Finishing audio

L

livelikeriley

Well-known member
Just a quick question - I'm finalizing the audio for my thesis film (I'll put it online as soon as I finish it since I shot it with the HVX and LEX). I'm finishing out to betacam SP (my school's too poor to put out for digi) and wanted to know what I should normalize the audio to? I'm thinking I should peak around -2...does that sound right to everyone?

Thanks,
Mark.
 
It's generally a bad idea to Normalize sound for film. I wouldn't peak over -3dB FS because of other issues but that is VERY hot. Video and broadcast standards usually call for peaks to not exceed -6 (broadcast generally wanted peaks no greater than -12) with you nominal "0" (and your average dialog level) at -20.
There are new "standards" coming out but they are still basically the same except that they want even less dynamic range.

I say "standards" because there really are no standards. Every network has their own standards. But as a group they are mostly pretty much the same.
 
I guess I'll knock it down to -6 ... it did seem excessively loud and I had to dial down my speakers to play it with "normal" sound - although the meters were reading like -14 etc. in the "normal (not high or low)" parts of the audio. But I'll play with that till I get it right.

Why would you say it's bad to normalize the audio? I mean, I've already done all the mixing to get the levels correct in relationship to each other, I figured I'd bounce an Aiff out and then normalize that to be sure it stays below whatever threshold (-6 etc.) Is that not a good way of doing it?

Open to suggestions, I'm not an audio engineer at all! :)

Thanks!
Mark.
 
Well there are a few reasons, some not so valid anymore. First "normalize" used to mean peaks went to 0dB fs and that is just a BAD choice. That is one of the not so valid reasons these days since almost all "normalize" options these days let you pick a peak value. Second is that the math can lead to errors since it can be very complicated to boost to a level rather than boosting by an amount, this is probably also mostly in the past.
A very valid reason that isn't what you want to do is that a lot of folks coming from music want to "normalize" every clip to make them all the same and this sounds BAD.
Also films are mixed by ear in a standardized environment not to a number. Broadcasters mix (and compress) to a number because over-modulation is illegal and they can lose there license. So they set low standards to make sure they don't over-modulate their signal and compress the audio to hell and gone to make it sound "loud". Things sound different at different volume levels. At lower levels bass and treble sound proportionately quieter than midrange sound. So it's very important to mix at the level that things should be played back at. Mix stages are calibrated to 85dB with dolby pink noise and in a small room with near-field monitors you should be at more like 79-80 dB. So if you mix at the right level and it sounds right you shouldn't change the track after the fact. I usually use limiters on output tracks and sometimes buses, but really the best choice is to do most of your limiting with mixing.
So needing to normalize the final mix either means that your setup was way off or something is quite wrong with the mix. In either case normalizing is not likely to be the fix since your setting a playback level that is different from your mix and things wont sound the same.

The last reason, and again this doesn't apply to you, is that people use it as a lazy solution. "I don't need to mix, I'll just compress and normalize".
 
Thanks for that! You probably just saved me from making a mistake. I'll go back through and tweak the hot parts of the audio, but I'll stay away from the normalize...oh, and the tip about a limiter is great, I hadn't thought of that!

Thanks again,
Mark.
 
I cut a film at my day job, mixing on headphones. Don't mix audio with headphones. Do you see any headphones at your local cinema? Just a bad idea all around.
 
I addition to Noiz2’s great points, normalizing raises the noise also. No one wants noise. You might get away with normalizing 99 times, then on the 100th, the time you really need it, the noise overwhelms your audio and the clip is unusable.

Needing to normalize is a sign of poor gain structure. You should record as hot as you’re able. If you can’t get the levels you need the first time, re-record it later. If you can’t re-record it, re-evaluate how you’re recording things. Do ANYTHING possible before you normalize. I understand, sometimes you need to normalize (I’ve done it) especially with sound effects, especially if they're free. But if you don’t have to, don’t. You’ll get much better audio.

About the comment of music people normalizing: music people who aren’t that good at what they do normalize. Many people get their levels red hot, then compress and limit all the dynamics out of their music. These are not experienced mixers. If you have to normalize music in a studio setting, you are doing something very, very wrong. Avoid these people. Now, again, I have normalized audio, but as an absolute last resort. Like Noiz2 said, many people use normalizing as a lazy solution put a band aid on poor recordings.

Light compression and proper use of limiting will go a long way to improve your audio levels in post. Also, (I don’t know if this is appropriate for film) if you have one available, you can use a gate or an expander to eliminate or reduce noise. And remember, gate first, compress second.

The most important thing though, don’t be afraid to have some quiet parts! Having a quiet part just before a loud part will make the loud part seem even louder.
 
Noiz2 said:
It's generally a bad idea to Normalize sound for film. I wouldn't peak over -3dB FS because of other issues but that is VERY hot. Video and broadcast standards usually call for peaks to not exceed -6 (broadcast generally wanted peaks no greater than -12) with you nominal "0" (and your average dialog level) at -20.
There are new "standards" coming out but they are still basically the same except that they want even less dynamic range.

I say "standards" because there really are no standards. Every network has their own standards. But as a group they are mostly pretty much the same.
Click to expand...


How is that measured during the editing process? Is that the level that FCP, for example, measures?
 
I'm not sure what FCP measures. In general on DAW's the metering is dBfs, 0 is as high as you can go and you count down from there. Analog systems used a 0dB that you set to a reference. The big difference is that analog tape saturates and is to a certain extent self limiting. It goes in to a soft distortion that isn't particularly nasty and then eases into bad distortion. Digital systems only go to 100% then it's garbage. So since to top is so hard it's a lot more important to not hit it. Most DAW's and NLE's have not great metering built in. If your doing sound you really need to get better metering, either an external box or a decent plugin. That said films are mixed mostly by ear in a calibrated room. So your meters are mostly as guides. Broadcast is a different beast and you have to hit (or rather not hit) certain marks and metering is the only way to get there.

I was too general in my "music" slander. There is a whole lot of music that never sees a normalizer. If on the other hand you look at just about any "pop" tune (and I include pretty much anything that gets a lot of airplay except on classical stations as "pop") you will see what looks like a solid block of a waveform. They have either "normalized" or effectively done the same thing by compressing the life out of the tracks. I also see this a lot with folks who grew up making music on the computer. It's a real PITA to try and "mix" a music score that has NO dynamics and not make it just plain boring.
 
That's nice to know - that the meters in FCP aren't that great - I always wondered why they refused to correspond well to what I was hearing :)
 
Normalize does nothing... N O T H I N G ... (basically nothing) to the dynamic range, noise floor, etc of your track. Normalize is functionally equivalent to moving the volume slider or adjusting the gain in whatever environment you are working. The only logic that normalize should employ is automatic greatest peak detection. It uses that to set the gain and thats it.

Sorry Noiz, but in your rant you are conflating normalization with compression and limiting.
 
I think I understand the reason for being careful at least with normalizing soundtracks, but can't it be helpful to approximate levels? Matching the approximate levels between shots before you resort to compression, eq, or reverb to match sounds (all of which alter the actual recording instead of just boosting it)

Naturally going over everything listening closely and changing it, but I can't think of a reason for that not to be a valid starting point. As tigeba said it doesn't alter the sound in anyway except increasing it's gain.
 
tigeba said:
Normalize does nothing... N O T H I N G ... (basically nothing) to the dynamic range, noise floor, etc of your track. Normalize is functionally equivalent to moving the volume slider or adjusting the gain in whatever environment you are working. The only logic that normalize should employ is automatic greatest peak detection. It uses that to set the gain and thats it.

Sorry Noiz, but in your rant you are conflating normalization with compression and limiting.
Click to expand...



I'm sorry, but you contradicted yourself in your own post. If normalizing is just like raising the fader, that raises the noise on the recording also.

Normalize raises the volume on everything, even the background/tape/circuit noise in the recording. It doesn't change the dynamic range, it raises the peaks and lows the same amount. so, relatively, the dynamic range hasn't changed.


What normalize does is takes away all your headroom, if you normalize to 0dB. IF you must normalize, I recommend normalizing to -3dB. You need 3dB of gain to make things sound twice as loud, so that should give you room to work with.

And don't worry Noiz, I know what you ment.:2vrolijk_08:
 
Hxd Ped said:
I'm sorry, but you contradicted yourself in your own post. If normalizing is just like raising the fader, that raises the noise on the recording also.

Normalize raises the volume on everything, even the background/tape/circuit noise in the recording. It doesn't change the dynamic range, it raises the peaks and lows the same amount. so, relatively, the dynamic range hasn't changed.


What normalize does is takes away all your headroom, if you normalize to 0dB. IF you must normalize, I recommend normalizing to -3dB. You need 3dB of gain to make things sound twice as loud, so that should give you room to work with.

And don't worry Noiz, I know what you ment.:2vrolijk_08:
Click to expand...

What I said is absolutely not a contradiction. The ratio of signal to noise is not affected by changing the gain on a digital mixer* (obviously this is not %100 accurate but for our purposes it is true). I'm sorry, but only a politician would call increasing the signal and noise by equal amounts "increasing noise" :)

Of course the dynamic range is unchanged because normalize does nothing but change the amount of gain applied to your track.


Sometimes this subject feels almost as unwinnable as describing the difference between stereo and mono to inexperienced audio folk. The reason for proper understanding of what normalize does (and most importantly does NOT do) is that is that most of the time it is not the solution for the problem being presented.

Most people realize that their recordings are not "loud" enough. They don't understand what loud means, but they realize they don't have it. They are fooled into thinking normalize will somehow solve the problem, when in fact it will basically do nothing. What they almost always want is something that will alter the dynamic range of their content (usually DECREASE it), and that means a compressor or a limiter.


See you in hell normalize!
 
Last edited:
tigeba said:
I'm sorry, but only a politician would call increasing the signal and noise by equal amounts "increasing noise" :)
Click to expand...


Well, I humbly disagree. But disagreements are what make this country great.


But I do agree that many people are using normalize for the wrong reasons. I have found that the Gain Plugin in ProTools is more useful than normalize and I have more control with it to even out different audio takes, like in ADR. I avoid compression on dialog, but I have used light compression in places where an actor yells suddenly or similar situations. When I say light, I'm talking 2:1 or 3:1 ratios with a high threshold.


You're right though. Discussions like this can go on forever. I agree to disagree.
:2vrolijk_08:
 
You must log in or register to reply here.
Back
Top