MeldaProduction

wildcafe · Post by **wildcafe** » Sun Oct 25, 2015 4:07 pm

Hi guys,

MeldaProduction is true!
Tried it as described with MMultibandDelay (just solo corresponding band in each channel and set everything else to 0) and it produces the same results as SplineEQ, although SplineEQ uses slightly less cpu.
The reason: right click on the multiband in the upper part. Then select as crossover "Hybrid".
If you don't do that, you'll get the problem of overlevelling again. Disadvantage: it costs cpu power and induces more latency (but not too much, so it's ok).

So at the moment there're two solutions that work for a clean and cpu effective multiband separation: SplineEQ and MeldaProduction Multiband (with correct set crossover type) plugins.

I've compared it now also to the likes of Waves etc. and no one can achieve the same results (always raising the level or inducing too much latency).

But I have one question.
- Take 3 channels
- Do the frequency separation as described above with the MMultiband...
- Mute (only) the low end channel (the complete channel, not only the fx!)
- Now the overall gain is exceeding 0dB, although we've CUT the low frequency channel (so there should be less gain instead of more)

Why is this?

Btw: the solution with SplineEQ has the same behaviour whereass it's level is ~0,5dB less/better than the Melda solution...
I'm still working on a video which will go online soon. I'll post it here then...

MeldaProduction · Post by **MeldaProduction** » Sun Oct 25, 2015 9:27 pm

It uses more CPU, because the crossover is more complex in our case. It also processes all bands in all instances.

As for increasing peak level - why do you care anyway??? It's most likely caused by ringing in case of hybrid crossover. If you use analog crossover, it can possibly cost less CPU and instead of ringing, the increase in peak level will be caused by the phase shift.
Anyway again, it does NOT matter! You are going to do something anyway. And finally there will be a limiter...

The crossover keeps the loudness intact, that's what counts. Measuring peak level doesn't make sense...

wildcafe · Post by **wildcafe** » Wed Oct 28, 2015 1:59 am

I assumed that these peaks are caused by phase differences. But why is ringing involved?
I care about these peaks because of the overall sound level. In other words: if I'm using multiband and maybe several instances of different multiband effects on the same channel, each effect causes higher and higher peaks and the overall sound gets quiet and more quiet (if I don't want to go over 0dB without using a limiter, which squashes the signal).
And if I use this on other instruments / channels, each channel adds peaks resulting in an overall "peaky" sound.
A nightmare from the perspective of a loud and transparent sound.

Maybe I'm wrong, but the fewer limiting I use the better and transparent the sound is (in the master).
So that's why I'm looking for the cleanest and most efficient solution and try to prevent unnecessary limiting in multiband situations.

vectorwarrior · Post by **vectorwarrior** » Wed Oct 28, 2015 3:13 am

The 'Linear' and 'hybrid' crossovers do not suffer from phase changes, but they do suffer from pre-ringing. That's the trade off. I believe that is what vojtech was referring too.

This is just the reality of linear vs analog filtering unfortunately.

MeldaProduction · Post by **MeldaProduction** » Wed Oct 28, 2015 3:54 pm

Correct. But also, linear phase and hybrid filters are steeper. There are also steeper analog filters (120dB/oct for example).
Simplified explanation: If you want some really steep filter at say 1000Hz, then you probably need to respond very strongly to that frequency right? (whether it meens amplify or attenuate) As a result, the stronger the filter (e.g. steeper crossover), the longer the pre/post ringing.
Also human brain doesn't really like steep filters, because they just don't exist in nature. Well, partly, try sticking your ear in a sea shell, which is essentially an interesting filter

. And just sound odd, doesn't it? That's why everyone is sticking their ears in there

. Anyway the point is, that no only the fact that strong filters ring scientifically is relevant, it's also the psychoacoustic fact, that these things are not natural.

As for increasing peaks etc. - I don't think that's true. If you put several multiband filters one by one and then a limiter at the end, it should be just fine. It depends on what each effect is of course. If you however do multiband -> limiter -> multiband -> limiter... then of course the volume would go down, but why on earth would you do that

.

Phase shift does NOT increase peak level on its own. The peak level rises if previously aligned waves get disaligned - for example take a square wave and put it thought a crossover and the peak level will probably go up. If you however run a sine wave, nothing will happen probably. In fact, each band will be lower in level as the energy spreads between them. That said, phase shift can also lower the peak level, the probability is probably lower though.

vectorwarrior · Post by **vectorwarrior** » Wed Oct 28, 2015 4:26 pm

This is why I was hoping you'd add 6dB crossovers to the multiband plugins and 3dB filters to the EQ

MeldaProduction · Post by **MeldaProduction** » Thu Oct 29, 2015 9:41 am

Hehe sorry, but that's kind of overkill. To be honest, as for me the 12dB/oct is already "not enough". I think the classic 24dB/oct is the best compromise between pretty much no ringing at all and enough filtering so that we could actually call it a crossover

wildcafe · Post by **wildcafe** » Fri Oct 30, 2015 12:33 am

Thanks for your explanations, I really appreciate that! Now I'm getting closer in my understanding...

MeldaProduction

KVR Audio

Multiband crossover filter?