Normalizing Digital Audio

PaulyV

PaulyV

Well-known member
Hey guys. shooting a commercial and network requires audio to be normalized to -20db. how can I do this. Im going to use Sony Soundforge to record the voice over and Premiere Pro to edit using a backing track royalty free audio clip. anyone have experience in this? im sure i could google my butt off about it but why not ask people with experience here first right? thanks in advance!

PS
Shooting with an EX1 however not using any of the audio!
 
-20? That seems kinda low to me, but whatever.

Soundforge has a great batch processor that will allow you to batch normalize audio files. I used it all the time when doing batch normalization of massive amounts of dialogue for video games. My audio files came from pro tools, so I would have the exported audio files from pro tools in a folder. Then I would create a new folder for the destination of the batch files, and then run the batch process. You can run whatever kind of process you want, like EQ, Reverb, or whatever, so you have to tell it you want to normalize and what to normalize it to. It's been a while since I've done it, so I can't give you a step by step on how to do it, but I read the Soundforge manual to figure it out, so it's in there.
 
davefisk said:
-20? That seems kinda low to me, but whatever.

Soundforge has a great batch processor that will allow you to batch normalize audio files. I used it all the time when doing batch normalization of massive amounts of dialogue for video games. My audio files came from pro tools, so I would have the exported audio files from pro tools in a folder. Then I would create a new folder for the destination of the batch files, and then run the batch process. You can run whatever kind of process you want, like EQ, Reverb, or whatever, so you have to tell it you want to normalize and what to normalize it to. It's been a while since I've done it, so I can't give you a step by step on how to do it, but I read the Soundforge manual to figure it out, so it's in there.
Click to expand...


Yeah the PDF spec sheet I got said -20db for digital audio -6db if using VU meters. its a local commercial so they can be extremely loud compared to national commercials. would that be why they want it at -20db?? this is new territory for me.
 
PaulyV said:
Yeah the PDF spec sheet I got said -20db for digital audio -6db if using VU meters. its a local commercial so they can be extremely loud compared to national commercials. would that be why they want it at -20db?? this is new territory for me.
Click to expand...

From the wiki article on dBFS, dB Full Scale, one has:

American Post: −20 dBFS = 0 VU = +4 dBu
 
I think that they are either talking about Dialog Normal, a common spec, OR they are talking about your normal (ie average) level. Most network specs run something like no peaks over -6 with your average level around -20. If you "normalized" it to -20 it would be WAY low compared to everything else.

A lot of these specks are pulled out of some execs backside. But if your not misreading it this one is just wrong. I did something for PBS a number of years ago and there were six pages of specs and most of them made no sense today. They were still using sheets from analog tape days.

If it's a Dialog Normal level they want you to hit, you can kind of fake it or you can go rent/ buy a $5K box from Dolby that will give you the number. -20 DilNorm would be a bit loud, that spec usually sits around -24.
 
Oh if it is DilNoprm you may want to head over to the ProTools forum (DUC). In the post section there is a lot of info on how to get close with out dropping the five grand.
 
Noiz2 said:
Oh if it is DilNoprm you may want to head over to the ProTools forum (DUC). In the post section there is a lot of info on how to get close with out dropping the five grand.
Click to expand...

HAHAHAHA dropping 5k!!!!! its a local commercial! just following their spec sheet which on the same page lists h.264 as an acceptable format to submit so im guessing this isnt "tape days" old.....so maybe its just what this local station wants. based on what ive read here it seems too low so i will call the station and clarify. thanks everyone.
 
PaulyV said:
HAHAHAHA dropping 5k!!!!! its a local commercial! just following their spec sheet which on the same page lists h.264 as an acceptable format to submit so im guessing this isnt "tape days" old.....so maybe its just what this local station wants. based on what ive read here it seems too low so i will call the station and clarify. thanks everyone.
Click to expand...

Your spec sheet is bound to be referring to a dialnorm spec, not normalization. Dialnorm is the average level of dialog within the program, measured by an integrating averaging meter system with specific characterisitcs. In a nutshell, if you put a VU meter on the dialog stem and average its readings over the entire program, you get the dialnorm. The standard calls for a dialnorm around -24 dBFS. You would set your average dialog levels in the mix first and lock them down, then everything else - music, FX, etc - is added to the mix with their levels set relative to the dialog. It's definitly not a holdover from the analog days but rather is a standard that is being legislated as part of the cut-over to digital TV broadcasting, partly in an attempt to control the wild level swings as the broadcast switches between program content and advertising or when users change channels. It also is part of the audio metadata inserted into the datastream that used to control the internal gain within Dolby decoder systems. http://en.wikipedia.org/wiki/Dialnorm
 
The new broadcast spec. (which I think is a legal requirement for broadcasters in the USA now) is to use a Loudness meter to EBU R128 at -23LUFS.

This new loudness meter spec. is designed for broadcasters to prevent the horrible level changes for adverts and trailers.

I have heard demonstrations and it actually works extremely well - including going from a classical recording with a large dynamic range to a heavy metal highly compressed recording.

Meter manufacturers like DK and RTS already include a Loudness meter in their meters and there is also a downloadable plug-in one available.
 
Normalizing

Normalizing

For the future work...You can normalize in Premiere Pro, Adobe Audition, and Adobe Soundbooth. It's probably similar in Sound Forge.

For Premiere Pro, you right-click on the audio track. It gives you the several options. You select the one pertaining to audio levels. In CS4, you are given the option to add a value to normalize to or you can normalize to default 0. In CS4, it's not as detailed but you get the same opportunities.

They definitely aren't normalizing to -20 db. That's lower than a whisper on there. db is what the software uses. Hopefully you got that clarified.

Tieuel Legacy!
 
j1clark@ucsd.edu said:
From the wiki article on dBFS, dB Full Scale, one has:

American Post: −20 dBFS = 0 VU = +4 dBu
Click to expand...

yup. and normalize means that the max peak in any file will be brought to that value, raising the rest with the same multiplier. The dynamics will remain the same. Thats why overcompression became so popular :-)
So now the new "audio engineers" from "schools" like SAE will insert a compressor and pump it to the max. The resulting visual of that audio file is one long square wave with 0 dynamics= yuck!
 
tieuellegacy said:
For the future work...You can normalize in Premiere Pro, Adobe Audition, and Adobe Soundbooth. It's probably similar in Sound Forge.

For Premiere Pro, you right-click on the audio track. It gives you the several options. You select the one pertaining to audio levels. In CS4, you are given the option to add a value to normalize to or you can normalize to default 0. In CS4, it's not as detailed but you get the same opportunities.

They definitely aren't normalizing to -20 db. That's lower than a whisper on there. db is what the software uses. Hopefully you got that clarified.

Tieuel Legacy!
Click to expand...

Note through that that sort of normalizing is not what is being talked about here. What the specs discussed here are calling for is a "dialnorm" of -20dBFS, meaning the average level of the dialog stem when integrated over the entire program will be -20dBFS, with all of the other stems (music, FX, etc) in proportion. That's a totally different thing from the normalization processing available in your NLE.
 
You must log in or register to reply here.
Back
Top