Suggested Avg RMS Level of Mixes

[Charles Dye]

Originally Posted by Doylemusic

Interesting. Guess I'll be checking the calibration tonight....

Hey Charles, so is PAZ RMS correct or off in PT?

I've not tried to use PAZ to measure average rms?

Does it have the ability to do that?


Let me be very clear...


As far as I know, the only way to measure average rms is in NON REALTIME.

IOW... by selecting the file + then using the measurement software to calculate the avergae rms over the entire length of the file or song or mix.


The way PAZ works, as far as I know is that you play the audio "thru" the meter + it tells you what the rms is.

BUT... it does not give you the average rms of the file that considers the average over the entire file.

If it does, I'm not aware of it. So, please correct me.

[Charles Dye]

Originally Posted by IntelDoc

Does that mean we are all slowly going DEAF?

nah... just that much of today's music is dynamicless.

[IntelDoc]

Originally Posted by Charles Dye

The way PAZ works, as far as I know is that you play the audio "thru" the meter + it tells you what the rms is.

BUT... it does not give you the average rms of the file that considers the average over the entire file.

If it does, I'm not aware of it. So, please correct me.

That is the way I have always used it. If I want the "average" I then throw it into Wavelab and analyze it that way...

Still think I am going deaf...

[tannoy]

Originally Posted by MacGregor

Hmh...
I use Sony Soundforge which calculates and shows peak and average RMS for both channels of the complete file individually.
And to me -11 db measured in Soundforge sounds more on the hot side.

I could imagine that the inaccurate results are a result of programming in a way that doesn't cause too much hype. Maybe they've been using some kind of weird m/s calculation to avoid double integration of channel analysing.

Beside that, 12.97 dB rms at the left channel and 12.94 dB rms at the right channel is what my tiny Wavelab analyzer is saying about the average rms at Mixerman's raw mix. Charles was measuring 10.05 dB. So obviously there are some more differences in interpreting the values. Considered that the 10.05 dB are a total value of L and R calculated in an accurate way, there seems to be little possibility to compare this number with the ones given by other analyzers.

Or do I have to substract 3 dB from each side of the signal instead of 6 dB - at least the result would be in the ballpark of the value given by Charles (9.97L / 9.94R to 10.05). Any ideas ?

Btw, do we get off topic here ?

[jaygro]

Originally Posted by Charles Dye

I've not tried to use PAZ to measure average rms?

Does it have the ability to do that?


Let me be very clear...


As far as I know, the only way to measure average rms is in NON REALTIME.

IOW... by selecting the file + then using the measurement software to calculate the avergae rms over the entire length of the file or song or mix.


The way PAZ works, as far as I know is that you play the audio "thru" the meter + it tells you what the rms is.

BUT... it does not give you the average rms of the file that considers the average over the entire file.

If it does, I'm not aware of it. So, please correct me.

That's how I thought it worked.

On a Mac what can I use to find the average RMS?

jay

[Doylemusic]

Originally Posted by Charles Dye

The way PAZ works, as far as I know is that you play the audio "thru" the meter + it tells you what the rms is.

BUT... it does not give you the average rms of the file that considers the average over the entire file.

If it does, I'm not aware of it. So, please correct me.

So, if I make note of the quietest RMS in PAZ while playing a song and take the final RMS as shown in PAZ at the end of the song (which it will always show the loudest RMS) and add those and divide by two, will I have "thee RMS"

[CaptainHook]

Originally Posted by Charles Dye

It's avg rms is -10.05 dB.

Here's the file. Measure it with your software. If you get an avg rms of -13.05 dB, your software is using the incorrect method to measure avg rms.

I tried the rough mix with audioleak.
It said -12.1 a-weighted and -7.1 unweighted.

Any software suggestions?

[Knastratt]

Originally Posted by tannoy

I could imagine that the inaccurate results are a result of programming in a way that doesn't cause too much hype. Maybe they've been using some kind of weird m/s calculation to avoid double integration of channel analysing.

Beside that, 12.97 dB rms at the left channel and 12.94 dB rms at the right channel is what my tiny Wavelab analyzer is saying about the average rms at Mixerman's raw mix. Charles was measuring 10.05 dB. So obviously there are some more differences in interpreting the values. Considered that the 10.05 dB are a total value of L and R calculated in an accurate way, there seems to be little possibility to compare this number with the ones given by other analyzers.

Or do I have to substract 3 dB from each side of the signal instead of 6 dB - at least the result would be in the ballpark of the value given by Charles (9.97L / 9.94R to 10.05). Any ideas ?

Btw, do we get off topic here ?

I got -13.07 with RMS buddy on Mixerman's. My own is -19. Which should mean that my is like -16 IRL. Am i fucked?
Or a hero?

[JariNi]

You guys are metering with different analysis window size. RMS Buddy lets you choose the window size while Audioleak defaults to an unknown fixed value.

RMS meters perform a root-mean-square calculation to derive an average power level over a period of time that is called an RMS window.

Smaller window sizes make the measurement more responsive to short-duration events and lowlevel peaks, while larger sizes apply more smoothing.

Massey has a free RMS meter for PT users. http://www.smassey.com/plugin.html

[Charles Dye]

Just to be clear... the offset is 3 dB, not 6 dB, as Knattsratt demonstrates above.

Sorry for the confusion.

[MacGregor]

Originally Posted by Charles Dye

Avg RMS Calibration

To also help you calibrate your avg rms software I've measured the MM's mp3 rough mix of "Aren't You Proud" (aka "622") that was posted here.

It's avg rms is -10.05 dB.

Okay, from memory Soundforge calculates something around -12.9.. for each individual channel, so if that corresponds to your -10 your original statement of -11 db as a target would translate to about -13.5 db calculated in Soundforge.
I can live with that. And won't report you to the TurnMeUp court!

Mac
.

[tannoy]

Hi there,

Originally Posted by Charles Dye

Apologies... the offset is 3 dB, not 6 dB, as Knattsratt found out

aaah...YES! Confusion finally terminated.

Originally Posted by Knastratt

I got -13.07 with RMS buddy on Mixerman's. My own is -19. Which should mean that my is like -16 IRL. Am i fucked?
Or a hero?

13.07 - 10.05 = 3.02! Congratulations!!! Referring to the fact that you're closest to the 3dB, I would say you're a hero.

Best wishes,

Marco.

[Charles Dye]

Again... the difference is exactly 3 dB. Not 3.02... etc.

If you get an avg rms of approx -13.xx dB with the MM's ruff, then just offset whatever avg rms reading you get from your measurement software by 3 dB.

Here's the backstory behind the offset...

A number of software apps actually measure RMS incorrectly. They do it in a way that mathematically appears to be accurate to the software programmers. But they're not incorporating the 3 dB offset established by AES many years ago for measuring Avg RMS of stereo signals.

[tannoy]

Hello,

Originally Posted by Charles Dye

Again... the difference is exactly 3 dB. Not 3.02... etc.

If you get an avg rms of approx -13.xx dB with the MM's ruff, then just offset whatever avg rms reading you get from your measurement software by 3 dB..

....hm...confusion not terminated though. I fully understand what you're saying. But why is it that Knastratt's RMS Buddy says 13.07dB, which makes 10.07dB respecting the offset or wavelab saying it's 12.97, which is 9.97 after offset consideration and your's saying it's 10.05 dB ? I consider the value given by you as accurate. So does that mean that our applications are not as accurate in measuring this or is your's doing it in a different way, e.g. calculating some total stereo value from the 2 channel values - ours are only delivering 2 values - L/R.

Originally Posted by Charles Dye

Here's the backstory behind the offset...

A number of software apps actually measure RMS incorrectly. They do it in a way that mathematically appears to be accurate to the software programmers. But they're not incorporating the 3 dB offset established by AES many years ago for measuring Avg RMS of stereo signals.

Thanks for that info .

Best,

Marco

[Charles Dye]

Originally Posted by tannoy

...why is it that Knastratt's RMS Buddy says 13.07dB, which makes 10.07dB respecting the offset or wavelab saying it's 12.97, which is 9.97 after offset consideration and your's saying it's 10.05 dB ?

Presumably, the difference is the in the software's window setting.

But as you can see in your example above the difference between the measurements is 1/10 th of a dB.

Which is pretty friggin accurate.

So, my advice is... don't worry about it.


Remember, my original suggestion was an avg rms of approximately -11 dB.

We got lot's of room in there.

Go -10 dB.

Go -13 dB.

Whatever sounds best to you.


I just would like to discourage you from crushing it to an avg rms of -6 or -5 dB.

[tannoy]

Good idea to create a dedicated thread.

Originally Posted by Charles Dye

Presumably, the
difference is the in the software's window setting.

Originally Posted by JariNi

You guys are metering with different analysis window size. RMS Buddy lets you choose the window size while Audioleak defaults to an unknown fixed value.

RMS meters perform a root-mean-square calculation to derive an average power level over a period of time that is called an RMS window.

Smaller window sizes make the measurement more responsive to short-duration events and lowlevel peaks, while larger sizes apply more smoothing.

Thanks to both of you for explaining!

[Audioboy]

I am using Cooledit 2k to do the analysis. It gives me an option of how to set the 0dBFS- either sine wave or square wave.

According to the help popup:
"This setting determines whether the RMS readings will be compared to a full amplitude sine wave or a full amplitude square wave. The comparison point determines the point at which RMS power is measured. When audio is measured 0dB RMS, it is the same as a maximum amplitude wave, either sine or square, depending on this setting. A full amplitude square wave is 3.01 dB higher than the full amplitude sine wave."

The same popup help thing in Adobe Audition 1.5 (never version of the software), says the same thing, except that a square wave is 3.02 dB higher.

So, it would appear I have the option to measure either way. The sine wave method seems to be the method Charles is describing he is using, and the square wave is what he is warning about as not being accurate.

To help illustrate, with the 622rough file, using the sine wave measurement, avg RMS shows as -10.36 left,
-9.91 right.
Using the square wave measurement, avg RMS shows -13.37 left, -12.92 right.

Again, these measurements were done with CE2K, which is a bit outdated. But it does show the two measurement schemes.
When I ran the analysis in Adobe (at work last night), the numbers were closer to what Charles listed. I cant remember them exactly, just that they were even closer than CE2K (at home) is showing me.

[Numb Thumb]

I think you guys are going nuts over nothing. I work in television post-production, whereas you have to have an average level of -20dB RMS and no peaks above -10dB FS or the mix might be rejected or simply be squashed in to oblivion by the TV broadcast limiters. Music is totally subjective. Just mix it as you like. My mix for this track has an average of -10 FS to-8dB FS with peaks at -1dB FS. But who cares? There are no rules with music. Just make it sound good. There is no better criterion than the human ear.

[IntelDoc]

Originally Posted by Numb Thumb

I think you guys are going nuts over nothing. I work in television post-production, whereas you have to have an average level of -20dB RMS and no peaks above -10dB FS or the mix might be rejected or simply be squashed in to oblivion by the TV broadcast limiters. Music is totally subjective. Just mix it as you like. My mix for this track has an average of -10 FS to-8dB FS with peaks at -1dB FS. But who cares? There are no rules with music. Just make it sound good. There is no better criterion than the human ear.

Ya, but then a commercial turns on and it is like the broadcaster went to take a dump and my home system goes up 15 db yet my remote was sitting cozy on the coffee table?

What gives with that???

[Trazan]

Do you guys measure average rms by analyzing complete songs? Since this average will be a result of song arrangement etc and not necessarily say much about the amount of limiting, I'm not quite sure why you would do it this way...

I completely skip anything but loud parts when measuring. I measured MM's "622 rough" to be around -10.5 dBfs rms in those loud parts, which is pretty loud in my book...but probably 2-4 dB less than the real offenders.

But...with this -3dB offset and all, Charles, are you actually saying that "622 rough" is rather -7.5 dB rms in the loud parts? Which means those really squashed ones may be -4dB rms?