How Do I Read Level Meters?

[Dr. Varney]

I'm mixing this with ears, not meters... but I am curious:

I'm told I'm supposed to set my volume levels around -12dBFS to leave some headroom for mastering. All of the tracks in this piece are voices and when I do look at the meters, some of them are hitting around the -12 mark on the DB meter, whereas others are hitting way above -6 yet sound the same volume as the ones reaching -12.

I don't get it.

Dr. V

 

[mshilarious]

Peak or RMS meter? If peak, well, probably somebody has some plosives or sibilance or something . . .

 

[SouthSIDE Glen]

How can a signal that reaches -12 be louder than one hitting above +6? It doesn't make any sense.

There's a few factors that could contribute to that:

1. Remember back in the thread on normalization how we mentioned that peak volume level has little to do with perceived loudness? Here is a nice case study of that very fact.

How loud something sounds to the human has far more to do with the average or sustained volume level of a signal than it does with just how high the occasional, temporary peaks rise. There's a good chance that the "oddball" vocal track has a more percussive character to it than the others, pushing the peaks higher in relation to the rest of the voice than the other tracks.

2. If it's the "husky" voice that's giving you trouble, there's a possibility that it has some low frequency content that you're not hearing properly on your monitors or cans because of weak low frequency response by whatever you're listening through. In the case of phones, they simply may not have the LF response or on loudspeakers is could be the speakers themselves or it could be room acoustics causing it. That lack of response could be causing you to hear the overall huskiness of the voice quieter than it really is.

If I had to bet, I'd bet on #1, but it could go either way (or both).

What you want is a meter that'll tell you the track RMS (a fancy version of "average volume") so you can see if it's the peaks vs. RMS issue or not.

G.

 

[Dr. Varney]

Peak or RMS meter? If peak, well, probably somebody has some plosives or sibilance or something . . .

Eh? What's the difference between peak and RMS? What does RMS stand for?

Dr. V

 

[Dr. Varney]

I'm sorry, I had to change my message because I wrote it entirely wrong. It's not just one that's different - when I checked, they are all different. And I thoroughly misnamed the title, too. Apologies.

The bottom line is, I don't understand what I'm seeing when I look at the meters because they don't sound louder or softer than each other, but they're all hitting different meter levels. I don't know what I'm supposed to be doing with them.

Dr. V

 

[mshilarious]

"root of the mean squares". It's a statistical method for deriving an average power level. You take the value of each sample (which can be positive or negative) and square it (squares are always positive), add up all of the squares, and find the average, then take the square root of that.

Or you get an RMS meter to do it for you . . .

Like this one:

http://www.naiant.com/vst/vu.html

 

[Dr. Varney]

"root of the mean squares". It's a statistical method for deriving an average power level. You take the value of each sample (which can be positive or negative) and square it (squares are always positive), add up all of the squares, and find the average, then take the square root of that.

Or you get an RMS meter to do it for you . . .

Like this one:

http://www.naiant.com/vst/vu.html

Oh dear. Something tells me I'm supposed to know this before doing anything. Maths was never my strong point... I don't have a clue what any of that "root square" stuff means.

So RMS is a meter that shows an average of what's going in? What's the difference with a peak meter then?

Dr. V

 

[Dr. Varney]

Can I just turn turn a blind eye to all these meters and just use my ears instead?

Or could I get problems if I mix something too loud or too quiet?

Glen, I can hear the peaks - like a sudden push at the ends of some words and on certain consonants but the meter is just confusing me and not helping me very much.

Dr. v

 

[philbagg]

Can I just turn turn a blind eye to all these meters and just use my ears instead?

Or could I get problems if I mix something too loud or too quiet?

Dr. v

You should learn how to use meters. They're an aid to your ears, and were invented for good reason

 

[SouthSIDE Glen]

So RMS is a meter that shows an average of what's going in? What's the difference with a peak meter then?

A peak meter shows you what the level is at every instant, an RMS meter tells you what the average level is over a period of time.

It's like the difference between a chart that tells you the high temperature for every single day in a month (peak meter) and one that tells you what the average temperature was for the month (RMS). You might have a 100 degree day in May and only a 90 degree day in July, but chances are that July will average out to be a hotter month (unless you live south of the equator )

G.

 

[Dr. Varney]

Ah! Gotcha. Thanks, Glen.

MShilarious, I downloaded those meters but the needle doesn't move at all and I don't know how to callibrate them. I've read the web page but can you shed any more light?

It suddenly occurred to me... Where should the meter ideally go, in a stack of effects? At the top (to measure pre-FX) or at the bottom, (to measure the resulting level coming out the stack)?

I'm thinking it should go at the bottom, so as to show the processed signal level presented to the output.

Dr. V

 

[mshilarious]

Well, that's a philosophical question, really. But as for calibration, raw vocal tracks peaking at -12dBFS are probably -24dBFS territory, and you probably won't hit 0VU too often. But you probably don't want the final mix to end up there, and you'll probably use some compression to whack down those stray peaks (editing is another useful method if you don't have too many of them).

 

[Dr. Varney]

Hmm... I think I'm getting the hang of calibrating these. Of course, I had to look at the DAW's native plugin ones to do this, so I'm still not sure of which way is up but at least I've got the LEDs and these new needle VU ones registering pretty much the same result. I'm assuming any meter will be accurate enough to keep it approxomately real? I mean - I really dont think I'm in any danger of clipping as such but one or two tracks are getting a little close to +0 for comfort.

If RMS shows change over the long period and peak shows the current peak, the LED ones I've been using up 'till now have a little bar which stays in the region, some time after the peak has fallen. I take it that sticky bar is showing me the RMS?

Thanks.

I think I've come to the point where I really need to learn how to start using a compressor. I've been putting it off because it looks like a bitch to learn.

Dr. V

 

[Dr. Varney]

Or you get an RMS meter to do it for you . . .

Like this one:

http://www.naiant.com/vst/vu.html

I downloaded these meters and they do not work in line with the existing meters in my DAW. No matter how I trim them, they show massively varying levels of discrepancy.

By selecting a small area of the playlist I can get the needle to hover and compare this with the fixed position of the RMS (?) bar on the LED meter. When I move to another part, the calibration goes out by about plus or minus 10dB and has to be done again. How can I rely on a meter like this? I have no idea whether it's true or not, without something to compare it with.

I wouldn't have even thought of this issue, had this meter not featured options for callibration. That part has completely thrown me. What am I supposed to calibrate it against? It makes absolutely no sense to me.

Well, that's a philosophical question, really.

No, it was a technical question. Multiple choice doesn't automatically make it philosophical.

Dr. V

 

[NYMorningstar]

Why does this thread remind me of Sesame Street?

 

[mshilarious]

If RMS shows change over the long period and peak shows the current peak, the LED ones I've been using up 'till now have a little bar which stays in the region, some time after the peak has fallen. I take it that sticky bar is showing me the RMS?

No, that sounds like a peak meter that has a peak hold feature. A peak meter isn't going to show the same result as an RMS meter, because peak and RMS aren't the same thing.

You set the meter's calibration to where you want it to end up, and then you manipulate your source (whether through gain change or compression, or usually both) until the needle is happily bouncing around 0VU. For choral music, that end result is probably -18dBFS, although for more compressed a capella style music it might be -15dBFS. This you would do on the master bus; you don't really need an RMS meter on every track. You might need individual compressors on each track though, or at least a few of the more dynamic tracks.

Then you use the peak meter to make sure you don't have any overs, drink a beer and call it a day

 

[SouthSIDE Glen]

"root of the mean squares". It's a statistical method for deriving an average power level. You take the value of each sample (which can be positive or negative) and square it (squares are always positive), add up all of the squares, and find the average, then take the square root of that.

Or you get an RMS meter to do it for you . . .

Like this one:

http://www.naiant.com/vst/vu.html

Jon, great job on the VU VSTs!

I don't want to hijack anything here, just a couple of related sidebar questions that I've always had regarding this subject:

I've always known that RMS is a type of average over time, but I've never known how much time when measured in real time as in a live RMS meter. Is there a standard for just how many samples or length of sample time are needed or taken to calculate the RMS? And if not, what sample length are you using and how did you settle upon that figure?

G.

 

[mshilarious]

Well a VU meter isn't a strict sample-based RMS calc (since it's analog!), although the VST is. 300ms is the old standard for ballistics. The VU VST has adjustable ballistics via the knob, but I think 12:00 is pretty close to 300ms, I forget.

I am also doing a compressor VST that has three (!) VU meters, I have a handful of people alpha-testing it That has a slightly better VU algo that is also a bit lighter on CPU, so when that's done I'll redo the VU VST with those changes

 

[SouthSIDE Glen]

Well a VU meter isn't a strict sample-based RMS calc (since it's analog!), although the VST is. 300ms is the old standard for ballistics.

Maybe I'm confused (no surprise there), but I don't get the connection between meter ballistics and actual RMS measurement.

Ignoring ballistics for just a second - let's pretend you had an instant-read RMS meter instead of one trying to emulate an analog VU meter - would that be showing the RMS measurement for the last 300ms of signal?

G.

 

[Dr. Varney]

Why does this thread remind me of Sesame Street?

I don't know - why does it remind you of Sesame St? One episode in particular or just generally? Any You Tube URLs you could point us to...?

MSHilarious, I didn't know you had personally coded these VSTs. I didn't mean to sound too critical and thanks for your help in understanding them - I'll certainly give it another go.

Cheers

Dr. V