MeldaProduction

mccy · Post by **mccy** » Sun Apr 03, 2016 10:03 pm

I just checked, if I need another noise-reduction and stopped searching after MspectralDynamics outperformed waves x-noise by far. Maybe it had to do with the settings, but Melda is all I need. PERFECT. It's nice to open the EDIT GUI where I could learn noiseprofile etc...

Thanks!

Boone777 · Post by **Boone777** » Sun Apr 03, 2016 10:17 pm

When do you use MspectralDynamics instead of MautodynamicEq ? Noise makes sense but other than that ? Exemples ?

jmg8 · Post by **jmg8** » Sun Apr 03, 2016 11:19 pm

one example is sidechain ducking, MSD gives much more transparent results as it is more able to match the frequency of a signal.
another is for more transparent compression and limiting.
In a nutshell spectral is more accurate than using a few bands.

astramistil · Post by **astramistil** » Mon Apr 04, 2016 4:14 am

So the ideal mixing situation would be to ''inter''-sidechain every instruments?

I know alot of people only sidechain kick and bass, or voice and mix.
but why not ALL instruments? Wouldnt it be the most natural way to mix?

Is it the next 'paradigm shift' in mixing?

So we would almost never have to compromise in sound shaping?

What are the risks, trade-off?

Someone tried and can give some advises?

vectorwarrior · Post by **vectorwarrior** » Mon Apr 04, 2016 5:12 am

I wondered the same thing, find and create frequency holes for everything. I would be really interested in this.

jmg8 · Post by **jmg8** » Mon Apr 04, 2016 10:14 am

Yes, this is how I mix. Obviously this method alone can't mix the track it is just another tool. The trade offs are only audible if you have overdone it. If used gently it is very transparent and really helps to make space in the mix for each instrument.
The problem is with the DAW and how it allows sidechain etc.
You can currently only remove frequencies in one track from another.
My ideal would be to have a signal sender on every track, that would feed into the master plugin. Here we could make decisions and shape the whole mix together, choose amounts of ducking and priorities of certain sounds.
This is an idea I've had for about 2 years now and nothing is out there! I wish Vojtech could do it, but he says that the signal can't be sent between tracks, although it seems to work fine in MCompare and MMultiAnalyser.
Hey maybe this is his mix revolution idea?

astramistil · Post by **astramistil** » Mon Apr 04, 2016 5:18 pm

I know I've been struggling for 3 years to find a way to do this... I had the intuition that it would be the next revolution in mixing but wasnt sure. But now you two jmg8 and vectorwarrior seem to approve so we could start a whole thread on this and share hours of experimentation.

The problem is that Melda or Reaper wont allow ''feedback loop''

*THOUGH, Ive found in reaper a way to ''trick'' it.. (but I haven't took the time yet to try this on a full mix). Maybe there is something wrong in my thinking, but you guy can tell me.

=======================
Here how it works:
(lets say Guitar 1 and 2)

Routing:

routing.jpg

Complex Mixing:

complex mixing.jpg

=======================

There are 3 main elements:
1) tracks (G1+verb and G2+verb)
2) main bus (:G1: (which contain G1+verb) and :G2:)
3) complex mixing bus ([I-G1] and [I-G2])

---

Now here how you can trick reaper, and properly analyses whats going on (I think):
a) Routing

1) remove all ''master'' routing from tracks; route to :G1: and :G2: accordingly
2) remove master routing too from :G1: and :G2: ; route to [I-G1]
3) leave master routing from [I-G1]

so now all tracks have a ''serial'' flow

-----
b) FX chain

1) add MMultianalyzer on [I-G1] pre and post MSpectralDyna; you can now compare with 'pre' [I-G1] and [I-G2] conflicts, then compare [I-G1] pre and post-processing
2) MAutodynaEq is great because you can use ''sidechain'' in Analyze window, also the 'Auto' mode as suggest by jmg8 could help for general overview of potential conflicts, but MSpectral used adequately for sure would do a better job I think (now, we have to define ''adequately'' in general, and ''subtle'' is not precise enough

**I know its a per-case issue, but**
(I mean, if my aim is to keep ''natural'' sound as possible using this kind of mixing, what parameter for lets say vocal works *generally* well (as a startup) for Slope, resolution, naturality, etc..)

-----
c) Sidechaining

1) Send I/O from :G1: in MSpectral [I-G1] tagged ''G2''
*watchout; with each new bus added, you have to route sidechain from a different input (G1 = 3/4, G2 = 5/6, Bass = 7/8 etc.)

Otherwise you wont be able to seperate instrument from each MSpectral instances

========
I tried, it worked. I could make a video, but if you follow all of this you should be able to try on your own.

So what do you guy think? Vojtech?

astramistil · Post by **astramistil** » Mon Apr 04, 2016 5:21 pm

Vojtech, would it be possible to eventually add an ''Auto'' mode (as in MAutoDynaEQ) to Mspectral that would set the threshold accordingly? (would it give similar results but alot more precise and natural?)

astramistil · Post by **astramistil** » Mon Apr 04, 2016 6:18 pm

(a video is coming)

Boone777 · Post by **Boone777** » Mon Apr 04, 2016 8:32 pm

astramistil wrote:Vojtech, would it be possible to eventually add an ''Auto'' mode (as in MAutoDynaEQ) to Mspectral that would set the threshold accordingly? (would it give similar results but alot more precise and natural?)

Makes sense!

astramistil · Post by **astramistil** » Mon Apr 04, 2016 8:36 pm

Here

(there is no sound; I chose random track)

astramistil · Post by **astramistil** » Mon Apr 04, 2016 8:46 pm

Boone77, why not?

If I only want most important pocket of Hz in G2 to sidechain G1, I have to manually set the threshold in G1's MSpectral to leave important pocket of Hz from G1 untouched.

Unless Im missing something

MeldaProduction · Post by **MeldaProduction** » Tue Apr 05, 2016 2:04 pm

astramistil wrote:Vojtech, would it be possible to eventually add an ''Auto'' mode (as in MAutoDynaEQ) to Mspectral that would set the threshold accordingly? (would it give similar results but alot more precise and natural?)

Ha I'm lost, how? Why? What?

As for keyword feedback I noticed when browsing this post - feedback is a very complex thing, it can be done in MXXX, but it basically means that the host/plugin needs to subgranularize the audio, so I'm not really surprised what Reaper cannot do that.

robertszalapski · Post by **robertszalapski** » Tue Apr 05, 2016 7:36 pm

Reaper does allow feedback, but that's a bad idea for what you're trying to do. You really want to create "control" or "detector" tracks that you send as a sidechain and don't use for the output audio.

astramistil · Post by **astramistil** » Tue Apr 05, 2016 8:03 pm

Lets say my Guitar 2 most defining frequency pocket are ''400-800'', ''1k-2k'' and ''4-7k''

And I only want those 'pocket' to make G1 duck

''Seperate'' mode in AutodynaEQ will ''draw'' the bands accordingly then dynamically cut, while leaving other pocket untouched

But when I use MSpectral, I have to set by hand multiple threshold to isolate (in that case) 400-800, 1k-2k, 4-7k. Otherwise a ''full-range'' Threshold will cut Hz I dont want to be touched in my G1

What I mean by ''auto'' (which is not the best term I guess in that case) is that based on an analysis of the G2 most important pocket of hz, MSpectral could ''draw'' these threshold to isolate the most defining (or should I say 'loudest') pockets, as I do manually.

Then I can start from here to fine-tune depending on my taste, and needs
-----

As I say unless Im missing something, what Im trying to do: *is to get similar results with MSpectral as when Im using ''Auto'' in MAutodynamicEQ to cut specific pocket with sidechain from G2 to G1*. I dont use ''boost'' in auto mode, only cut.

So how can I achieve this in MSpectral other than setting multiple Threshold?