Recording Levels, Noise and Normalization

MH_Stevens wrote on 8/3/2005, 3:10 PM
I am recording voice-overs into Vegas6 with an AT3035 mic (7" from side of mouth) and the m-audio Firewire410 preamp/interface (and Spot's box). If I raise the gain as high as I can without cliping (0dB) as recommended in the Vegas manual I am getting amp noise that I can plainly hear in my monitoring cans. For any of you who have the same setup the gain knob in this setting is at 3 'O' clock. To remove the noise and get clean audio I need cut back the gain to the 2 'O' clock position. This gives a wave form in Vegas of only half full-scale and I need to "Normalize" every recording. Surly a good recording should not need to be software enhanced in Vegas with normalization.

Any thoughts on my equipment, settings etc?

Comments

Catwell wrote on 8/3/2005, 4:19 PM
I think you may be trying to record too hot. For voice I usually set the peaks to -6 or -9 dBFS. This leaves a little headroom for the unexpected. Also it is unusual to have the voice as the hottest signal in you mix. I will typically use normalization in Sound Forge where I can set the voice tracks to -20 dBFS with RMS (average level) Vegas does not have the RMS setting. Vegas will normalize the peaks to -0.1 dBFS unless you set a different maximum. I hardly ever use the normalization in Vegas since I bought Souind Forge.

With Digital Audio you have a great deal of dynamic range so you do not have to push your levels to the max. You also have brick wall that destroys the signal if you ever exceed 0dBFS. If you have any experience with analog audio your 0 dBu is about the same as -20 dBFS.

With analog mic preamps, I have often found problems if you push close to the maximum gain settings. The noise goes up much faster than the signal. You may be running into this fairly common problem.
farss wrote on 8/3/2005, 5:49 PM
I use the 410 but with Rode's NT1A, quietest mic you can buy.
Only noise problems I have are birds two doors and 100 yards away!
The NT1A is a high output mic though but from memory I usually only need the gain at 12 O'clock, still this shouldn't make much difference assume your mic is fairly quiet.
Now monitoring with cans can be a trap, any noise is far more obvious than it will be to the average listener because their environment is much noiser naturally. I know we all want things as good as possible but don't go overboard!
What do the level meters in Vegas say when you're setup to record?
I can manage about -65dB when everyone holds their breath and as I say that's mostly from ambient noise.
Bear in mind speech sounds quiet and just bumping up the level is not the way to get it to sound louder, level and loudness are not the same thing. You need to use compression. I've done VOs for use in a store kiosk and even turned down 20dB they still sound LOUD. Wave Hammer is very good for this but do monitor things in the same environment as the typical listener, you might think it sounds way cool blasting them out of theri chairs, they're more likely just to turn it off.
Also the actual voice itself has a HUGE bearing on how VOs sound, that's why the pros get serious money for their voice.
Bob.
Catwell wrote on 8/3/2005, 6:44 PM
Bob,
You are absolutely right about the Pros. I was just working on a set of interviews about the history of Hydroplane racing in Seattle. On of the people was a radio annoucer from years ago. His levels didn't change. It was uncanny. No need for compression, just match the levels to the other clips.
MH_Stevens wrote on 8/3/2005, 6:48 PM
Bob:

I am using Wave Hammer. I am using it after Track EQ and before a Noise Gate. My concern is the Normailization. I am doing it ALWAYS as step one in the voice-over audio work flow because of the flat wave forms. I am wondering if I need stop using Normalization and let the gain or auto gain in Wave Hammer handle it.

Chers,

Mike S
B.Verlik wrote on 8/3/2005, 8:56 PM
You might try using compression. Set it up so it only kicks in, if a sound is up to a certain level. (a specific db) This way it will automatically bring up the level of talk, only when you talk, and should basically shut off, when ever there's a pause, lowering the noise level whenever there's silence. This will have to be done in the mix down unless you want to spring for an external compressor.
JohnnyRoy wrote on 8/3/2005, 9:56 PM
I have a FW410 that I use for voiceovers with an AT4033 with the input gain set at 3’oclock and I have no appreciable noise from the FW410 at all. I have all my FW410 mixer inputs at 0dbfs. I also keep a proximity of about 5" from the mic with a pop filter. Perhaps 7" to the side is a bit too far. Have you tried bringing the mic closer? This should give you a stronger input signal.

~jr
farss wrote on 8/3/2005, 10:11 PM
First things first. "flat waveforms" isn't a very specific term!
Just how much gain is Normalize having to apply to get the peaks to -0.1dB?
Anything over 12dB as a general rule of thumb and you're getting into troubling territory!
Here's the thing. Record something using any recording system, even good old cassette tape and you should aim to get the maximum possible signal onto the recording system. There's no way that turning the gain down will reduce your noise, in fact you're making it worse in fact. Sure you mightn't hear it so much but that means nothing.
Every recording system has an inherent noise floor, it's fixed. Let's say it's -80dB. Record something at -10dB and the resulting recording has a maximum S/N ratio of 70dB. Record the same thing at -30dB and you've got a maximim S/N ratio of 50dB!
I've simplified this no end but you can see what's happening, if you record at 24 bit your noise floor is pretty low so you can afford to record at a slightly lower level to have more headroom but even so if it looks like a flat line in Vegas you've got issues.
One thing to remember about compressors, they effectively increase the gain of the low level portions of the recording and hence the noise.
Also I think you should have the noise gate before anything, compressors should go last in the chain.
Can I recommend a good book? Audio Postproduction for Digital Video by Jay Rose, huge amount of good info in there. Not just how to do it but how things work.
Bob.
MH_Stevens wrote on 8/4/2005, 8:26 AM
Bob & all: thanks for good replies. I have Jay's book - I am also studying Spots's "Audio Mastering in Vegas." Have not seen my issue exactly addressed but I now see I might drop Normalization and let the compressor/expander do the work. I will take Johnny Roy's advice and get closer to the mic.

Reading a lot I noticemany contradictions in audio theory. Here its has been said, as in the Vegas manual, to get the most out of your recording equipment. To me that means to peak at 0dB. Others here say to peak voice at -6bB or it will be too hot - any comments? Also nearly every book or thread I read has a differnt opnion on the order of processing tools. I note Vegas uses as a "default" audio tools chain 1)Noise gate, 2)EQ 3)Compressor. Does this imply that a non-expert like me should stick with these three processors (unless there is a problem that needs special attention) and to use them always in this order?
Spot|DSE wrote on 8/4/2005, 8:34 AM
You need to leave some room if you're processing. Look for averages in the -18 to -12dB dB range when you're recording, depending on what it is. Voice should be averaging at -12dB, peaks no more than -3dB.
The order of FX is indeed significant; you almost always want compressors last in the chain. Regarding the noise gate first...I actually tossed the noise gate from my default chain, I don't use one, and find them generally useless, but some folks like em'. For drums and certain effects, they're great and necessary. Anway, you want to cut noise before it's processed in the EQ, which is why it comes first. Compressor comes after EQ, because even though an EQ is adjusting frequencies, you're modifying the overall gain of the signal in the EQ, which means if it came after the compressor, you might be modifying gain enough to undue the dynamic control the compressor is doing. There are times that EQ comes post-compression, but not very often.
farss wrote on 8/5/2005, 3:58 AM
To get the most out of your recording equipment yes you should peak to 0dBFS as read by a true peak reading meter like the one in Vegas. Confusion comes from not everything having and everyone not using peak reading meters. Decades ago when I used VU meters they were normally calibrated so that 0VU was 20dB below 0dBm, simply because a) true peak reading meters didn't really exist and b) that left enough headroom for any kind of content you would find back then. However our content (Muzak, yes literaly) was already highly compressed so we could afford to calibrate our meters to -10dBm. With the advent of modern music that could have a peak to average ratio of just about anything and digital recording technology VU metering became problematic, what you truly had to watch out for was exceeding 0dBFS on the peaks and that could happen even if your VU meters were reading -10VU. Also the old analogue gear saturated rather than clipped, going too hot didn't sound all that aweful like it does with digital.
Now what Spot is saying is also very true, if you record hard against 0dBFS then you do have to be careful what you do in post. Most Eqs lie, those nice looking curves you see in Vegas are telling you fibs and this isn't unique to Vegas either. You put in a high or low shelf with 12dB slope and it doesn't just drop away, there's actually a bump at that knee, i.e. the gain goes up. Now if a note at 0dB hits that knee it's level goes up, but up to where, you guessed it, it gets clipped! But this you will see, it ain't the end of the world, you can reduce the level a few dB prior to the Eq to leave enough headroom.
Or you can do as Spot suggests and leave 6dB to spare in your recording, if you're recording at 24 bit you can easily afford to loose 6dB, after all a few dB too low is way, way better than clipping any day.

But as I said before 20dB is getting to be a worry, flat lines are very problematic and do need be addressed.

Bob.