Here I go again: Melda vs. Sonible, Oeksund Soothe, Gullfoss?

[Bullfrog001]

GF certainly doesn't sit there trying to recreate some ideal curve - quite different to that.

Hmm. I think that is more or less what it’s doing. You actually said so yourself earlier in slightly different words: “Part of what Gullfoss can do is to egde the frequency distribution towards an idealised form…”

After testing it a bit that does seem to be what it’s doing based on the various settings. “Recover” gives a slight smiley-face curve (lows and highs boosted) while “tame” does the opposite (boosts the mids). But especially on reasonable settings they don’t stray too far from pure pink noise. When both settings are equal (recover at 50%, tame at 50% say) the target curve *is* pink noise with a very slight (one dB when it’s at 50%) wide dip around 8k or 9k.

It’s certainly not performing any special magic with psycho-acoustics or Fletcher-Munson curves or whatever you might gather from their marketing.

The other thing I noticed is it’s reacting a lot faster to material than I’d assumed so if you’re trying to match MSpectral to this disregard the settings I recommended earlier in the thread. It actually is reacting quite dynamically. Settings a lot closer to default on MSpectral make for a closer match. Try a smoothness of 3%, naturality of 5%, slope +3dB, att/rel 30/100, knee size 25% and a ratio of 2:1. Pull down the threshold until MSpectral needs 5dB of makeup gain to match the bypassed signal.

Those settings should roughly match what Gullfoss does with recover/tame both at 100%. (Adjusting the dry/wet on MSpectral to 50% mimics recover/tame settings of 50% on Gullfoss.) Level match it with Gullfoss and I don’t think you’ll hear much difference. Follow MSpectral with an EQ if you want some of the other curves Gullfoss offers.

[jmg8]

Cool, nice homework.
To be honest, and I mean no disrespect to the developer......but even if we got it almost spot on (what is happening under the hood) I doubt we would get any confirmation from the developer, as they clearly what it to be secret.
Bullfrog001, your idea to get the same sound will work on the whole frequency range. If we want to use it only on certain frequencies, we could use your settings and then use the threshold curve to decide where to apply the effect.
As for Fletcher Munson curves, I have given it a lot of thought and I cant really decide how I feel about it being an ideal. I understand what you say that if we look at any great songs frequency curves they are not similar to Fletcher Munson. However from a scientific point of view, i will ask the question........would the most natural/pleasing sound be to have an even loudness across all frequencies based on human perception?

[pumafred]

would the most natural/pleasing sound be to have an even loudness across all frequencies based on human perception?

This, IMHO, is the key question, and I suspect the answer is “no”.
On the one hand, you have centuries of much-loved (Western) music you can analise, and it generally has the downward slope that looks somewhat similar to pink noise.
On the other, I also find it significant that the “relaxation/meditation” world seems to find white noise (and sometimes pink) useful for their exercises. Either generated, or taken from nature (rain, sea waves, etc.)
Finally, consider we are already perceiving these pleasing music and sounds through our ears’ and brain’ F-M filters. (Which, btw, I understand we should more properly be calling ISO 226:2003 equal loudness contours, since these are the most recently accepted re estimates of the old Fletcher-Munson experiments.)
Equalising to these curves may reasonably sound “horrible” to us, as Bullfrog001 points out.

[pumafred]

Thanks, Bullfrog001, very interesting! (Especially since I cannot test the plugin in my Windows setup.)

The other thing I noticed is it’s reacting a lot faster to material than I’d assumed so if you’re trying to match MSpectral to this disregard the settings I recommended earlier in the thread. It actually is reacting quite dynamically.

“Capable of changing its frequency response more than 100 times per second”

Also, have you tried to invert the polarity of your MSD version, to hear how much cancels out with Gullfoss?

Finally, since you seem to be good at this , have you looked at Sonible’s Smart EQ Live, which, as I see, when left constantly processing (rather than settling on an eq curve) , also tries to constantly equalise the incoming material to some unspecified benchmark?

[Bullfrog001]

...even if we got it almost spot on (what is happening under the hood) I doubt we would get any confirmation from the developer, as they clearly what it to be secret.

I think that’s almost certainly the case, yes.

If we want to use it only on certain frequencies, we could use your settings and then use the threshold curve to decide where to apply the effect.

Yep, that’s right. Though that could be one area where G (I’m tired of typing out Gullfoss by now) has an advantage. MSD only cuts (whereas G cuts and boosts) which is fine if you’re working on the full spectrum—you can use post-gain to bring the levels back up. If you only do the compression on part of the spectrum—let’s say 20Hz to 1k—then that area will be well balanced but it will also be lower in level relative to the 1k to 20k region which received no processing. There are ways around that of course (the easiest being to place a low-shelf EQ after MSD and bring up the 20Hz to 1k region).

[Bullfrog001]

...have you tried to invert the polarity of your MSD version, to hear how much cancels out with Gullfoss?

No, I haven’t. This was a fairly quick and dirty match. If I have time later I might play around with it a bit more. For all that though, at least on the material I tested it on, MSD sounded remarkably similar.

Finally, since you seem to be good at this , have you looked at Sonible’s Smart EQ Live, which, as I see, when left constantly processing (rather than settling on an eq curve) , also tries to constantly equalise the incoming material to some unspecified benchmark?

Ha ha no, I haven’t. I think this area of plug-in development is really interesting though and it is nice to know how various plug-in developers have tackled these issues. If I have some time in the next little while I might try to have a look.

[jmg8]

I have made 2 presets for MXXX called "Transient EQ" and "Spectrum EQ" they work like Sonible's proximity and entropy EQs. They offer less as there are active preset gui limitations. Vojtech will release these at some point. Also I am working on a version of their smart eq and a new concept of my own which is an anti resonance EQ.

[pumafred]

I have made 2 presets for MXXX called "Transient EQ" and "Spectrum EQ" they work like Sonible's proximity and entropy EQs. They offer less as there are active preset gui limitations. Vojtech will release these at some point. Also I am working on a version of their smart eq and a new concept of my own which is an anti resonance EQ.

[pumafred]

I have made 2 presets for MXXX called "Transient EQ" and "Spectrum EQ" they work like Sonible's proximity and entropy EQs. They offer less as there are active preset gui limitations. Vojtech will release these at some point. Also I am working on a version of their smart eq and a new concept of my own which is an anti resonance EQ.

Have you checked their new Live version? One interesting thing about it is that there is an option to leave it adjusting constantly, rather than designing a static eq curve as smart eq does.

[pumafred]

Whoa, whoa. What’s everyone doing here???

I’m all for experimentation but balancing a mix to Fletcher-Munson curves sounds like an objectively bad idea. Balancing to pink noise sounds like a less bad idea but be aware if you’re mixing anything other than modern pop you’ll end up with overly bright mixes. You might get decent results if you limit yourself to the area between 100Hz and 3k or so but in most genres you’ll end up with too much deep bass and waaaay too much top end (especially above 10k where even most high-energy pop songs roll off a lot more steeply than 3dB/octave).

Take a look at the curves presented here for a handful of different genres:
http://www.harmonycentral.com/articles/ ... ical-style

None of them follow a strict 3dB/octave (pink noise) slope and they definitely don’t look like Fletcher-Munson curves. If equal loudness curves or "psycho-acoustic pre-filter" curves (I think Vojtech’s is just a rolled off low end with a high shelf but it’s been a while since I checked) were some kind of ideal to shoot for in mixing and mastering then you’d find songs that follow these curves. But you don’t. Frankly if you mixed a song to a Fletcher-Munson curve I think it'd make your ears bleed.

As I said, experiment away, but I think you guys are going down a bit of a dead end.

A better reference, supporting your point: https://www.researchgate.net/publicatio ... _1950-2010

[Soundtheory1]

A better reference, supporting your point: https://www.researchgate.net/publicatio ... _1950-2010

Nice reference! I'm not sure how FMunson's got involved in this - and you certainly wouldn't wish to equalise to an FMunson curve to improve the feel of your track. You'd use for loudness compensation. Gullfoss's BOOST parameter uses a perceptual model equivalent of FMunson to allow for mastering in different listening environments. Very naively speaking the BRIGHTNESS parameter plays with an equivalent of those AES curves, adjusting for genre expectations. Which then leaves the two main parameters of Recover and Tame to alter the relationship between perceptual elements in the signal.

[Kalamata Kid]

The few Gullfoss videos I have seen take a well mixed track and try to improve it using Gullfoss.
I would like to hear a badly mixed track to be improved with Gullfoss.
This will help me with my tracks

I do ambient and experimental will Gullfoss help?
I use Windows 10 so I will have to wait a while but want to be ready to buy if on release there is a special price.

[pumafred]

Forgive me, but here I am back with Oeksound Soothe. (For the record, I own all Melda plugins, so I raise the issue a) out of curiosity and b) to see if some of the experts in this forum can help me see if I can get similar results with Melda plugins.)

This is the best and most detailed review I have seen of Soothe, with the reviewer examining how it works, and exchanging mail with the creator, to validate or no. Hopefully, these details can be of use:

http://admiralbumblebee.com/music/2018/ ... eview.html

[MeldaProduction]

I checked the review and if that's all correct, then it looks simply like MSpectralDynamics with some threshold and infinite ratio, pretty much like the spectral flattening device just with a different threshold (fletcher-munson's based). Personally I think it makes no sense to use these curves, since it all depends on the loudness etc... But hey, who knows

[pumafred]

I checked the review and if that's all correct, then it looks simply like MSpectralDynamics with some threshold and infinite ratio, pretty much like the spectral flattening device just with a different threshold (fletcher-munson's based). Personally I think it makes no sense to use these curves, since it all depends on the loudness etc... But hey, who knows

Great, thanks! Those plugins look so easy to use (with ears), that they are tempting, compared with "start to explore the thousands of Melda controls and parameters". But, since I own all of your products, I will try to do this, hoping other forum members perhaps can help a bit.
I asked the reviewer who, to me, seemed smart, about the difference between Soothe and MSD, and he said MSD can do much more, but what Soothe does, it does well and much more easily.
The typical dilemma you must have come across sooo many times by now.
Cheers!