Suggestions for VO echo removal

[TimS]

Does anyone have any working tricks to remove (or certainly limit) echo from audio..
I've been given some VO that was recorded on a lavalier mic in a bright, live room and there is a little echo in the recording.
I know there is probably very little chance that anything can be done, but it was part of a 25 advert shoot and they cannot redo to VO.
I was asked to tidy it up a little.
After having a listen to the samples, it's not going to be a easy task, if even worth attempting.
Do any of you POST guys have any suggestions??
Appreciated..

[Chinagraf]

This is a bit left o centre, but have you heard of a program called "photosounder"? What it does is convert audio files to spectral bitmaps that you can then open in photoshop, edit and then convert back to an audio file. I've seen one example where a dude pulled the synth solo out of a mix, but apparently it took a few hours. It's complicated and a but tricky to get your head around but you can do some amazing things with it. It's actually more of a sound design thing, but a few guys have been using it to isolate things from a mix, some with success, some without.
Depending on where the reverb sits spectrally relative to the voice it may work. It's not a simple set and forget type thing though. I find the whole concept really interesting. Website is www.photosounder.com

Worth a look just to see the concept.

[TimS]

Thanks Andy - I'll check that out - good idea too!

[ChrisW]

Wow, sounds like you need a PHD to edit sound in Photoshop.
Voice is so naked, I think it's going to be hard.
What about the old Transient Designer?
Cedar Sound Restoration?
A friend of mine has been restoring historic recordings of the early blues musicians and says the Cedar stuff has been incredible:
http://www.cedaraudio.com/products/products.html

[Chinagraf]

[quote="ChrisW"]Wow, sounds like you need a PHD to edit sound in Photoshop.
[/quote]

Probably, but you can also just open pictures of your face and play it back. Mine goes "bwub bwub swish tish wah wub wub"

The Cedar stuff is a good call. Made me also think about maybe you could use one of those Dolby CAT things the film studios use for pulling out background noise etc.

[Kurt]

ozone rx?

[Thirteen]

As the vocal will presumably be panned centre, you could try using an MS decoder plug-in to separate the center info from the side info, and EQ some of the vocal FX out of the left and right channel information but not out of the centre information.

[Kurt]

Wouldn't a lavalier be mono?

Ozone RX is a spectral editor, apparently very good. Adobe Audition has a similar but more basic one built in.

As the vocal will presumably be panned centre, you could try using an MS decoder plug-in to separate the center info from the side info, and EQ some of the vocal FX out of the left and right channel information but not out of the centre information.

[Thirteen]

Wouldn't a lavalier be mono?

Ozone RX is a spectral editor, apparently very good. Adobe Audition has a similar but more basic one built in.

As the vocal will presumably be panned centre, you could try using an MS decoder plug-in to separate the center info from the side info, and EQ some of the vocal FX out of the left and right channel information but not out of the centre information.

Oops, of course you are right, I need to pay more attention when I am reading.....

[jkhuri44]

umm, gating ...with perhaps a Waves SSL channel strip (great gate) with natural reverb following sometimes can do the trick....however, on some shitful material, the gate is falsely triggered by random sounds like kids running around and screaming or other nuisanceful things :P Damn those kids!!

Another pain in the arse trick is to make a duplicate track, phase reverse....then fade in and out the sections where the acho begins and ends...so that you're effectively phase cancelling the tails. The above is a pain in the BEEP...but it really depends on how much you want that recording to sound good :-)


Another trick is to cut a bit of lame mids in the...lame mids area....just fiddle around until it doesnt sound like it was recorded in a tin shed...your reference should be a VO recording...in terms of EQ...you'll find the best field recording for vocals sound like they were done in a studio..so use that as an EQ benchmark. This eq trick doesnt get rid of the echo, but in conjunction with either of the above 2 tricks, can really help make the recording acceptable.

[Chinagraf]

Actually, I've used waves trans x to dry some stuff up before. Kinda a bit like the TD. But a little different.
With Jamil's suggestion about phase reversing the tails, maybe you could do the fading up thing on the phase reversed track by using a compressor with a fast attack slow release thingy goin on? Then maybe key it off the voice and eq the sidechain?
I'll try to think of something even more complicated if I can.

[jkhuri44]

yeah, what andy said...!

thats the less dumb way to do what i was saying, LOL

i would just hope the compressors response is smooth, otherwise u get BS sidechain stutter, which sounds like ballz.

[heathen]

Add heaps more echo till you can't hear the one that's bothering you. :)

Probably very little that can be done to make a big difference, that's why we are supposed to go to so much trouble not getting stuff like that on a recording in the first place, but yeah you know that.

[Chinagraf]

Tell that to a location sound recordist...haaaaaaaaaaaaaaaaaaa
(I'm talkin about location recordists working in difficult situations with testy directors who don't care about sound, not a lack of skill btw)

If it was music and Mutt Lange was producing, he would just re-record the voice overs with him doing it and not tell anyone, then say "yeah that's how it sounded when i got it"

[TimS]

Thanks for the suggestions guys.
I'm currently away at the moment and cant do anything on the samples, but I will certainly try some of these suggestions.
RX - I have that so I might give it a whirl when I get back sunday night.
Appreciate the responses guys - all good ideas to try..
I know its not great, but the sound was recorded in a very reflective room and there are quite some reverb tails in the dialogue..
Shame, I dont know why they didnt do the VO in a booth in the first place??

[Linear]

Try using an SPL transient designer. It (somehow) allows you to alter the attack and release envelopes of audio independent of level. Works wonders. I have a friend who does alot of VO stuff and he uses it to remove unwanted verbs on VO's. Apparently you can get it as a plug in.

I sometimes use it to change the room size on overheads. It's a magic box!

Chris

[jkhuri44]

i think transient designer comes as a plugin added to the UAD 2.

[tunetown]

I've got the transient designer plugin on uad2 if you want to try it Tim. Not sure how it works on full mixes but works a treat on solo tracks.

[jkhuri44]

just as a side note, going to a big post production place in Sydney, cant remember where...

they have a 120 input Harrison desk feeeding 120 protools channels.,...its a big place..put it that way...has a garb load of 40bit (internal processing) DSP plugins made for post....

for the rest of us, there's the above stuff we're talking about :-)

[TimS]

I've got the transient designer plugin on uad2 if you want to try it Tim. Not sure how it works on full mixes but works a treat on solo tracks.

Thanks Pete - tracks are solo at the moment - unmixed..
See how I go..

[Barney Loveland]

There's also the Sonnox Trans Mod plugin which does a similar thing to the SPL transient designer. I've used this to dry up roomy sounds. Not sure how it would go on a solo exposed VO. Might be worth a listen.

[TimS]

Once again, all these suggestions are great.
Thanks again guys..

[heathen]

Transient designer, hmmmm I've been meaning to get one of those things. Still though it might kill the reverb tails but on long vowels you're still probably going to hear the reverb building up, I dunno I could be wrong.

I remember having to mix a band where the drum kit was like this, was a real PITA but it did sound fine after messing with it for a while with heavy gating and using a good gated and non-gated reverbs blended and setting the pre-delay right. Could have used TD here I reckon with good results.

[Linear]

what makes a transient designer great is the fact that it is independent of level - so unlike a gate or expander (which work on a threshold) the TD manages to work the same envelope on all waveforms.

a very clever bit of kit. doesn't work on mixes (obviously) but also does great things when you link two channels and make the signal of one waveform shape another (ie kick triggering bass to make it tighter).

If you go to the SPL website you can read the PDF of the manual which explains it pretty well.

Chris

[heathen]

what makes a transient designer great is the fact that it is independent of level - so unlike a gate or expander (which work on a threshold) the TD manages to work the same envelope on all waveforms.

a very clever bit of kit. doesn't work on mixes (obviously) but also does great things when you link two channels and make the signal of one waveform shape another (ie kick triggering bass to make it tighter).

If you go to the SPL website you can read the PDF of the manual which explains it pretty well.

Chris

Ahhhhhhh yes very cool. Ok I'm sold I'll have to get 1, when I have some spare cash that is.