Levels???

graywolf

graywolf

Well-known member
I know a bit about audio, but not a lot about video. A youtube about setting mixer levels has me confused. Not going to send you there because, I do not want this to be about him particularly.

OK, supposed expert, on how to on how to set your levels on your $1000 mixer feeding your $150 digital recorder (I always wonder about that kind of setup). Technique seems about right, but...

His youtube audio voice levels showed +10db on my vu meter (My workstation uses the Behringer Xenyx 1204 USB as a sound card feeding audio to a pair of 5" monitors). Now that is too hot, in my opinion, even for tape, much less digital as a recording level. But he had it as an output level for his voiceovered video. I had to turn my control room fader all the way down to keep from blowing the monitors.

Now, I learned long long ago with tape to record hot with peaks hitting -7 db to overcome tape hiss (that frying bacon sound), but then when I got into digital I learned that it is not good to run your recording peaks above unity (0 db). And I would think the mix output on the video should be somewhere about -20 db average and voice peaks more like -10 db (Yes, I know the Euro/PBS stardard is -24 db, but I usually am not working to standards) with music/fx peaking around 0 db. I, personally like soft voices down around -30 db, which puts them 16 db or more above the noise floor on my admittedly not too expensive equipment.

So, is the expert, not such and expert, or am I way off the mark myself?
 
graywolf said:
I know a bit about audio, but not a lot about video. A youtube about setting mixer levels has me confused. Not going to send you there because, I do not want this to be about him particularly.

OK, supposed expert, on how to on how to set your levels on your $1000 mixer feeding your $150 digital recorder (I always wonder about that kind of setup). Technique seems about right, but...

His youtube audio voice levels showed +10db on my vu meter (My workstation uses the Behringer Xenyx 1204 USB as a sound card feeding audio to a pair of 5" monitors). Now that is too hot, in my opinion, even for tape, much less digital as a recording level. But he had it as an output level for his voiceovered video. I had to turn my control room fader all the way down to keep from blowing the monitors.

Now, I learned long long ago with tape to record hot with peaks hitting -7 db to overcome tape hiss (that frying bacon sound), but then when I got into digital I learned that it is not good to run your recording peaks above unity (0 db). And I would think the mix output on the video should be somewhere about -20 db average and voice peaks more like -10 db (Yes, I know the Euro/PBS stardard is -24 db, but I usually am not working to standards) with music/fx peaking around 0 db. I, personally like soft voices down around -30 db, which puts them 16 db or more above the noise floor on my admittedly not too expensive equipment.

So, is the expert, not such and expert, or am I way off the mark myself?
Click to expand...

You are probably mixing up dBFS and DBu, which are not at all the same things, even if they use the same units (dB).

Your expert was probably using a Sound Devices mixer, the meter of which shows you dBu (it's a holdover from the old VU meters of tape days). Running that into a recorder whose meter shows you dBFS is probably the reason for your confusion.

When I use my own SD MixPre-D into my camera, I have the same situation more or less. I use the MixPre-D's calibration tone to set the camera's level to -20 dBFS, then forget the camera -- all control is done from the mixer from then on. And I set my dialog peaks to run about +8 dBu on the mixer (note the "+"), which results in the recorder (my camera) seeing peaks around -12 dBFS (note the "-"), which is just what you typically want. Then I have the mixer's limiter's threshold set to +18 dBu (again, that's a "+"), which gives them 2 dB of headroom before the mixer starts clipping. The result is that my recording chain is virtually unclippable.

Works like a charm. And it sounds like what your expert is doing.
 
Location (capture) levels and broadcast levels are two different things. Most recorders these days use full scale peak meters where 0.0 dBFS is the highest level w/o clipping. Theoretically, this yields the best signal to noise and quality but one must take headroom into consideration. Mixers and such can still be VU or other scales and vary by manufactures. The EBU R128 standard and AES A/85 broadcast levels are simular.. but use the loudness 'K' scale, which is quite different from, dBFS, VU or other measurements and actuate meters are needed (if its not within specs, a submission is usually rejected by the network) Typically the b'cast standard is -23LKFS (EBU) and -24LKFS (AES), there is also parameters within the LU scale, such as max short term, momentary and integrated. The EBU meters even use a noise gate, so the long-term integrated levels are not skewed abnormally low. For web based videos, DVDs and such, it's currently 'recommended' to be around -18LKFS (integrated). But there's really no enforcement, except the laws of physics (clipping).
Music CDs on the other hand are usually maxed out to 0dBFS and even clipped (loudness war). I use a dynamic range meter for music mixes these days to confirm I'm not squashing things excessively.
 
Last edited:
My set up is almost exactly the same except I set the limiters to leave -3dB.

Digital doesn't have any tape hiss so you do not have to maximize levels for that reason and digital clipping is very bad and pretty much un fixable. So if you have good limiters then it's fine to push the level. I tend to record pretty hot. But you can with out sacrificing too much also record lower if you don't have some clip protection in place.
We are talking about video here so I would highly recommend recording at 48/24. The added bit depth will help in post.

Bruce basically said what I would have about scales. You actually mentioned three different ones in your post. It is complicated when you are talking about Apples and Oranges, and limes... Because you have to translate them to know where they are relative to each other. Not only are the scales different but what they are aimed at measuring can be different. dBfs is all about level and how close to digital clipping you are. VU is about perceived loudness and tells you very little about you digital level unless you know what it's zero was calibrated to. In truth you may have been saying VU out of habit when the meter was reading something else but?
 
Well, doing some research shows it is about as confusing as solid state electronics. I did know that dbu, dbfs, and vu, not to mention dbV are all different. However, it seems that dbfs is not directly related to the others, although if you know just which FS scale is being used you can approximately relate it to the others. GAGH!

BUT! My problem with the "experts" video is that his choice of final output for a dialog level was apparently zero dbfs. To me that shows that he knows less about film/video than I do. Treating it as Rick says Music CD's are treated.

Except for the input gain (trim), my system is set for unity gain. There is no way I should have to turn down the volume for conversational level audio in a video. AT LEAST TO MY WAY OF THINKING!

However, what you guys are saying is helping me understand the technicalities better.
 
Regarding the Sound Devices mixer meters, I often use the full scale measurement method, feeding the recorder a +20dbu reference level and setting the recorder just below 0dBFS, (which is a little more accurate than the 0VU = -20dBFS calibration) With the mixer's limiter set to +17dB which usually allows for unexpected peaks.. and limiter overshoot which can sometime occur on HF transits even with the SD limiter's fast one millisecond attack time. Not to be more confusing, but when feeding a recorder from my cart-based Mackie Onyx, I prefer to use -7dBu on the Onyx with the recorder's meter set to -20dBFS. The Onyx does not have any internal limiters but has a high max output of +27dBu. Same for the ISO tracks which are via analog direct outs via the Onyx' DB25 connectors.
 
dB (deciBels) is a RATIO, not an absolute measurement. If you want to use it to measure actual values, you must always have a REFERENCE point. (Like dBV or dBu or dBmw, etc.)

dBFS (Full Scale) is absolute. It came about with the digital era where there is no concept of the "headroom" we had back in the bad old analog days. When you hit the maximum value of the analog to digital converter that is an ABSOLUTE limit beyond which there is no "there" there. It is equivalent to slamming into a solid rock cliff face.

Perhaps the only source of confusion is between old-school gear that is labeled in the analog tradition (including digital Sound Devices gear) and how you relate that to a digital recorder where dBFS tells you absolutely where you are at all times.
 
Still you guys are talking about recording levels. While I am talking about mix playback levels.

A scenario: two people are talking in your war film, the music hits a cresento telling the audience something is about to happen just before the bomb goes off. Now if those people are talking at 0 db, the music and bomb are going to be kind of lost in the mix. You do not run your dialog at 0 db, or +7 db, or +24 db because then the music and the bomb mean nothing.

For the same reason you do not run your voiceover at 0 db, you run it at maybe -20 db. Unless you are supposed to be yelling then you run it at maybe -10 db.

You do your recording at a high level to minimize noise, but when you mix it you dial it back to the proper levels for the final mix.

You do not make the poor guy looking at your video have to turn the volume all the way down to keep his neighbors from banging on the wall!
 
graywolf said:
Still you guys are talking about recording levels. While I am talking about mix playback levels.
Click to expand...

Not how I read your OP where you said "how to set your levels on your $1000 mixer feeding your $150 digital recorder". That made me think you were talking about dialog capture. If you are in fact talking about post, then I can understand the confusion.

graywolf said:
You do your recording at a high level to minimize noise, but when you mix it you dial it back to the proper levels for the final mix.
Click to expand...

Yes. There are no real standards for this. Levels for the production tracks are a Sound Design decision; the final tracks are often comprised of dozens if not hundreds of layers, each layer's level set against the others to create the final track. That way the ringing telephone effect can be heard against the sound track, but below the dialog, and the distant dog barking can be a nearly subliminal suggestion that "something bad is coming". If the levels are too high it's an annoying dog, and if too low a missed cue. But if it's right, it's ominous (or whatever it's supposed to be to support the story).

That said, about the only time dialog is the loudest sound is when there is no background at all. Just like a face is seldom the brightest thing on screen.
 
I guess ", but..." was too subtle (friendly grin).

So, are you saying that in your voiceovers you would set the voice to +10 db VU output? That is what I am trying to understand. To me that is shouting at your audience.
 
Meter scaling varies according to what device or software you are using but for the three most common ones in the UK if you apply a 1khz line up tone at .775v the following applies for broadcast:

dbfs -18
PPM 0dbu (ppm scale 4)
VU -4

The max level for broadcast mixed content is PPM +8db so the following will then show

dbfs -10
PPM +8dbu (ppm scale 6)
VU +4

If I am doing on line content you tube type I tend to mix hotter by another +6db so this would then show on the three meters

dbfs -4
PPM +14dbu (off the ppm scale)
VU +10

I also have some behringer mixers and their meters may well show +10 for an on-line video that is mixed to the above, I think final cut pro also follows old VU meter scaling so would also show +10.
 
Last edited:
We keep Going back to the "Bel". As Richard says, it is a ratio, in fact it is a logrhythmic power of ten ratio. Zero Bel is 1:1, One Bel is 10:1, Two Bel is 100:1, and so forth. Of course the Bel is too large for most measurements so we usually use 1/10th Bel (deci-Bel, or db). Of course the Bel is actually a power measurement, so when we use 1/2 the equation "Volts" or "Amps" instead of "Volts x Amps (watts)" it shifts to 0,20,40,60,... db.

Confusing as hell, but I learned all that 55 years ago.

I seem to have two scales to deal with here VU on the physical meters and apparently dbfs in my audio software indications.

I guess, most people must have a problem with the deci-Bel as that seems to be what you all think I was asking about. It is not!

What I was asking about was where does a knowlegable expert set his "final" mix levels.

As an example, the following is the very first voiceover I ever did (also music and credits), before going into the hospital for a supposedly routine overnight procedure that I am now 2-years later just getting over, while trying to teach myself filmmaking:


So, is my audio in that way too low?
 
I set my final mix levels as I stated above in accordance with the calibration of my meters. They are scaled in decibels so that is why we all quote such readings. I can not tell you how loud things are on my speakers as you may have a totally different set-up to me.

That you tube video sounds too low to me but may be Ok for broadcast but as I don't have any reference metering on it I don't know as it is as loud as I turn my speakers up. Calibrated meters like f stops and shutter speeds for video using light meters or vectorscopes allow us to find the same readings regardless of other influences such as the volume of the loudspeakers or the contrast and brightness of a monitor.

But in comparison with this one that I mixed for on-line your clip sounds too low to my ears: https://youtu.be/8-KpV46AEyI

At a guess the difference is about the 6db I also mentioned.

So to summarise if I am mixing for broadcast I peak at -10dbfs and for on-line you tube etc -4dbfs, if I was to mix without any reference metering then I will not have a fixed point and as I also said it would be only relevant to how loud or quiet I have my speakers set.
 
Last edited:
Gary, by my VU meter, mine is peaking about 0 and yours about +4. Considering content (yours kind of urgent, mine kind of laid back), I would say we are at about the same levels. Of course the laid back voiceover is quite a bit lower on average, but that fits the content too.
 
Yes +10db VU for you tube or +14 PPM or -4dbfs as that corresponds to what I said above, how loud it is to the end user depends on the volume of their speakers.

The metering is only an indication and VU meters tend not to be what is used for accurate referencing these days.

Your behringer mixer is not what I would call a pro calibrated mixer and the VU's are really just there as an indication not an accurate way of measuring levels.

Have a look at this video I did and you will see that the behringer mixer VU does not even correspond to what I said above as -18dbfs seems to be about -6 ??? on their scaling : https://youtu.be/pJWKLbVZVU4
 
Last edited:
According to the download from YT, I'm getting an integrated loudness of around -22LKFS with a true peak level of -5dBFS, which is fine for web and DVD. It's right on the edge for the current ATSC broadcast levels: -24LKFS, plus or minus 2 .
Note: Prior to the CALM act, most NTSC network audio submission levels were @ -20dBFS reference, with program peaks not exceeding -10dBFS.
I'd prefer the music a little lower, but that's a personal preference and I'm basing it on cheapo laptop speakers.
 
graywolf said:
Gary, by my VU meter, mine is peaking about 0 and yours about +4. Considering content (yours kind of urgent, mine kind of laid back), I would say we are at about the same levels. Of course the laid back voiceover is quite a bit lower on average, but that fits the content too.
Click to expand...

Bingo and you hit the nail on the head there by using the phrase "about" !

As VU meters are not very accurate I would say that as I stated above my video is "about" +6db hotter than yours!

Therefore as I don't know how your VU's are calibrated that difference could even correspond to "about" +4 and +10 VU. The behringer is a domestic piece of kit so unless you have a true reference of what the meters are calibrated to with a .775v test tone they are purely for indication only.

Rick we use loudness meters here in the UK too but I still tend to mix to PPM and to my ears which seems to be OK for -24LKFS + or - 2.

Also I find that You Tube tend to normalise all their audio so a downloaded video may not be accurate enough for realative measurements.
 
Last edited:
Just a quick question what editing software are you using as it would be possible to put some bars and line up tone on the edit timeline and see what the level is on your edit meters and then set-up your behringer VU to a calibrated known source so at least you can get a common reference between the two?
 
I know this is confusing you but you probably should stop saying VU since with the possible exception of yourself nobody is likely to actually be using VU anymore.

You started this post referring to a video you didn't link to that was about setting production recording levels that you thought the levels on the video it's self were too hot. So it is actually fairly confusing about what you were asking. Post and production are separate worlds that don't often play by the same rules.

IF what you really want to know about is post levels then it's all about dBFS (as in digital levels) and you should probably look up LUFS because that is what most post is heading to if it isn't already there. It is the required measurement for NPR, PRI and quite a few broadcasters.

And probably look up Bob Katz and see why you never want to be above -3 dBFS.

As to what you learned 55 years ago, it's good background but it's based on a completely different technology and workflow and is likely to get you in trouble. I learned more like 45 years ago but it was pretty much the same stuff. Somewhere between 1983 and 1996 everything changed. I started film post in 95 and the "gold" of production was still 1/4" tape with a DAT as a safety. Post was edited digitally but the mix was all analog to analog masters. By 98 the analog dubbers were mothballed and video tape was mostly history, by 2000 everything was digital even the projection for mixes. Film sound is generally 10-15 years behind everybody else so the music people were there in the late 80's.

As a general thing production should record as hot as possible with out distortion or clipping. What that is depends a bit on what your gear is. You won't be laying down calibration tones etc, because everything is set from zero dBFS, there is nothing over zero so the sooner you stop thinking old analog and start coming to the new digital where there is no +dB the less confused you will be.

Post has delivery specifications that vary (A LOT) depending on who and what you are delivering too. Web stuff tends to want to be bouncing off of zero dB with compression and hard limiting so you don't clip. Yes it is screamingly hot but it's competing with a lot of other screamingly hot compressed to hell and gone stuff and that is what you will be asked to deliver.

I mix for the screen and then crank and squish for those who demand it. If you are in control then you can do whatever you want, but if you are delivering for a client (especially one under 30) expect a "it's kind of low" if you deliver a "good film mix".

Radio and a lot of broadcasters will spec a mix, and a LUFS number you have to hit.

One big point is that what comes out the end of the pipe (level wise) has very little to do with what went in. So in essence you started a discussion about production levels because you had an issue about deliverable levels, and they just don't relate much.
 
Last edited:
You must log in or register to reply here.
Back
Top