Dialogue editing is driving me nuts

M

MeAgain

Carbonite Member
I hate to be a pest again, but since I'm an audio amature I really don't know if I should under process things to stop myself from overprocessing-

I'm editing ADR dialogue that was done because the original audio was not good, and my sound guy got my ADR stuff WAY too low because he was afraid of clipping. Thankfully, the only noise is hiss, but I'm having trouble knowing what frequencies to boost (how to get the "Warmth" in a much deeper voice) and how to control syballance and noise without hurting the voice.

I'm using Noise Print in STP but it seems to be backfiring on me. I comb through with a preview and it only removes a very small section of freq from the actual voice, but it still sounds crappy to me.

ANYWAY, I exported a section of my timeline with the edited audio synced up. I also added some roomtone, which is the hiss you're hearing, and I'm wondering if I should just leave the original voice, hiss and all, and it will blend in with roomtone...

Anyway, it's a .mov file, real quick, I simply need a second opinion on how it sounds as an audience member (terrible, OK, it's fine, etc?)

http://www.megaupload.com/?d=GCSK8AXL

(I tried installing dropbox but I'm on a school computer and it won't let me).

Thanks.
 
That clip is sure low level IMO you can't use that.

A good voice track being the basis of the show and depending on how much ADR there is .. a consideration might be to rerecord it. Having done it once your actors should be quicker next time.

If you can't do that .. I'd process the same 60secs in all the restoration styles you prefer and run it right through your system to a DVD for test playing on real world speakers.

You still might want to go back to redoing it .. it's going to be around for a long time. I assume it's going to Film Festivals. hth.

Cheers.
 
Really good noise reduction units cost over $30K so don't expect miracles from a plugin Apple tosses in for free. As a rule more passes of light reduction sounds better than one pass of heavier reduction.
Also it's a bit of an art with some science tossed it so it's not simple. SO, you will be better off probably by getting someone who has better tools and more experience. If you listen to what is being removed and there is any of the voice in it it's going to sound crappy. Also you should probably aim to get the level up around -12 dB.

I really hate dealing with MegaUpload so I may not download the clip. But yes a school won't let you install software on their machines, but you don't own a computer?
 
Adr

Adr

Well, I found the sound usable, but you need to live it up a lot, with eq and some conv. reverb.

If I were you, I´ll check the lipsync too, its a little off and doesn´t match right the action of the actor. If you´ve the chance, maybe doing some rerecording having in mind the advices from the guys here at dvx, u will a better/realistic ADR.

If you can´t do anything of the named above, you might want to loop just a noise part, check with a Spectrum Analyser and look at that freq with some parametric Eq´s, check if the frequency that´s bothering you is in the same spectrum that the voice, and try to remove some with a Notch filter with a Parametric Eq.

my .02
 
I'm going to download it and see what I can do with it...

It's bit short... can you upload something longer ?
 
Last edited:
This clip is an exported quicktime that has already been adjusted by me (roomtone added) so this isn't even the original sound file.

I'll upload the original sound here, exactly how it was recorded (peaks around -30db to give you an idea)...

Basically, I cannot re-record it (I would have gotten good levels, but I had to direct the actors AND boom, so I had to trust it to someone who I thought was more experienced than me).

I basically need someone here to give me an idiot's work flow- because I am more or less a sound idiot- using stuff I can find in STP. Something where I won't overprocess the voice so much. I realize each clip will vary in what it needs, but I'm talking specific tools and processes I should look into in STP. Noise Print isn't working all that well for me.


Here's the link on the original sound file:

http://dl.dropbox.com/u/6428385/MD1T01_1.WAV

EDIT: And here's a longer clip- http://dl.dropbox.com/u/6428385/MD4T01_1.WAV

(Please ignore the slew of weird takes, the guy was unfamiliar to lip-synching to himself)
 
Last edited:
Thanks for the drop box!

So it doesn't sound bad (at least on my laptop - the video played but wouldn't download). It is pretty soft but that should be fixable. What I find missing and makes it sound like a low budget film is the lack of "place". Where is this room? It's not roomtone it needs but the sounds that should be in there if it was a "real" space. Are there neighbors watching TV that we can hae a bit of? Is there a window in the room? Is it in a city? If so we should hear some of that city. Is it in the country? If so we should be hearing some of that. Once you add in some of the world the ADR (and the track in general) will sound much more "real".
 
I played with the last few takes of the longer clip. This is just cleanup, if you layer some BG in you may not even need to fill with RT. As long as you can sync it I don't think there is any reason to rerecord it. It's low but seems pretty clean.

www.askinc.net/stuff/MD4T01_tst.aif
 
Yeah I should mention this isn't the FINAL mix- right now, I'm just focusing on voices.

I've got leather jacket rustles, sheet rustles, chair creaks, dog barks- the whole nine yards (all recorded MYSELF with decent levels), so thankfully I don't have to worry about fixing or re-recording my sound FX- just mixing them, which will be enough work without the voice cleanup. I should also mention this scene happens right after the main character (the girl he's talking to off screen) just woke up from a dream, so I want it to be a bit quieter and surreal at this part.

So I appreciate your work and opinion- I'm really disappointed in myself for the sound levels on the voice, and if I could re-record, I would, but time's run out on that.

So with the sound cleaned up, I could boost it enough (around 20db) to get to nominal levels without distortion? And do you have any recommendations of how I can approach this through the tools at my disposal (STP, FCP audacity, etc?) Obviously I can't do what you did, but something that can remove that hiss?


Thanks again Noiz
 
Everything I did was done with iZotope Rx in stand alone mode. All I did was apply some NR, boost +20dB, another NR pass and some final EQ to de-ess a bit (a real de-esser would have been better) dipped some low mid to take out some of the mud and rolled off the top around 12K or so and the bottom around 80.

You can get a 10 day trial for Rx so if you plan your time right you might be able to do it all on the trial. You can pick it up for under $350 (which is unbelievably cheap for the quality, but still might be too much if you don't use it a lot). And since your a student you can get it with an academic discount so that's 50% of (from their store so that would be $175.
And it will work as a plugin in all the apps you have (RTAS, MAS, VST, Audio Unit, Direct X) or as a stand alone application.
 
I actually DO need to finish this in the next ten days- so I'll download the trial tomorrow- AND since I'm graduating soon, I may purchase it as well- I'm out of cash because of the film but I could throw it on my CC. Even if I don't use it much in the future, half off something like that is a real bargain.


So is the program somewhat easy to pickup if I spend a couple of hours in it? Would I be able to get the hang of it quickly (I'm not dling it until tomorrow, hopefully it won't be blocked by my school, we'll see)?
 
I find it also ok.

What can help is cutting really on words, so that the dialogues "cover" the noise.

EQ: around 10kHz with a parametric eq, high Q factor (so narrow) pull down (in volume)
lift a little with little larger Q factor around 3kHz.

It was a quick ckeck, so you'll have to play with it, but indeed with adding of roomnoise, ambiance, it should be ok.
 
So OK as in just barely manageable or OK as in it'll stand up on a good system as long as I put some work into it? Sorry, I'm pretty paranoid.

But thanks for checking and thanks for the EQ tips- in his voice specifically I was going around the 4-5k range and found a lot of syballance when I boosted it, and found more luck boosting around the 1k range due to his deep voice.


One more question- is it better to do the "overall" 20db boost first, then boost particular levels in the EQ, or boost those first, then do a smaller overall boost?
 
MeAgain said:
So OK as in just barely manageable or OK as in it'll stand up on a good system as long as I put some work into it? Sorry, I'm pretty paranoid.

But thanks for checking and thanks for the EQ tips- in his voice specifically I was going around the 4-5k range and found a lot of syballance when I boosted it, and found more luck boosting around the 1k range due to his deep voice.


One more question- is it better to do the "overall" 20db boost first, then boost particular levels in the EQ, or boost those first, then do a smaller overall boost?
Click to expand...

Well if you boost the signal fi!rst (you can do it with "normalize") you will hear the noise better. So the EQ will be better also.

At least that's what I think :-p Some migth do it the opposit way.

It's a bit like "do you add a compressor before the EQ or after ?"
 
It's also a bit like me not knowing when to compress or what it means in that context!

In the meantime, RX's trial won't let me save, so so much for using that this week- and I can't find where the half off student discount is- maybe they aren't doing that anymore noiz?

Looks like it's STP for me- do you guys think STP is better for noise removal or Audacity (even though it's free)? Remember, I'm new to sound, so the more basic for me to understand the better, for now.
 
The trial mode (not demo) should be fully functional for 10 days according to the site. Look at the FAQ for academic pricing. As a general note you should check that for everything you buy. A lot of companies have very good academic pricing and many are full products with no restrictions (some are not upgradable). Red Giant makes some really great plugins (visual ones for FCP, AE etc) and they are also 50% off for students and teachers and are the full versions.

I found the sibilance around 3200. You can see the bright blip in RX. Rx doesn't have a de-esser so I just did a sharp cut but a real de_esser would be better since it's level dependent and won't notch except when needed.

It doesn't really matter if you boost first or last, in theory. I NR'd first then boosted then NR'd again on the logic that at the original level I would get the "truest" print and that after the boost things that were not noticeable before and any artifacts of the boost would get cleaned up a bit. But I don't know that that is any better than boosting then NRing.
 
FYI

Obtaining an Educational or Academic Discount

We do offer academic discounts direct on our web site at 50% off of our retail price. Only one academic discount will be issued per user. As such, please make sure that you are ready to make your purchase before inquiring.

Note: The EDU discount is good only for FULL product licenses and cannot be applied to upgrade pricing from LITE or DirectX-Only versions of iZotope products.

This discount doesn't stack with other discounts or bundled offers.

To qualify however, we do ask for proof of your academic status in the form of a valid and dated school document confirming your proof of current academic status.

We must receive a scanned, dated official document from the registered school or university detailing your current student teaching relationship with that organization. The document must detail clearly your name and title, as well as dated proof of current student or teaching status.

When you are interested in obtaining an educational discount, please e-mail us at sales@izotope.com with your attached scanned document. We can then provide you with a unique Coupon Code which you can use on our Online Store to make your purchase and download your products.

Note: Your Coupon Code will only be good for one purchase, so if you would like to purchase more than one product, please make sure to add them all to the same purchase.
Click to expand...

and
--------------------

What's the difference between trial and demo mode? Do I need to authorize to run the plug-in demo?

The Authorization process is only used after purchasing iZotope software, to transform a demo product into a fully-functional version. If you'd like to try out an iZotope plug-in before purchasing, authorization is not needed, and you can press the Trial or Demo button to exit the Authorization Wizard.

For the first 10 days after installation, an unauthorized plug-in can run in Trial mode, which offers full functionality with no restrictions. After 10 days, the plug-in will continue to run in Demo mode. In Demo mode, the plug-in's output will mute briefly every 40 seconds. Aside from this periodic silence, there are no other limitations in Demo mode, and both Trial and Demo mode offer the same feature set as the authorized plug-in.
Click to expand...
 
You must log in or register to reply here.
Back
Top