Normalize or Compression for final audio post work?

E

eco_bach

Well-known member
Hi
I'm learning by testing to see what quality I can get out of my limited audio gear.
My main audio for a shoot i just completed comes from an ATR3550 lav into a Zoom H1.

My audio 'sweetening' consists of
1-equalization
2-slight reverb to add some room presence
3-normalization

Only noticeable when wearing headphones but AFTER normalization I can hear some slight noise...I assume I'm hearing mostly the noise floor of the inexpensive lav mic.
I've avoiding using ANY noise reduction filters since they all seem artificial(but perhaps I don't know how to use them properly).
I've read on other audio forums that compression INSTEAD of normalization is a recommended last step.
Does anyone else agree-disagree?
If yes, what are some recommended compression settings?
If not, what are some recommended normalization settings?
Another 'tip' I recall reading for equalization is to only apply half the amount of equalization you need, reverse your track(so it plays backwards) apply a 2nd time, reverse back to normal.
Supposed to canceled the phase anomolies caused by EQ's.
Anyone try this?
Thanks in advance!
 
Couple things...
1. Films are mixed by ear, so there really aren't "recommended" settings for anything. Honestly, everything is different.
2. That said, when mixing by ear, that assumes high quality monitors in a treated sound studio, and at a specific sound level. When you start adjusting EQ/reverb/levels/noise reduction outside of this environment, the biggest danger is that the changes you make sound good in your headphones (for example) but actually make it sound worse on say... laptop speakers or stereo speakers. So when in doubt, less is definitely more, and check any work you do on multiple playback systems (laptop speakers, TV speakers, headphones, home stereo, ect...) so that it sounds good in all of them.

Normalization, as Dave mentioned, is extremely rare for dialog tracks.
Compression is occasionally used near the end of a mix, but it's more common to raise and lower the level of the audio tracks as they play (called volume automation) by hand riding faders on each channel. If that's not a realistic possibility, using some compression can help ensure that all of your actors lines are heard. Obviously, story comes first in the end.

One last note, if you're looking for a cheaper mic for use on a boom in the future to get more natural sounding recordings, consider this...
http://pro-audio.musiciansfriend.com/product/CAD-CM217-Condenser-Mic-Buy-One-Get-One-Free?sku=271324
You'd need a preamp before your H1, but just something to keep in mind.
 
Neither.

Normalization simply takes the loudest point in your sound clip and makes that 0dB. If someone was mumbling but then there was a loud sound for a split second, the loud sound will be 0dB and their mumbling will probably not be loud enough.

If you add compression, you are squashing the louder parts of your recording (e.g. the signal), making the overall volume quieter. Now, to get the levels back up, you have to raise the volume of the entire track. This raises the noise floor right along with it.

What you should do is decide how loud you want the dialog to be, and then make it that loud using volume automation and possibly sound editing.

Honestly, with sound the simplest tools are usually the best.
 
In a completely different application - final disc music mastering - I don't use normalization at all. I use a limiter to keep everything from peaking 0db while still maintaining a strong signal. That said, it's best to keep these things (compression, limiting) as minimal as possible and if you can get good results "flying" the faders or automation like robotsound says - do it that way. I just use limiting in my application on songs that are finals and need just a bit of fine tuning.(where overall volume level is concerned)
 
It depends a lot on what you're recording, and how much time/effort you want to put into it. The advice above applies to narrative film work, where you are doing the audio editing as well.

Normalization, just to be clear is a two-stage process. Stage one finds the highest peak in the chunk of audio that you are normalizing, and stage 2 then applies a fixed gain to the whole chunk in order to place that peak at a specific level. Standard "normalization" would put the peak at 0dB; but most normalizers will let you specify any target peak level, and will normalize up or down.

If you are recording events, or in any uncontrolled situation, and leave a lot of headroom in the field, normalization is a quick and effective way to get the audio into roughly the range you are interested in working in, without worrying about adding in peaking distortion anywhere.

If you are doing quick turnaround of events/conferences/music, the normalization will help you safely maximize levels.

If you are recording high-dynamic range stuff, like classical music, normalization is often a key part of the finalization process... DESIRED peaks are few and far between, but mist be kept intact, and you want to raise the level of everything as much as possible without squashing dynamics.

ProTools, e.g., has a normalization plug, but I find for many things the gain plug more useful... it actually provides the two stages of normalization separately... you can select a range of material, and ask it to tell you the peak level, then you can choose a level of gain to apply. This is great for situations where there are some peaks you don't care about (e.g., in applause, or someone dropping something), but where you want a quick way to raise the level of what you do care about without having to do it piecemeal.

Compression I'd say to avoid except in quick-turnaround conference situations, etc. It throws away information; normalization does not.

Cheers!
Barry
 
Now ironically in feature films compression is used a fair amount but normalization almost never.

BUT film lags behind many things and Normalization is not the sound destroyer that it once was. One note about levels though. If you really want to make sure you don't introduce distortions you need to keep your peaks below -3dBfs. The reason is that in the D to A proses the analog signal is created from the samples and it can crest over the highest sample.

Any way the artistic reason to avoid normalization is that it's kind of useless with dialog. Riding the volume is a much better way to get your levels correct. Normalize will get the peaks at zero (or other level) but most of the time your trying to bring up the low bits and leave the louder parts alone. Compression can in theory do this but it often leaves artifacts that are a problem and it's "stupid". Some quiet things want to be quiet and some should be louder but plugins don't know the difference. What often works better if you have a good one is a limiter. With that set correctly you can push your dialog up and keep the loud bits from clipping.

But as Ullanta mentioned what you should and shouldn't do has a lot to do with what your doing. Fast turn around has different rules than narrative films. DVD is one set of rules and Broadcast another.
 
The only place I've used normalization is for game sound fx and game dialogue. Other than that, I've never used it for TV or film.
 
Forget normalization. Don't use artificial reverb unless your scene is in a cave but sounds like a small room or something. Match dialogue levels by region based gain or volume automatisation, whatever your DAW offers. You can mix all dialogue to peak (now and again, not just once) at 0 dB, put a limiter on the master output and finally lower the output volume by say 9dB for headroom. With these few rules in mind, mix with your ears, make it sound good to your ears, and you'll do alright.
 
The comment about reversing the audio and reapplying EQ is one of the strangest things I have ever heard. If you are honestly getting phase anomalies from your EQ that require that treatment, either you are doing something wrong or you are using a terrible EQ.
 
markblasco said:
The comment about reversing the audio and reapplying EQ is one of the strangest things I have ever heard. If you are honestly getting phase anomalies from your EQ that require that treatment, either you are doing something wrong or you are using a terrible EQ.
Click to expand...

Most EQ's have some phase shift going on but this sounds like one of those things someone thought up who decided the math would make it "perfect" but never actually did any comparisons by listening. I have come across a number of very whack "recommendations" to cure problems that are only problems if you pay more attention to the math than the sound.

As to EQ if it really bothers you you can get plugins that have no phase shift.

RE: reverb. It's used quite often. If you don't want it to sound like a cave take the time to get to know the reverb (or get a better one, Altiverb is excellent for post) and how to set it up to get what you want.

It is not however used (in film) as a "mastering" stage. It is used to make some sounds, ADR etc fit the space that is in the film. Little bits here and there. The only time you might use it as one big pass is if your on a film/ short where it all takes place in the same space. I do know some folks who have a stack of reverb busses and automate preset changes and sends so they are essentially doing one pass but it is effectively many different reverbs and pretty complicated to set up.
 
You must log in or register to reply here.
Back
Top