Spectral Casting - when is it useful and how to use?

[bencuri]

I have seen this technique demonstrated in a video of Spectralayers Pro. Seemed to be an advanced method, but years have passed and except for that video I have not seen it to be used anywhere (maybe I checked too few tutorials).

I have been planning to use it in situations when I have the lead vocal and the lead guitar going together at the same time. I found this can cause one cloaking the other. And this technique seems to be able to solve that. But I am not sure, just an assumption.

Also I don't understand how you mix those layers? For example the lead guitar is the base layer, you remove parts from it where the vocal will be casted in. How do you effect them afterwards? You effect them separately? But won't it cause an issue that portions of the lead guitar frequency spectrum will be missing? Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

Or you need to merge them together? But how do you effect the vocal then alone later? Or maybe you cast the vocal as the very last step, after the mastering phase?

So, all in all, what is the normal workflaw of this in the context of a full project in a DAW?

There are so many aspects of it that raise questions for me.

[Tommy Dahlman]

Caveat, I haven't done a deep dive into the workings of this particular software.
But to me it sounds like you could achieve the same thing
using the old school technique of side-chaining the vocals into a multi band EQ or compressor, like the FabFilter plugs, on the guitar channel.
This would defeat the masking at selected frequency bands.
I did use the Spectral editing features of Adobe's Audition about 20 years ago,
to remove someone in the front row coughing during a sensitive solo violin piece.
Very time consuming since there was no automation feature,
so I had to do each burst of cough separately.
And the toughest part was to get rid of the coughs in the concert hall ambience.
But I got a result that was good enough to broadcast. So mission accomplished.
If I had to do it today I would look at iZotope's RX restoration/clean up suite.
But for a close miked, well recorded vocal - guitar set up
I believe the EQ/compressor trick would do it.
The best of luck

[James Perrett]

Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

You would normally only use a technique like this if there was a problem that couldn't be solved by simpler means like gain riding. I wouldn't do this until I was close to the end of the mixing process with the vocal effects already close to being finalised and I'd use something that changed the spectrum in real time so that I could adjust it at any point in the mixing process.

[The Elf]

Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

You would normally only use a technique like this if there was a problem that couldn't be solved by simpler means like gain riding.

...or arranging the song such that the lead guitar and vocal *don't* play together!

[Sam Spoons]

I think the style/genre dictates that the guitar is playing pretty much throughout the piece?

[bencuri]

Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

You would normally only use a technique like this if there was a problem that couldn't be solved by simpler means like gain riding.

...or arranging the song such that the lead guitar and vocal *don't* play together!

I see you don't know, but there are genres in the world that involve solo and vocals going together. If you deal with them, rearranging is not an option.

We can learn something new every day, right?

[bencuri]

Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

You would normally only use a technique like this if there was a problem that couldn't be solved by simpler means like gain riding. I wouldn't do this until I was close to the end of the mixing process with the vocal effects already close to being finalised and I'd use something that changed the spectrum in real time so that I could adjust it at any point in the mixing process.

When you watch that tutorial from Magix on Spectralayers, you assume it is some kind of simple and advanced technique. And up to the editing part it may definately be but I am surprised the reapplication of layers into the DAW has not been addressed by them too much. It is like having a car without the key.

They should have provided some guidance at least at which stage it is useful to be done and merging how many tracks.

[The Elf]

Also, if you add effects to the vocal later, that might mean extra spectral spaces that were not removed from the guitar spectrum. So the casting won't be accurate.

You would normally only use a technique like this if there was a problem that couldn't be solved by simpler means like gain riding.

...or arranging the song such that the lead guitar and vocal *don't* play together!

I see you don't know, but there are genres in the world that involve solo and vocals going together. If you deal with them, rearranging is not an option.

We can learn something new every day, right?

Of course. What genres are these?

[James Perrett]

When you watch that tutorial from Magix on Spectralayers, you assume it is some kind of simple and advanced technique. And up to the editing part it may definately be but I am surprised the reapplication of layers into the DAW has not been addressed by them too much. It is like having a car without the key.

They should have provided some guidance at least at which stage it is useful to be done and merging how many tracks.

Most of the tutorials that you find on YouTube are created by people who have little experience of the real world of recording - or else they are trying to sell you the latest trendy effect that you don't actually need. You can safely ignore 90% of recording tutorials on YouTube (I had a great chat with a very well known recording engineer at Gearfest this year after I heard him say exactly the same thing).

I'll bet some sales person at Magix came up with this notion, gave it a fancy name, and made a video about it but didn't really consider how it would work in the real world of recording.

[alexis]

...
I have been planning to use it in situations when I have the lead vocal and the lead guitar going together at the same time. I found this can cause one cloaking the other. ...

the lead guitar is the base layer, you remove parts from it where the vocal will be casted in. ...

I'm not sure if you're saying that, because the guitar track is clashing with the vocals, you want to remove all guitar playing at certain points in time.

If you are, that is certainly one way to make room for the vocals, but there are less drastic ways that still allow the guitar to be heard well even while the vocal is getting top billing. One way is by reducing in amplitude the guitar frequencies that are stepping on important vocal frequencies. I think some suggestions on how to do that were made in another thread you recently started.

If you happen to have access to the individual piano and guitar tracks (not just the final stereo track), you can of course reduce the volume of the guitar when the vocals are playing.

A less sledgehammer- like approach in that situation (i.e., if you have access to the separate guitar and vocal tracks), if you know how to identify the more offending/colliding guitar frequencies, is to insert a compressor on the guitar track that acts mainly on that band of frequencies, and send the vocal track to the side chain input of that compressor, the end result being that when the vocals play the most competing guitar frequencies are compressor-attenuated, while the rest of the guitar signal gets through unprocessed. You could even EQ-boost those frequencies in the vocal before sending to the compressor to make the compressor more sensitive to those frequencies.

I agree with the above posters that there's a lot of chaff on YouTube regarding engineering techniques. For learning SpectraLayers I recommend Steinberg's own YouTube channel, or probably anything by Dom Sigalas, or Chris Selim on his MixDown Online YouTube channel.

I hope this is helpful, write back of course if more questions!

[alexis]

Holy canoli: Firesonic's Firespacer, one of the "SOS 2024 Gear of the Year" Award winners, looks like it does exactly what I was talking about in my last post:

"...A typical application might be to duck the spectrum of pad or distorted rhythm guitar sounds in the presence of vocals, where there are likely to be significant overlapping frequencies, especially in the midrange ...

https://www.soundonsound.com/reviews/fi ... firespacer

https://www.soundonsound.com/reviews/2024-gear-year

It looks so good that I would be tempted to try it out if I didn't have Fabfilter's ProQ3 and Cubase's own Frequency plugins, as well as Izotope Neutron 4, all of which seem to include addressing the same problem of competing frequencies, with ducking, as Firesonic's Firespacer.

I have Soothe 2 which looks like it addresses the same overlapping frequencies issues that Firespacer does, but I am not really learning much from its GUI.

[bencuri]

...
I have been planning to use it in situations when I have the lead vocal and the lead guitar going together at the same time. I found this can cause one cloaking the other. ...

the lead guitar is the base layer, you remove parts from it where the vocal will be casted in. ...

A less sledgehammer- like approach in that situation (i.e., if you have access to the separate guitar and vocal tracks), if you know how to identify the more offending/colliding guitar frequencies, is to insert a compressor on the guitar track that acts mainly on that band of frequencies, and send the vocal track to the side chain input of that compressor, the end result being that when the vocals play the most competing guitar frequencies are compressor-attenuated, while the rest of the guitar signal gets through unprocessed. You could even EQ-boost those frequencies in the vocal before sending to the compressor to make the compressor more sensitive to those frequencies.

Thanks, this can be one possible solution for the problem if it occurs. I am not familiar with the resulting sound characteristics of this method, but we'll see. It seems to me it is similar to the approach of speech over music during broadcasts, isn't it?

The methods you refer to from other posts of mine, those posts were question about different problems I am not sure which part you refer to in this regard.

You referred to turning down volume from the guitar track sometimes, if you mean a solution like this, that way I don't want to go:
https://www.youtube.com/watch?v=lqVdQR6fysA

I don't like this method.

"if you know how to identify the more offending/colliding guitar frequencies"
Can you link a tutorial how to do it properly because I am not a pro in this. I can do it only with trial and error devoting lots of time, but a more precise and effective method would be useful


I'm not sure if you're saying that, because the guitar track is clashing with the vocals, you want to remove all guitar playing at certain points in time.
No, Spectralayers Casting method doesn't work like that. You don't remove all the guitar during specific vocals moment. So if you imagined it as opening the wave of the guitar and cutting out segments at specific points, it doesn't work like that when casting. The software detects the frequency spectrum of the vocal, the whole spectrum involved, and only deletes those parts from the guitar spectrum. So at a given point in time when they would clash, not the whole 0hz-20Khz spectrum of the guitar will be deleted but only a segment of that, where the vocals are. The rest of the guitar spectrum stays unaffected. So in practice casting is not a "attenuating the guitar' method.

[bencuri]

When you watch that tutorial from Magix on Spectralayers, you assume it is some kind of simple and advanced technique. And up to the editing part it may definately be but I am surprised the reapplication of layers into the DAW has not been addressed by them too much. It is like having a car without the key.

They should have provided some guidance at least at which stage it is useful to be done and merging how many tracks.

Most of the tutorials that you find on YouTube are created by people who have little experience of the real world of recording - or else they are trying to sell you the latest trendy effect that you don't actually need. You can safely ignore 90% of recording tutorials on YouTube (I had a great chat with a very well known recording engineer at Gearfest this year after I heard him say exactly the same thing).

I'll bet some sales person at Magix came up with this notion, gave it a fancy name, and made a video about it but didn't really consider how it would work in the real world of recording.

The Spectralayers tutorial about casting that I referred to was not uploaded by a third party individual, but it was on the website of Magix at the Spectralayers Pro page. There were a bunch of tutorials there, and casting was one of them. So it was an official tutorial, not done by some outsider.

I have just wanted to link that tutorial, but I noticed that page doesn't even exist by now, and Magix doesn't even sell Spectralayers Pro any more. This is new to me because I purchased mine many years ago, when it was not even Magix but Sony Spectralayers.

I have just noticed the software was purchased by Steinberg and they distribute it, I am not sure if a standalone or as a segment of their other softwares. But it seems to be involved in Cubase now. So maybe there is a solution now to the problem of workflow. There seems to be changes to the casting option, too, they call it imprinting now.

I can see in the videos they are working with it inside Cubase. It is a long video, I still need to check it in full. But just previewed it fast and found there is a point where they compare sidechaining and casting:
https://www.youtube.com/live/xu4Dv_3j9x ... v74&t=3071

This tune that the presenter is testing, the change the casting makes is a bit drastic to my ears for the first listen, but keep in mind in my case I don't want to wreck the whole mixing, only want to adjust the coupling of the lead guitar and lead vocal.

[ManFromGlass]

Aside from the programs Alexis mentioned recently I have been experimenting with Melda’s AutoDynamicEQ to duck frequencies that are the same as the vocals.
It’s working great, I actually set it to remove too much from the backing track so had to ease off the degree of removal a bit as you could really hear the effect.

There are a few youtubes that show how easy it is to use. Melda has a lot of sales.

[alexis]

Thanks ...

"if you know how to identify the more offending/colliding guitar frequencies"
Can you link a tutorial how to do it properly because I am not a pro in this. I can do it only with trial and error devoting lots of time, but a more precise and effective method would be useful

You're welcome!

I think the tutorial is called something like, "Put 10,000 hrs in to learn a new skill". The ability to do this well on the fly is an acquired skill, and one of the many things that often separates someone with pro abilities from bedroom/home studio dudes and dudettes. I say that as one at least currently firmly ensconsed in the latter group! There are lots of folks on this forum who can do things like that almost in their sleep, their advice is golden!

For those of us in the cheap seats, things like Neutron 4 https://www.youtube.com/watch?v=U1sje7v ... gWywxGQFWO , and Fab Filter Pro Q4 https://youtu.be/IXWkViqU2K8?si=5L21E-Z9udwC2AzN4 can help get the job, or at least a fascimile thereof, done. I don't think any of these are more "precise or more effective" than knowing how to get the job done by understanding what's coming into one's ears, but at least it's a start! Similarly with the programs mentored by Man From Glass and others mentioned previously.

I don't know that any of them are particularly inexpensive. Voxengo does however make a plugin called Span Plus that can do this job for 40 Euros (current sale price) https://www.voxengo.com/product/spanplus/, and they also have a freeware version called Span that I'm almost positive can do that as well.

[bencuri]

Thanks ...

"if you know how to identify the more offending/colliding guitar frequencies"
Can you link a tutorial how to do it properly because I am not a pro in this. I can do it only with trial and error devoting lots of time, but a more precise and effective method would be useful

I think the tutorial is called something like, "Put 10,000 hrs in to learn a new skill". The ability to do this well on the fly is an acquired skill,

You see now why I noticed spectral casting as a method to use when I saw it before, LOL

[Drew Stephenson]

Might be worth clarifying a couple of things here.
Using one signal to reduce the volume another by setting up a compressor triggered by an external side chain is generally called "ducking". It's used where you have two elements playing at the same time and you want to emphasise one. Ducking a bass line when the kick drum is played is a typical example.

But ducking doesn't have to be done over the full frequency range. If you use a multiband compressor, or dynamic EQ, you can set it up so that only a specific frequency range is ducked. So you can set the range of the guitar between 1kHz and 3kHz, for example, to be ducked when the vocals are coming through to help with clarity. But this ducking will be triggered whenever the vocal goes above the threshold, irrespective of what frequency the vocal is at that point - so you could have a high guitar frequency ducked by a low vocal frequency where there is no real clashing.

What the spectral casting does, and other similar tools, is allow you to set it so that only the clashing frequencies are ducked and cause the ducking. So a low pitched vocal won't cause ducking on the high frequency guitar parts, but a high vocal will.
There are, as Alexis suggested above, ways to set this using conventional tools if you're only tackling a single frequency range but it would get very complicated very quickly if you wanted to tackle multiple frequencies over the same two tracks.

[Stuart79]

Trackspacer by Wavesfactory is a simple tool that would allow you to achieve what you're talking about.

[Drew Stephenson]

UnMask by Sonible is another cheap but effective option.

[ManFromGlass]

I discovered to my happiness that the Melda one can do multiple frequency ducking in one instance of it.
On one track it set up 3 nodes - 2 of which ducked the frequencies and 1 which increased the frequencies.
I assume the frequency increase was because those frequencies were low or non-existent in the vocals. Took me by surprise. But then I still have much to learn!

[bencuri]

Thanks for the various software ideas I will definately give them a try.

[bencuri]

What the spectral casting does, and other similar tools, is allow you to set it so that only the clashing frequencies are ducked and cause the ducking. So a low pitched vocal won't cause ducking on the high frequency guitar parts, but a high vocal will.

From what you write it seems to me this can still be a useful tool in my case, provided I manage to figure out the proper workflow. If I can, to do the ducking can be actually very simple because I have just found there is a percentage control for it, too, so you can set the ducking (attenuation) level as well, you don't need to remove the clashing frequency of one track completely. I did not notice this before.