I am currently editing) I had worked out great with three cameras and a seperate digital recording tapped off the church FOH. I'd have done the same setup for the reception - multi cameras and an audio tap off of the stage mixer - if I had my druthers, but as it turned out I wound up being the camera operator on a single tripod-mounted camera and audio from one of the guests with barely enough time to actually eat my meal :o. So I gotta work with what I got to work with.

It's kind of a shame in another way. The reception actually featured 5 (count-em!) different bands in a kind of Weddingpalooza...6 if you count the solo song the bride's sister sang and played solo on the keyboard. This is what happens when both the groom and his cousin are musicians and everybody in the bride's family is one as well. It would be great to have a top-shelf A/V of the whole thing; especially since there wasn't a single "Hokey Pokey" or "We've Only Just Begun" in the whole mix :D.

Anyway, I'll do what I can with it. The newlyweds know and understand the situation and will be happy with what I deliver to them; I just want to try and tweak it the best I can.

And all your help is indeed a help and is appreciated. Keep it coming! :)

G.
 
SouthSIDE Glen said:
Nah, he just likes trying to get a rise out of people. He especially likes poking me in the ribs. I've learned to just gate that noise. ;)
Click to expand...


An asshole like me is pretty easy to ignore, but I'm afraid you're stuck with that mess on your video tape. It is what it is. Short of getting all the members together and re-recording the show, using your current audio as a guide track so you can stay in sync, that is. :D Just have some fun with it, but I wouldn't worry about it a second longer. I am actually giving you good advice right now, believe it or not. Probably the best advice I've given anyone all week.
 
chessrock said:
Just have some fun with it, but I wouldn't worry about it a second longer. I am actually giving you good advice right now, believe it or not. Probably the best advice I've given anyone all week.
Click to expand...

I think "You're screwed" was better. :D
 
Yeah Chess...You need to learn show biz the way they do it in Vegas...If you walked out on "You're Screwed", the crowd would have went nuts and you would have become a legend...You see, the Beatles knew when to quit. :) :) :)
 
You must log in or register to reply here.
Back
Top