Using pink noise to set initial mix levels: Is there a smarter way to do this with Melda?

[pumafred]

This seems interesting to set INITIAL mix levels, although cumbersome if you have many tracks.
Is there a smarter, more automatic, way of doing this with Melda plugins?
(Just in case, note this is not the same as equalizing a mix against pink noise.)

[jmg8]

I've not heard of this technique before, very interesting.
I think there might be a different way of achieving this.
So the pink noise has all frequencies even and therefore masks the frequencies of the sound until it is loud enough for it's highest frequency to just poke through the noise.
Well this "loudest frequency" is the same as a peak value on a meter. However as humans we hear the frequencies different we could use the psychoacoustic pre-filtering. So just a meter until it peaks at the same as the noise rms.
I might give this some experiments.

[steve2KVR]

I've used this technique on/off for a couple of years...it's simple enough to set up:
1/ on an empty channel put an instance of MN set to pink/100%wet and raise the gain until the channel meter reads your target level
2/ pick another channel with whatever you want to measure and increase the gain until you can "just hear" it, this will vary by material and listener
3/ then mute the channel leaving the fader just as you set it while listening
4/ ...etc, until all channels are accounted for...
5/ unmute everything and give it a listen ...see what you think hopefully...
6/ (especially interesting after you've "finished your mix")...sigh...ymmv... ...s/~

[pumafred]

I've not heard of this technique before, very interesting.
I think there might be a different way of achieving this.
So the pink noise has all frequencies even and therefore masks the frequencies of the sound until it is loud enough for it's highest frequency to just poke through the noise.
Well this "loudest frequency" is the same as a peak value on a meter. However as humans we hear the frequencies different we could use the psychoacoustic pre-filtering. So just a meter until it peaks at the same as the noise rms.
I might give this some experiments.

I assume you are remembering that the frequency distribution of pink noise has a 3db per octave slope, no? Therein lies the "magic" of this technique. And this is why I couldn't think of how to "automate" this somehow with peak metering.

BTW, I would think Vojtech would use something like this for his "mixing revolution".

[jmg8]

Could we not just use a 3db per octave pre filter before the peak meter?

[pumafred]

You tell me! I'm the amateur here!

[jmg8]

I'll try to make a preset in MXXX. I will do it by ear and then also using meters and see if i can achieve the same results. I will post on here with progress.

[MeldaProduction]

The idea was here a few times, but to be honest I just think it's more like a waste of time. You'd eventually move the faders anyways and you are likely to start with one track, say drums. Adding a plugin, performing some kind of analysis and stuff, would take much longer, than just moving the fader.

[pumafred]

The idea was here a few times, but to be honest I just think it's more like a waste of time. You'd eventually move the faders anyways and you are likely to start with one track, say drums. Adding a plugin, performing some kind of analysis and stuff, would take much longer, than just moving the fader.

Well yes, that's why I asked for ideas. I had thought it could be done with the MultiAnalizer, using the frequency distributions without normalising, each instrument against pink noise. But that would be even more work!! But if some semi automatic procedure existed...
Some proponents do it with pre gain, so they can then continue mixing with the faders starting at zero, where they have more definition.
Anyway, it was just curiosity, no big thing.
Curious to see what jmg8 concludes, however.

[MeldaProduction]

Here I wonder, why would you need the spectral profile anyway?

[pumafred]

I am probably thinking wrongly. I thought that since the above method involves raising the gain until you can just barely hear the track against pink noise, this would be the same as raising the gain until the highest frequency of the track barely surpasses the pink noise spectral profile. Is this mistaken?

[MeldaProduction]

Hmmm, no idea really . Interesting thinking though!

[dmbaer]

I am probably thinking wrongly. I thought that since the above method involves raising the gain until you can just barely hear the track against pink noise, this would be the same as raising the gain until the highest frequency of the track barely surpasses the pink noise spectral profile. Is this mistaken?

No, not the highest frequency, I think. To automate this incorporating frequency-dependent considerations, you'd probably need to factor in Fletcher-Munson characteristics of human audio perception. But that might actually get you close enough since this technique is all about getting to a reasonable starting point quickly, not an optimal final balance.

Of course, the bigger problem is giving this automatic process control over the fader position of your DAW tracks. That would require the cooperation of the DAW developers.

[pumafred]

To automate this incorporating frequency-dependent considerations, you'd probably need to factor in Fletcher-Munson characteristics of human audio perception.

I suspect not, because you are benchmarking against pink noise that will also be affected by these characteristics of human perception.
Anyway, I am venturing too far from my areas of expertise...

[steve2KVR]

...for those with interest, I strongly suggest setting up a small experiment and then judge for yourself if pursuing this line of thought is/will be productive. I don't think talking theoretically about this will get you anywhere useful. The "concept" on it's face is indeed very appealing so give it a try and let us know what you think...Cheers!.../s~