I'm new to dialogue, I'm wondering is my dialoge premix too muddy? It is a film project, would be output to DCP and play back in an normal cinema. The whole film is redubbed, any one can give me some opinion?
To my ears, it does indeed sound as if there is too much build-up in the lower mids. I wouldn't say "muddy", but rather more boomy, boxy, or honky - those are three words I would use to describe it. Cutting these will clean it up for you, but as you said, this was definitely miked too close.
Two more things to keep in mind for dlog editing: 1) You will almost ALWAYS get more natural sounding results by cutting what you don't want, as opposed to boosting what you do want. As you carve away some of the lower freqs, the spectral balance of the dlog shifts to the highs, and as a result will sound brighter, more airy, etc. 2) Generally speaking, tighter Q factors will cause the dlog to sound more unnatural. Usually I try to stick with broader Q factors, and subtle changes. In my opinion, a slightly under-corrected problem often sounds more natural, than a slightly over-corrected problem. 3) Don't be bashful about cutting the low end. I am constantly amazed at how much of the lows are cut in film dlog.
Tips For This Situation: I uploaded an example of your dlog, with some changes made to it. The first clip is your original dlog, the second clip has a +2.7dB boost @ approx. 460Hz (to highlight the offending frequencies), and the third clip has the following changes.
I cut roughly -1.3dB, @ 1kHz, with a medium Q factor.
I cut about -2.7dB, @ approx. 460Hz, w/ a medium Q factor.
I cut the lows, with a -2dB low shelf, @ approx. 115Hz, w/ a pretty gentle Q factor.
I cut more lows, by adding a medium HPF @ 70Hz.
There were no other changes made to these clips, except a universal volume boost of around 4dB (before processing), and I used some make-up gain to roughly match levels (post processing). The only thing in the signal path was one EQ.
If you still feel like you're missing "air", I would suggest using a harmonic enhancer, as opposed to a traditional EQ. Slate Digital's "Revival" is an excellent one, and best of all, it's free. Obviously, you have to be subtle with things like this, but it does a really great job of lifting the highs, without adding a bunch of harshness. Potentially, you could also add a de-esser if you're still having issues with the harshness.
In hindsight, I probably should have cut another -1dB with the low-shelf, and maybe eased off on the cut @ 460Hz, and left it at -2dB.
Well what Joseph Santoyo said is pretty right, i mean it can get you in the ballpark and have a decent sounding vocal recording, but in no way a serious one.
If you take 2 voices make a big Q around 250-500 in your situation you will be a lot less muddy but what i have seen is that when i EQ stuff that were recorded in the same room (and the room/mic/recording has problems) they tend to sound alike thus losing their character and to my understanding actual clarity in the mix.So if i was watching a scene with 2 guys talking i really want to be able to separate those voices in my ears.
So what you can do to avoid this is use a distortion box really subtly with a compressor after it but EQ will be mandatory for this, you have to feed a pretty nice vocal signal into the distortion box.
The subtle effect you are going to create with the distortion box is going to individually add some grit just on top of the vocals making them much more separated and colorful.
Now of course this is not a process that you are going to be able to make right away , it needs a nice balance between your EQ - Distortion - Compression , plus you have to already have the voice "in your head", i mean when you hit that spot you have to understand it and stop , you have a good voice, it's just a matter of turning 3-4 buttons really , when you master it :)
You can use whichever compressor / EQ you want , but i would recommend you use Decapitator VST as a distortion box OR some other really hi end one that is for mixing don't just go stick the vocals in a guitar box, save that for later.
One last thing, if you had let's say a very colorful voice you know with grit and nice bass and generaly nice qualities you may not wanna touch it just soft EQ Co
To answer the comment on my answer:
You can put a frequency analyzer on the speech channel , try to see the dominant frequencies should be 2 , 1 around 5k where the "s" are and a bigger Q area from 200-450 where the body of the vocal is (or at least those are the frequencies to look for).You want the S to have more clarity.Note that this is just to see the freqs , do nothing else on the vocal channel.
So now go add a multiband compressor like C6 to the music channel and use the 2 band pass controls which just focus on a specific(variable if you like) bandwidth. ***c6 plugin (or other) must be SideChain mode !
So now feed the sidechain multiband with the vocal signal as key, have semifast attack and smooth release , let it cut around 3-5 db depending on the music behind and you should have a pretty good separation effect. Try to play a bit with the settings until you get a transparent effect,you dont want the compressor to pump!
Bypass the other bands on the multiband and leave only the 2 specific band areas affected by the multiband.
If you need more info on sidechain multiband check some youtube video:)