Why Does Ratio Default to Equal Power Crossover?

[dmbaer]

In the Ratio module in MXXX (and I assume in MRatio in the free bundle) the default Mode is Equal Power, the one that looks like the first quarter cycle of a cosine and a sine superimposed.

When it comes to pan laws, this makes perfect sense. Take a mono source and pan with this crossover and you hear a constant volume across the panorama.

However, when mixing two stereo signals (using Ratio), the way to keep the volume consistent is to use the Linear Mode. Try it – it’s trivial to set up a simple experiment in MXXX.

So, having discovered this, I’m realizing all my previous uses of Ratio in MXXX used the wrong mode. I didn’t think it through and figured the default Mode was the one most appropriate in most cases.

Am I missing something? Why was Equal Power chosen as the default?

[jmg8]

This is a funny one.
So take MXXX and insert a noise generator. In its default state it will give a meter reading of -10db.
Now add another in a separate lane and place a ratio after them. Use linear mode as you suggest and you will see that the 2 signals together give a meter reading of -10db.
Perfect right?
No.
Try to use your ears instead of the meter. It is easy to here that linear mode does not give an even volume from A to B.
However equal power mode sounds the same throughout the whole crossfade.

[MeldaProduction]

Actually this is sort of a statistical topic . Basically the equal power IS the statistically the best bet, unless the signals are heavily correlated and in phase.

Simply up - if you have 1 signal split into 2, with say just a volume change or gentle filtering, then the correlation is nearly 100% and linear version is the best. But if you have 2 independent signals, the change they would be correlated are pretty minimal. Some frequencies will add, others will negate, most will do somewhere in between. And then the equal power is the best bet, since it will compensate for the typical average power loss.

[jmg8]

That makes sense now. I used noise as my example, so basically because of the random nature of noise, the two signals were completely uncorrelated therefore the equal power mode worked best. Cool.

[dmbaer]

Actually this is sort of a statistical topic . Basically the equal power IS the statistically the best bet, unless the signals are heavily correlated and in phase.

Aha! OK, that explains it entirely. Clearly my test using two oscillators at default setting feeding into Ratio was the *worst* possible one I could have constructed. That seemed appealing because the audio sources were very well behaved ... too well behaved, it turns out. Thanks.

[jmg8]

Hehe, that was fun. Nice post dmbear.