Final edit: Average audio levels?
 

Hi,

What is a good average audio level for an edit?

For example:

"-18" for quite noises

... everything inbetween ...

"-2" for the loud stuff

I have heard that when you are monitoring your audio when filming, you want to ride the levels (I typically shoot solo with XHA1, the shotgun mic and/or wireless)... In other words, keep the levels from peaking, but keep them up in the yellow range.

Is the same true for a final edit?

What do the pros do? I am just looking for a few tips. :)

Thanks!
Micky

i myself want the fullest range without going over, EVER going over, because digital cant handle going over one bit, like analog could. so when you say -2 we will assume that is not an "averaged" -2 and that when you get close enough you can see peaks of waves that slammed into the top :-(

Many final output things i have analised never go over 50% of the full possible ranges , which is strange because they are probably the same people who had to edit in 32bit with 96k :-) so in comparison your awareness is already higher than some :-) were they Paranoid or just safe?? if they chose 75% of the total range, it would have made some sence.

most importantally it should be similar to the movie they just pulled out. being similar to broadcast is quite impossible :-) as you never know what they are setting to, each channel has thier own thing going on there. Again if they were being aware, why is it when you switch channels you must adjust the volume?

The Client isnt compelled to have to adjust thier volume during the viewing. Dang i cant hear that whispering. turn it up, dang the music is to loud will you turn it down, dang now i cant hear the dialog. Whispering is so completly different than talking i know they ARE freaking whispering, so tone it down a bit, but dont overdo it. there is a balance to everything. Different people watch all kinds of things in a multitude of different situtaions, not everyone has thier 5000W stereo up in thier soundproof theatre room, dedicating thier existance to just watching that. (we will leave out hearing the theatre next doors audio even IN a real theatre:-) Sometimes there are other humans in existanace that are sleeping. Like when daddy has to work nights, and the kids want to hear what is going on with the video.

also volume/level isnt everything, some "sounds" are "full" have many many frequencies and ranges going on, like say a full orchastra will walk all over some dialog not because its level is so high but because the range of frequencies is filling.

so take everything into concideration, and have something that is both pleasing on your 2" speaker, and works too when you rattle the roof. stay within range, and just try to do better than everyone else, and you will get close enough.

for the moment i will leave out compression and expansion, but if your doing commertials, you should learn how to be extreemly irritating :-) that way people will mute them.

Depends on several things you didn't mention. One big factor is where is it going? For example when mixing for making a DVD, average peaks at -10dB.

Other factors include the dynamic range of the dialog, and whether there is music SFX, or whatever behind the dialog. More often than not, I end up compressing, normalizing, or otherwise adjusting the dialog levels.

There are mixed standards for your peak levels. Some networks want them at -4, some at -6db.

I think on average it's -4dbfs for peaks.


For example:

"Vision TV's standard for digital audio level is -20 dBfs (full scale). The equivalent analog level is 0 VU, 1000 Hz in the NTSC analog standard. The analog peak value is +16dB."

Which basically translates to -4dBfs for peaks on your master.

Then again some say -12dBfs.

Discovery HD in Canada:

2.1.3 Headroom

"Transmission limiters clip at +8 dB. For broadcast stereo tracks, transient audio peaks must not exceed +8 dB above reference tone when measured on an audio meter using the "True-peak" ballistic set (0 ms rise, 200 ms fall). For 5.1 surround mixes, audio peaks may rise as high as +21 dBm (-3 dBfs). When mastering to a digital format and/or using an Absolute Scale or Peak meter, where "0" is at the top of the scale and reference tone is at -20 dBfs, broadcast stereo tracks should peak at no more than -12 dBfs. "

I see short duration peaks ranging from -12dBfs to as high as -3dBfs. So check with your broadcaster as to what they want. I can't remember what maximum acceptable peak audio levels are for a film master. I think it's -3dBfs.

What are you outputting for?

As Denny mentioned, it depends on the final destination of the mix ... there are different requirements for SD broadcast, HD broadcast, DVDs for home viewing, theatrical delivery, etc. It also depends on whether it's going to be in Europe or North America as European broadcasters generally follow EBU standards while North American broadcasters and networks follow SMPTE.

Normally for broadcast submission (at least most of the networks here in NY) Simplified PCM version: Reference level of -20dBFS. (test tone) with program audio peaks not exceeding -10dbFS.
For home DVD release, I go to around -5.0dBFS.
Perceived loudness is another can of worms, and is currently being litigated in congress with bills soon to come to complicate things more. At least for broadcasters.

Wow! Awesome replies folks!!!! Thanks you!

Sorry that I did not specify the output.

I hate to say it, but I mostly do web video with the possibility of going to DVD/broadcast tv in the future.

I can't wait to look over all of your replies in detail. I am at work now, but will post a longer reply later tonight.

A billion thanks everyone! I appreciate you all helping an audio noobie out. :D

Spot on Rick and just what I do with a slight difference of -18dbfs and -12dbfs that are the UK broadcast levels. DVD and web videos are the same as yours at around -5dbfs.

I try and get my voice over peaking at -6db with the average between -12db and -6db.
Background music under VO peaks around -18db but this differs with different styles of music. After this has been cut in FCP and mixed I then run the audio through Sound Track Pro with a compression filter set to -6db and this brings those peaks down a bit and tightens up the whole track.

Thanks to everyone!!! This forum has been a life saver. Hopefully one of these days I will feel comfortable answering questions instead of just asking them.

Excellent advice Marty! A great read too. :)

Many thanks for sharing your thoughts. I will be sure to test my audio on different pieces of equipment... I usually only test on via my headphones and/or my external computer speakers. Now that you mention it, I probably should broaden my horizon a bit and test on a few other systems. :D

Hi Richard, thanks for the reply!

Interesting! Sounds like you use your pro instincts to tweak the audio when needed to fit the current edit/situation/scene... Man, I would love to intern with you on your next project! :D

Know of any good online video tutorials, webpages, and/or books that might help me out? Hmm, maybe I should look around the Premiere and FCP sections of Lynda.com?

Hi Denny! More great info! Thank you!

Sorry that I did not mention sooner what I am outputting for... Web for short-term, with possibility of DVD and/or broadcast later. I guess my goal would be to create an edit that I can put on web, but will be optimal for other formats later.

Interesting! I am in USA, but I did not even think about standards and the differences between countries. Thank you for pointing that out. :)

Hi Rick, thanks for sharing your numbers. :)

Also, thanks for mentioning the tidbit of info about perceived loudness. That is an interesting topic. I plan on googling around to find more info on that topic.

Excellent! That gives me something to shoot for. Thanks for the reply Gary! :)

Ooooh, cool! Good tips. :)

I have been doing a ton more editing in Premiere lately, so I will have to figure out a similar workflow for Soundbooth. Any Soundbooth tips?

A ka-billion thanks all! I really appreciate the pro advice!

Have an excellent day/night all!

Cheers,
Micky

Don't ever, EVER, rely solely on headphones for your final mix! Evaluating the raw sound in a take to look for noise, etc, and perhaps cutting dialog, fine. But the mix MUST be done on decent speakers. Phones introduce a whole range of distortions that mean the track will often sound completely different when heard on speakers. Some flaws, like phase reversal between the channels, are only audible on speakers. And don't forget to listen to a mono mixdown (again, ON SPEAKERS) because not everyone hearing your video will have stereo and some surprising things can happen - don't know how it made it through QC but not too long ago I saw a show broadcast on National Geographic where the dialog went completely missing for the entire program. Music was there, SFX were there, but the spoken dialog and narration was gone. Likely cause was a phase reversal in the dialog tracks between the left and right stereo channels. That sort of thing can escape notice when you only listen to the stereo mix because phase cancellation only occurs when you mixdown the two channels to mono.

Why ?
 

I'm curious. Why is it neccesary to keep the peaks at -12 for digital audio for broadcast?

I can see not letting the levels get past -3 but what is the reasoning for the -12 limit for broadcast? Does this not limit the dynamic range of the audio and introduce noise? Was it an arbitrary limit imposed by networks or is there a technical reason for the -12 limit? Not arguing, but I really am curious. jm

Hi Steve, thanks for the reply and help.

Oh, wow! I never even considered that before. Thanks for bringing this to my attention.

Now, this might completely silly for me to ask, but when you say "mono mixdown", is that something you do within your NLE, or is that something I would do via the playback equipment? I assume that I should do a mon mixdown via FCP or Premiere using one of the stereo > mono filters?

Again, sorry to throw a noob question right back at ya. :(

Thanks again for the help! I really appreciate the pro tips and advice (from you and everyone else).

The worst-case analog broadcast chain often only has 8 or 10 dB of headroom after allowance for safety margins.

You should be able to do a mono mix in your NLE and some monitor controllers (Mackie Big Knob, for instance) allow it in hardware as well.