Dialogue editing - too much room now

[dcwave]

So I am editing (cleanup mostly) several video promo shorts for a company. The recording sounds like someone turned on the pink noise machine and set it next to the narrator. ADR is not possible. I can clean up 95% of the noise with MB compression and several NR filters. My problem now is that the room tone is very prominent - boxy and in your face. I've been playing with EQs for the past hour - my ears are fried.. Any thoughts? Live with it and tell the customer that a turd is a turd?

[Michael Tibes]

Maybe the transient designer (UAD/SPL) can help it a bit?

Michael

[MattiasNYC]

Yeah, for the most part garbage in = garbage out.

I encounter this regularly. They think it's a great idea to lug in a power generator and record in a castle and then expect it to sound like a studio environment.

Just tell them you did what you could and it is what it is.

Getting rid of room tone can be wicked hard. Use an expander maybe?.....

[LEX]

Expanders only work when there is some light BG noise you want out.

It sound like the room presence is too over bearing, so expanding is going to make the problem more noticable.

Fine a section of the room and fill in the blank spots, and treat the same as with the dialog.

If they are expecting it to be "clean", then they are out of luck.

LEX

[kdm]

My approach is to strike a balance between cutting noise and maintaining a decent voice tone. If you kill the voice tone, or get a boxy sound, you might need to back off on NR.

As m2 noted, you can only do so much with a bad recording. It's ADR or the client will have to live with it.

[MattiasNYC]

It really sucks, because there's only so much we can do. And trying to educate our customers is difficult too since we're straddling the fence between truly educating them so they can make better production decisions next time, but on the other hand we risk sounding like we're whining, pushing blame or can't deal with the problems.

I suppose it usually takes care of itself though in the long run since those making such poor decisions and then won't listen to our advice won't amount to much anyways, meaning as we hopefully grow into a higher bracket of work they "won't follow"......

[LEX]

Do you have a Cedar DNS1000 or 2000?

You'd be amazing at what it can do for dialog clean up. Maybe you could rent one.

It is worth a shot. If you can upload a sample, I could try it on mine and see if it works on my stage.

LEX

[TAFKAT]

The TC Powercore Restoration Suite is good also , but of course that is a little thinner on the ground.

I have the suite and could give it a go for you as well if needed.

[dcwave]

Thanks for all the ideas. I came to a balance around 1am last night and the producer seems happy. Tried some different MB and NR methods and left a small bit of the noise which masked the ringing from the boxy room sound - a little EQ around 500Hz and small dip around 1k helped it to sounds better. About 85% cleaner but the voice still sounds decent. it doesn't help that the speaker has a heavy accent and a resonate baritone voice and talks in a way that makes it hard to tell if I cut off the heads and tails of words.

Lex thanks for the use of the Cedar - rumor has it the closest one to me is in Vegas. With as many films getting shot in Utah you would think someone would have one on their post stage.

Vin - I've tried the Powercore suite at a friends place once trying to help him out. I just couldn't get it to feel right - you get used to certain tools in my case a custom written EQ, C4 and X Noise and a way of working.

I am interested to hear what a Cedar can do. It will be a 35sec clip.

You can pull the file down from here: https://www.onlinefilefolder.com/1siudJgrHm80rF

[LEX]

PM me where to upload.

I created 3 mono files, sorry I don't work with stereo dialog.

One with the Cedar only.
One with the Cedar and 4 filters on the Euphonix S5
One with the Cedar and 4 filters on the Euphonix S5, with a bit of expansion from the Euphonix dynamics section.

they are 24 bit. I thought your file was 24bit, so I just created a 24bit PT session.

They are 4.7megs each. I can email them.

The cedar completely gets rid of the pink noise without taking too much out of the dialog.

Took me about 10 minutes to set, filter and expand.

With the expander, you might need something underneath to help it.

LEX

[Andrew J]

Maybe try the demo of reNOVAtor (http://www.algorithmix.com/en/renovator.htm). Or maybe NoiseFree (from the same site) would be better suited.

(Haven't tried them myself)

[dcwave]

LEX ran it through the Cedar with and without filters and expanders.

He suggested I post a file with comparisons. Keep in mind I spent about an hour or more tweaking settings and starting over, whereas I doubt Lex spent that much time (I hope you didn't).

www.dcwave.com/dcwaveaudiofile.mp3 (http://www.dcwave.com/dcwaveaudiofile.mp3)

This is the cedar with filters and then my chain of EQs C4s and NR. I like what the cedar does with the filters. My chain has some artifacts which make the Sss more pronounced.

[LEX]

The cedar is really good at getting out broadband noise in certain areas of it. While the filters themselves can get into more detail.

It is usually a combination of both. I suggest you post the original file so everyone can hear how bad it is.

I spent 10 minutes BTW and whether any noise reduction plugin would have worked or not, I would have went with either version with filters or with filters and the expander.

If you post all 3 versions, plus the original and yours it might give some more options to see what was going on.

Remember, I was on a dub stage, so I had the advantage of the screen curve.

I didn't listen on the near fields.

I put in a HPF at 100hz, low pass filter at 10.5k, and notched two mid hi frequencies fairly steeply and narrow.
If you post all versions, you'll hear the ring disappear with the filtered version.

A little noise doesn't both me as long as the dialog isn't compromised to where it is thin, or dull.
Sometimes you have to live with some noise.

LEX

[dcwave]

Source File - www.dcwave.com/dcwavesourcefile.wav - 48kHzx16

Cedar: www.dcwave.com/CedarOnly_01.wav
Cedar with filter: www.dcwave.com/Cedar+filters_01.wav
Cedar with filter and expander: www.dcwave.com/Cedar+Flters+Expander_01.wav

My Tools: www.dcwave.com/Audio_cleaned.wav

[TerryG]

Try this:

http://www.heirapparent.com/real/dcwavesource-denoised.wav


Single pass with Isotope RX denoiser on your source file.
I took a noise print in a gap between words.
-12dB on tone noise; -18dB broadband reduction.

No audible gating/pumping artifacts as in your cleaned example.
And, the volume of the voice remains identical to the source.

Another pass or perhaps some EQ and/or compression to balance out the syllables/phrasing and you're good to go.

Here's the second pass with identical settings:
http://www.heirapparent.com/real/dcwavesource-denoised-2pass.wav

[LEX]

Clean but you can hear the audio clipping off at the end of many of the words.

LEX

[TerryG]

Clean but you can hear the audio clipping off at the end of many of the words.

LEX

These exist on the original source file... probably assembled from edits.