Two Mic questions- 12-bit and in camera gain

J

jpastuch

Carbonite Member
1: On one of my shoots I accidentally shot in 12-bit (long story). I was wondering- the camera and mic are not amazing to begin with, so it is already low-res- will the difference be that noticable? I was testing this on my other camera, the DVX, and I could BARELY notice any change. If I'm not going to be playing this in any theaters, will this matter a whole lot?

2: I noticed the Mic gain on my DVX and never bothered to mess with it, but out of curiousity, is there a difference in quality if I use the 10 db boost built into the camera or just boost it up 10 in post? Obviously, both of these functions are just GAIN and not magic, and if you FREALLY wanted to boost audio you could use both gains (which I imagine would make the sound unlistenable anyway) but I'm just wondering if bumping up 10 db in camera would provide better quality than bumping it 10 in post. I tried it while recording some quiet SFX tonight and I don't really like it.

Sorry if these are repeats- I truly searched over and over but it's hard to find these exact questions on the boards.

Thanks.
 
Analogy time: underrecording audio and then fixing it in post is a lot like underexposing the picture and then cranking the brightness in post. You may end up with audio/video that's loud/bright enough, but the results will sorely lack detail and depth. All that information that wasn't recorded when you rolled tape was lost for good, and no amount of "fix it in post" will ever bring it back.

It is almost always better to boost the gain on the mic preamplifier than it is to crank up the gain in post. By underrecording and then boosting the gain in post, you're just increasing the noise of a low bit-depth recording. Even though you're recording at 12-bits, you're likely only using the first 8 or so bits. When you're setting the gain on a mic, try to keep the dialogue sitting between -20 and -4. That way, you'll have enough headroom for momentary spikes in gain, but the levels will be substantial enough that you only need to make small occasional increases in gain when you're in post.

As for recording 12-bit audio: stop that! 12-bit 32 kHz digital audio hearkens back to the early days of DAT recording, and as far as I know was only ever popular with AM radio (where the limitations of the broadcast made low-quality digital recording acceptable). I have no idea why they built the miniDV spec to include it, since many non-linear editing applications will choke on it (since they expect 16-bit 48 kHz sound).

Of course, I'm making it sound like the differences in noise level should be night and day, which doesn't jive with what you found experimentally. Well, there's a good reason for that: you're starting off with an extremely noisey mic. All mics have inherent self-noise (its a quantum thing), just as all preamplifiers and other audio circuits have noise that they introduce into the signal. Some mics have a lower signal-to-noise ratio than others, and you'll usually find that cheap mics have higher S:N ratings than expensive mics that were built to higher specifications. Although the DVX is actually a very solid camera audio-wise, it happens to have a rather cheap and noisy on-camera mic. Whats more, putting a mic on top of the camera is about the worst possible place, since it'll be much too far from the subject to get a decent recording. Both of those factors are compounding to give you recordings that are so noisy that its tough to know where the tape noise begins and the sampling noise ends.

If you swapped out the on-camera mic for an inexpensive shotgun mic on a boom, the difference between underrecorded 12-bit and well-recorded 16-bit will bring tears to your eyes.
 
I didn't use the on board mic when recording in 12-bit, nor did I do it on purpose. Nor did I use the DVX. Did you read my post?
 
You will barely hear the difference because depending on how hot you normally record your probably only getting around the same number of effective bits and the high frequencies where this would show the most are being filtered out by the 32k sample rate. And your hearing is probably not close to flat anyway. BUT you don't want to record that way for a few reasons. First it's not the DV standard and somewhere down the line it will be an issue. And second is that anything you don't capture you loose forever. All through the filmmaking chain you taking small quality hits, every filter, non digital transfer, format conversion etc takes a little hit on your film. You want to start out with as much as you reasonably can so the hits don't end up noticeably impacting your film. Since you CAN tell the difference right off the bat you can count on a noticeable loss by the time the film is finished. So yes it will matter and it may matter a whole lot. Either way you want to convert that audio to 48/16 ASAP. The reasons are the compatibility issue I mentioned and because as you process the sound (changing a level in post is processing) it will degrade slower at the higher sample rate and bit depth. The conversion won't improve anything (since it's just going to fill in blanks for the missing bits) but the first time you change a level or ad a filter it's going to use the extra bits to give a smoother result.

And as to gain. The hotter you record (with out clipping) the better your sound will end up. The more information you send to post the better the files can be processed. If it's clean and loud lowering it in post will just make it cleaner. If it's noisy and loud then there is a lot to work with and it will be much easier to reduce the noise with out impacting the dialog.

The exception to this would be if the gain circuit on the camera adds a lot of noise. On DVX100 and better that should not be the case. The second "catch" is if you are shooting HD AND your camera records audio in a compressed format (rather than PCM audio) then you need to do some tests to find out what happens when you over-modulate. Some of the encoders on HD cameras "freak out" easily, so hot may not be the best choice in that case.
 
jpastuch said:
I didn't use the on board mic when recording in 12-bit, nor did I do it on purpose. Nor did I use the DVX. Did you read my post?
Click to expand...
My apologies. When you mentioned testing 12-bit recording on your DVX I assumed that the original camera in question was also a DVX or comparable. Still, bits are bits; the math doesn't change when you go from one camera to the next. Anyway, to address your original questions in a more pithy manner:

1) Yes.

2) Yes, provided the incoming audio doesn't clip.
 
Thanks, again, I didn't MEAN to record in 12-bit- dumb mistake on my part. I changed batteries and for some reason it reverted to 12-bit mode (I guess it's the default for this one) and I didn't even realize until later. All I was saying was the camera I used was 1CCD camcorder with NO audio meters or adjustment on camera...and the mic wasn't amazing either.

While I know I'll lose quality, I'm just wondering if it's worth freaking out over and trashing the project or if it's still technically salvagable...it's sound fairly clear for what it is and I don't plan on showing it anywhere besides youtube and SD TVs.

So if anyone has some (somewhat) reassuring stories that'd be cool. Again, I wasn't recording in HD or anything, or with any kind of amazing sound set up to begin with, so while there will be less quality as opposed to 16-bit, I'm hoping it's not going to be AS glaring. To repeat, I'm NOT condoning having awful quality from the start- just, obviously in my instance there ain't much I can do but salvage in post. If the dialogue is clear and understood, even if it's flatter, I'll be happy (this time).

And I learned something about the mic gain, though I do have to say with the mic I had, I noticed a LOT of noise increase on the boost....not an amazing mic either, so...I avoided it this time. Should I generally use the gain boost, except in noisy situations?
 
jpastuch said:
While I know I'll lose quality, I'm just wondering if it's worth freaking out over and trashing the project or if it's still technically salvagable...it's sound fairly clear for what it is and I don't plan on showing it anywhere besides youtube and SD TVs.
Click to expand...
I think you'll find that TV viewers are no more forgiving of an SD program's audio than they are of an HD program's. Youtube viewers tend to have lower expectations, audio-wise, but that's changing. Distorted, noisey, unclear or muffled audio will alienate viewers pretty quickly.

Surely that won't mean scrapping the entire project, though. If you're unable to coax the tracks into a form that you find acceptable, you still have the option of rebuilding the tracks with effects and ADR.

jpastuch said:
And I learned something about the mic gain, though I do have to say with the mic I had, I noticed a LOT of noise increase on the boost....not an amazing mic either, so...I avoided it this time. Should I generally use the gain boost, except in noisy situations?
Click to expand...
That's kind of the nature of gain: add another 10 dB, and everything gets louder, including both the dialogue and the noise. Cutting the gain on a recording may bury the mic's noise floor in the track, but you're introducing a lot more noise if you have to boost it back up to a useable level in post. You'll probably get better results by addressing mic choice and placement; use a mic with a higher S:N ration and/or get it closer to the sound source you're recording.

Any given mic's self-noise level is (mostly) a fixed amount, regardless of what you're recording. Get that mic in closer to the sound source, and you won't have to turn the preamp up as high. If you've already positioned the mic at the optimal distance and you're still experiencing unacceptable levels of noise, then you should probably switch to another mic.
 
Honestly, I'm not worried about the audience forgiving anything- as long as they don't have to struggle and they can hear it clearly. By that I mean, no, it isn't professional- I went in with very very bad equipment. Again, I'm not condoning laziness or mistakes- but the audio and picture were always going to be not amazing. Very low-tech.

AGAIN, I'm just saying...this was never going to be an audio technician's dream, even if I recorded in 16-bit. Sorry if that makes some people mad. I don't plan on making the same mistake again though, so I guess lesson learned.

And also, I couldn't see doing ADR possibles. Lots of dialogue in a very long take- the point was for the actors to sound natural. So...
 
jpastuch said:
so while there will be less quality as opposed to 16-bit, I'm hoping it's not going to be AS glaring. To repeat, I'm NOT condoning having awful quality from the start- just, obviously in my instance there ain't much I can do but salvage in post. If the dialogue is clear and understood, even if it's flatter, I'll be happy (this time).
Click to expand...


Dude... no offense, but, use those things on the sides of your head. There's nothing anyone here can tell you about whether your dialogue is "clear and understood", or whether your audio is "glaringly" bad... you have the audio there to listen to... just listen to it...


I mean, you're not gonna do this again, right? You just need to know if what you have already sounds good...?
 
Haha, sorry. I'm neurotic, and when I learn that I've made a technical mistake that can't be fixed I beat myself up over it pretty bad. No, I won't do it again. I just keep thinking I've destroyed my entire project and this is the end of the world. I did this last week when I realize a fraction of a shot I accidentally captured while filming last week was much better than everything else I had planned out, and I was trying to think of ways to stretch that fraction out longer, like I had missed some sort of oppurtunity. That's how I feel with this, even though there are much worse things.

Sorry, Im a really neurotic guy. I want to learn, and until recently I've been doing that by making terribly dumb mistakes. All of my nightmares will have 12-bit audio now.
 
No worries. As it happens, I'm currently working on a feature where 80% of the dialogue was underrecorded by about 16 dB. Pulling dialogue out of the mud isn't much fun, but its still better than dealing with clipped recordings. Its better to be 16 dB too low than 1 dB too hot.

If you want a real nightmare scenario, try this on for size: a couple years ago I was cutting dialogue on a documentary where noone bothered to put fresh batteries in the wireless. As the film progressed, the interviews became increasingly distorted. By the end of the film, it sounded like they'd hooked a talkbox up to a masonry saw.
 
Yes, no worries... if I can say that to a neurotic guy...

That's the issue... trust your ears... or work on trusting them, and you'll be able to put these worries to bed faster.

I will interject, that in addition to the stupidity of including this mode, many camera manufacturers (especially of consumer cams) set it as the default. I often do compilations or edits for other people, and I can't tell you how often it's at 12-bit. And these are usually classical-music related... ugh! But, frankly, as you say, given the wretched nature of audio recorded from too far away, next to snuffling coughing audience members, with AGC, through crappy built-in mics and preamps... 12/32k audio is the least of the problems....
 
You must log in or register to reply here.
Back
Top