I was testing the correlation of a sample from Splice (sample in stereo) and found out that the correlations meters from Levels (masteringthemix.com) and the one from Ozone (Izotope.com) don't show the same result ! and by far...

The levels shows a fluctuation from 0.4 to 0.6 and the Ozone shows a 1.0 all the time... Which one to trust in ?

However both show the sample is slightly on the right

 

Hi add this one to the test you are doing and if any two match out of the three it should help sort it out,
http://www.fluxhome.com/products/freewares/stereotool-v3

 

... or add Brainworx bx_meter; Voxengo SPAN has a correlation meter too.

 

bx_meter ftw.

 

The vectorscope shows the stereo information in the middle, the izotope one is correct.

 

It's exactly what i did, it's not clear because with the sample in question, i had to put it on a loop and play it during 5-6 times until the correlation rises up each time a little bit more to the +1... it's not dynamic or i don't know how to analyse it. I'll test it with Bx_meter i come back. I heard that i should also check the settings of Ozone, sometime it's checked on 5.1 and not stereo

 

Very interesting, I'd always trusted Ozone over everything else, looks like I'll have to run some tests myself now

 

I love the quality of Izotope, maybe it's only a bug from the installation, i should re-install Ozone to be sure... i think it's a bug

 

What is going on with these correlation meters, the Levels and Flux are the same and the izotope and BX are the same, but don't have the same value.... i don't understand

 

Last edited by a moderator: Aug 16, 2017

I'm an ignorant but could it be they have different methods of calculating it, but the end result over longer times is the same?
Have you tried different samples? Shouldn't the correlation be +1/almost +1, since it's almost all mid?

(I also use T-Racks Metering)

 

i don't know and with these 4 correlation meters i don't understand... the 2 who shows the meter near the 0 should be on the +1 that's it, but it's not the fact... i'm going to try with other samples

 

They have different signal thresholds and integration times.

The ones that seem less sensitive are just operating at a higher gate value and maybe a slower integration time. For all practical reasons unless you need 100% correlation and you are working with very quiet under -60dbFS program material they are identically useful. If you are making boom boom music that is going end-up relatively loud, this technical difference is just mathematical.

As always when it comes to phase cancellation - your ears and a mono-sum mix are worth more than any meter.

 

Thanks Mozee. Now i have a 3rd question about correlation : should the stereo signal (only the sides) be at "0" on a correlation meter to be called a perfect stereo signal, without phase issues ?
I've analysed a few stereo samples and when i solo the sides, the correlation always is between -0 to -1, almost never at 0. So i exported in false mono and created by myself a stereo signal and now the correlation, when i solo the stereo sides, is sometimes on "0".

 

reverb + microshift it's what i'm using, but i don't have the same wide stereo effect, however i don't have phase issue on the stereo signal...

 

Last edited by a moderator: Aug 17, 2017

Try with a hardware oscilloscpe. Good cables, 1Khz, 10 khz sinus tone and see what's happen in real life. Dac must be calibrated as well.
Good luck !

 

+1 for ozone

 

To resume, as long as i am between the 0 and +1 side, i am ok. Even if sometimes it flips 1 second under the 0. Right ? And more i have a wide stereo signal, more it would come to the 0. More i have a narrow signal, more it will come to the +1, right ?

 

Yes

For loud music with a lot drum and bass content you should stay around +.5 (45 degrees) while the low end is bumping. During the quieter parts small dips into negative territory are fine.