Db level peaks matter in this situation?

M

MeAgain

Carbonite Member
I know I will be beaten for this, but I'll ask anyway-

My sound guy (student, not pro) was using a 702 and shotgun mic (sennheiser 416) to grab some wild lines on set. He said he knew what he was doing with the db levels, but they seemed a little low to me. We got the mic very close (within a foot) and pointed straight at the talent's mouth. The voices are clear and rich, but are peaking at -30db in FCP when I bring them in.

I KNOW sound is much more complex than that and db is relative, but out of curiosity, would it have made a difference if he boosted the db level up to where it peaked at -12? This is normal room conversation, no yelling etc.. so it should be somewhat quiet, but I always feel like I have bad sound unless I have to lower the levels rather than boost.

Basically- I want to know if it'll be a problem to boost up the wild lines voices a bit...we did get it incredibly close and the voices sound great with no background noise, I'm just worried they were captured too quietly, though I have a feeling that makes no sense.
 
Assuming your FCP system it set up correctly
yes, he recorded your audio way too low.

As far as you bringing it up now, you really have no choice right?
It needs to come up..so bring it up..see what you get and hopefully the noise you are
also bringing up isnt too too bad.

If it is, then you will have to look into noise reduction plug in's.

Also make sure you let your sound person know their mistake so next time
it might not happen.
 
Yeah, well, I thought he was recording too low but he said it would be OK.<br />
<br />
WEEEEeeee



EDIT: I'd just to like to add how great it feels to mess up audio when you have $3000 worth of audio equipment pointed straight at the talent's mouth about 10 inches away.

GO ME!
 
Last edited:
Rippie,

Are you ever asked to show your levels you're recording when working with people for the first time?
 
Well... recording at a to low level is not that uncommon...

I had it also my first time. It's because of the rule "clipping digital = throw away".

So I think the soundguy was a little "afraid" of clipping, so he give it some "headroom" (little to much though).

But anyway... it's better to record it to low then clipping (because that's ...throw away..)

Noise you can handle with some noise reduction, EQ...

But better was to record it around -15db<->-10dB
 
Gohanto said:
Rippie,

Are you ever asked to show your levels you're recording when working with people for the first time?
Click to expand...
Never.
I see camera op's take peaks at the side of the camera all the time
to make sure everything is going ok. (timecodes moving, audio levels are bouncing, battery life is ok..etc)
but I have never been asked or questioned about it.

Honestly, recording audio at the proper level to a device is about as
basic 101 as audio gets in my opinion.

Its like teaching a guy how to focus the camera. You can turn a knob and there ya go but its what you decide to do with that focus,
when to rack to something else, when to draw your eye to another part of the screen, etc that makes the image worth watching

Audio is so much more than "record to -12 to -8db and bam thats it"
It's more of when do you feel its right to hit the limiter? When is it right to NOT ride the levels
and let dynamics play a part? When is it the dance of balancing out 2 voices with
only a single microphone smoothly with 1 fader?
And a huge one which just comes with a lot of time, is actually anticipating the volume and punchyness of
how a person is going to start a particular sentence even though you just met them 20 mins earlier so you do hit that volume pocket you're going for.

Can you imagine how boring dialog would be if you rode the levels so extreme that
a persons whisper was the same volume as when they were having a fight?
Which is what would happen if we stuck to the 1 sentence answer of "always record your dialog at xxdb"

I know this isnt want you asked, but just felt like explaining while its ok
to just "record at xx volume" and be done with it, it can be so much more if you let it.
Production sound isn't the most artistic dept on a tv/film set by any means
so enjoy what you CAN do with it.
 
Of course there has to be a difference between whispering and shouting.

And you don't have to turn the knobs as a crazy guy indeed.

But it was more like a "direction".... You don't have to focus blindly on those levels.

But you first have to walk....before running...
 
Here's the thing- I KNOW audio is complicated- AND important, and I know none of us are good enough to record pro sound or understand it fully. We gave it a shot with booming the scene the best we could, but we couldn't get the mic close enough the first time because the shots were wide and that couldn't be changed.

Thankfully, my script has a lot of short lines that I wanted to be spoken calmly and without a lot of inflection (as part of the atmosphere of the film, plus we're using non-actors). So I figured "We can redo the sound by looping each line and having them lip sync, that way we can get the mic nice and close and pointed DIRECTLY at the mouth, eliminating a lot of the noise". Which we did, and I advised the sound guy to make sure he gets good levels this time as this was the ideal situation.

So, I figured we have 4 grand of equipment that we only basically know how to use- but I also figured this is the IDEAL situation to get good, clean audio because even if it wasn't perfectly done, we had good headphones to monitor, amazing equipment, and we don't need to worry about picture- we're ten inches away!

But, I still screwed it up. I feel like I should have ignored whatever he was saying and boosted the levels myself- granted, the actors talk rather low in this, but I could have easily turned it down, no?

---------------------------------


So ultimately, here is where I'd like to learn one of the fundamentals that I can take away from this mistake...what, exactly, are the differences between boosting the db level on the 702 and the db level in soundtrack pro, considering my mic placement in this situation (which was the most ideal)? Did I basically lose voice information by not boosting the levels on the recorder? This is a dumb question, but I'd really like to understand the difference to help me with levels in the future. When I boost in soundtrack pro the only noise I get is some slight hiss, the voice sound clear, but the hiss is hard to get rid of without sacrificing voice quality.
 
I don't think you'll be so bad off, so long as there wasn't too much noise at the location. The 702 and 416 together have a pretty low noise floor, so by bringing it up in fcp or soundtrack pro you will be bringing up the noise as well, but I'll bet that since it's not much in the first place, it will be manageable. Use the "gain" audio filter in fcp.
If you were recording into the dvx it would be quite noisy, but still probably salvageable with some eq/ noise reduction.

Every device, (recorder, mic etc..) has some self-noise, or noise floor.
If you record with proper levels it keeps your dialog farther from the noise floor of the device, and since you won't need to increase those levels so much, if at all, in post, you won't be increasing the noise either. Also, digital recorders have greater sampling accuracy the closer you get to 0db, but that's not a huge factor anymore unless you're way down there like at -80db, which you should not do.

This story of low levels is a classic one, and I've found that the culprit is usually a combination of fear-of-clipping and headphone levels too HIGH. If the phones are too high, it makes you think the levels are great because, "hey, I can hear it loud and clear!" Turn those phones DOWN.

The limiters in the sound devices mixers and recorders are great, and don't be afraid to use them occasionally. If every other word is slamming the limiters, yeah, too hot, but once in a while they will be your best friend to get better levels overall, without clipping. I'm speaking mostly for the one-man-band sound guy that can't really boom and ride levels simultaneously.
 
If your signal going into the recording device(whatever it is) is too low, you are dipping further into the noise floor. Every device has a certain amount of noise, so if you can get your signal (mic volume) higher you have a better "signal to noise ratio". Once it's recorded you can't change that ratio. If you turn it up in FCP, you are turning up the noise as well. Of course you can use noise reduction filters, which are wonderful but they only go so far. It's best to get it right going into the camera.

That being said my EX1 is pretty quiet if I leave my camera audio levels at 50%-60%. Beyond that adds noise. If you are recording dual system with a dedicated audio recorder, recording at 24 bit gives you more headroom, and a more forgiving noise floor. So when possible record at 24bit.

And if you are the director, you make the call. If your inexperienced sound person is doing something wrong, have them change it. If you're recording one line or one word in a controlled situation there is no reason to leave a lot of headroom (-30db).
 
OK, I DID record 24-bit. When boosted up to around -15db-10db range, the voices still sound clean with some noticable (but not overbearing) hiss. What's the best route for this? I try to lower some of the highest frequencies in Soundtrack Pro, but it DOES take away very slightly from the voice's full bodied sound.

In general, do you think it's better to keep the slight hiss or to take away a bit from the voices? Or if it's just a simple hiss can I get a better fix with a plugin? I know this is a personal judgement call, but I don't want to mess with sounds more than I have already.

Also, I'm using "amplitude" adjustment in Soundtrack pro rather than FCP's gain, though i have to amp twice since I need about a 17-18 db raise on this, unfortunately. Don't know if I should adjust certain frequencies first, then bump it all up. I've read in Jay Rose's book that boosting certain frequencies around the 250 hz area can add some "warmth" for instance.


Anyway, is it worth posting a sample for you guys to listen on here?
 
I believe there is a noise reduction plugin in STP. Use that before trying EQ. You select a portion of the noise with no voice in it, then it pulls down only that sound. If you go to far it starts to sound a little like a flanger, or like little bird chirping artifacts. Try pulling all the noise out first, so you know the sound of going too far, then adjust to taste. It's best to just take the noise down a bit rather than remove it completely.
 
Yeah, I could use the noise print method, but I'm worried about the accuracy of me getting the noise. I'll mess with it though, and see what I can do.

I really wish I could back in time, grab the 702, and turn the dial up.
 
I always wind up catching a second of clothing rustle or breathing too, which I imagine messes up the print. I'm sure I'll be able to get it though.
 
MeAgain said:
In general, do you think it's better to keep the slight hiss or to take away a bit from the voices? Or if it's just a simple hiss can I get a better fix with a plugin? I know this is a personal judgement call, but I don't want to mess with sounds more than I have already.
Anyway, is it worth posting a sample for you guys to listen on here?
Click to expand...

Definitely worth it to post a sample here. Just as a picture tells a thousand words... and audio clip tells all to the guys here.

In general, I would say it's better to take away from the voices a little bit. I'm talking a Low pass filter that drops cuts out above ~ 15-17 kHz. Many older people don't even hear past 15k anyway! Plus, you're audience is not going to hear the 'before', so they won't know what they're missing.

But yeah, post a sample and someone will be able to tell you right away.
 
Thanks guys- sound is really important to me, which is why I'm obsessing over it and my mistakes- really dumb ones- and why I'm trusting someone who had used the 702 before- I figured he had more experience using it and knew more than me (which is true up to the point of db levels anyway).


Anyway, I'll post a sample in a couple of hours. Thanks.
 
OK- I saved both takes separately as stp files into a zip folder and uploaded it that way to avoid any type of compression or meddling problems since I'm clueless-

As you'll be able to see (if they open correctly), the first one is just the straight take. The second I lowered everything below 120hz and amped it up around 17db I believe.


Thanks, and let me know if I should upload it some easier way.

http://rapidshare.com/files/375972707/Compare.zip.html
 
MeAgian,

stp files can only be opened in Sound Track Pro so most audio people won't be able to listen to them. You'll need to convert them to WAV or AIFF files for anybody who doesn't have FCS2.

"Hiss" should be very removable with very little impact on the voice. You can use the NR in STP it should work very well on this kind of noise. There is also an old Behringer box (SNR 202) that works really well on hiss. A lot of folks who had analog gear have them kicking around (no resale value to speak of). Or depending on the length of the clips you might be able to get someone on this list to run them through something higher end like RX from iZotope.

I would try the STP plugin though. And yes it is important to be sampling ONLY what you are trying to get rid of. A trick I just learned is to EQ the noise sample and remove from it the freq.'s you need to preserve. I was trying to clean up an interview that was done on a shrimp boat and the low end was so pervasive that it kept being a big part of the noise print and chipmonking the voice. I just filtered out the low end from the noise sample and it worked like a gem and the voice was virtually untouched. I ended up doing two more passes. One just ti EQ down the BIG low of the motor and a second NR with a new noise sample that I eq'd everything out but the lowest lows. In your case I would grab the bit of just noise and copy that bit to another track and EQ everything below 6K or so out of it. Then use that as you noise print and the NR should leave your dialog alone and just attack the hiss.
 
You must log in or register to reply here.
Back
Top